Proceedings of the 5th International Conference on Water Resources (ICWR) – Volume 2: Current Research in Water Resources, Coastal and Environment (Lecture Notes in Civil Engineering, 365) 9819935768, 9789819935765

Citation preview

Lecture Notes in Civil Engineering

Ilya Khairanis Othman Mohd. Ridza Mohd. Haniffah Mohamad Hidayat Jamal   Editors

Proceedings of the 5th International Conference on Water Resources (ICWR) – Volume 2 Current Research in Water Resources, Coastal and Environment

Lecture Notes in Civil Engineering Volume 365

Series Editors Marco di Prisco, Politecnico di Milano, Milano, Italy Sheng-Hong Chen, School of Water Resources and Hydropower Engineering, Wuhan University, Wuhan, China Ioannis Vayas, Institute of Steel Structures, National Technical University of Athens, Athens, Greece Sanjay Kumar Shukla, School of Engineering, Edith Cowan University, Joondalup, WA, Australia Anuj Sharma, Iowa State University, Ames, IA, USA Nagesh Kumar, Department of Civil Engineering, Indian Institute of Science Bangalore, Bengaluru, Karnataka, India Chien Ming Wang, School of Civil Engineering, The University of Queensland, Brisbane, QLD, Australia

Lecture Notes in Civil Engineering (LNCE) publishes the latest developments in Civil Engineering—quickly, informally and in top quality. Though original research reported in proceedings and post-proceedings represents the core of LNCE, edited volumes of exceptionally high quality and interest may also be considered for publication. Volumes published in LNCE embrace all aspects and subfields of, as well as new challenges in, Civil Engineering. Topics in the series include: • • • • • • • • • • • • • • •

Construction and Structural Mechanics Building Materials Concrete, Steel and Timber Structures Geotechnical Engineering Earthquake Engineering Coastal Engineering Ocean and Offshore Engineering; Ships and Floating Structures Hydraulics, Hydrology and Water Resources Engineering Environmental Engineering and Sustainability Structural Health and Monitoring Surveying and Geographical Information Systems Indoor Environments Transportation and Traffic Risk Analysis Safety and Security

To submit a proposal or request further information, please contact the appropriate Springer Editor: – Pierpaolo Riva at [email protected] (Europe and Americas); – Swati Meherishi at [email protected] (Asia—except China, Australia, and New Zealand); – Wayne Hu at [email protected] (China). All books in the series now indexed by Scopus and EI Compendex database!

Ilya Khairanis Othman · Mohd. Ridza Mohd. Haniffah · Mohamad Hidayat Jamal Editors

Proceedings of the 5th International Conference on Water Resources (ICWR) – Volume 2 Current Research in Water Resources, Coastal and Environment

Editors Ilya Khairanis Othman Faculty of Civil Engineering Universiti Teknologi Malaysia Johor, Malaysia

Mohd. Ridza Mohd. Haniffah Faculty of Civil Engineering Universiti Teknologi Malaysia Johor, Malaysia

Mohamad Hidayat Jamal Faculty of Civil Engineering Universiti Teknologi Malaysia Johor, Malaysia

ISSN 2366-2557 ISSN 2366-2565 (electronic) Lecture Notes in Civil Engineering ISBN 978-981-99-3576-5 ISBN 978-981-99-3577-2 (eBook) https://doi.org/10.1007/978-981-99-3577-2 © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023 This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors, and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd. The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721, Singapore

Preface

This book is the 2nd volume following the first one published online in October 2022. Although published later, this volume compiles selected high-quality papers presented during the 5th International Conference on Water Resources (ICWR 2021), held virtually in November 2021. The articles are, arguably, better than the 1st volume. ICWR 2021 was jointly organised by the Department of Water and Environmental Engineering, Faculty of Civil Engineering, Universiti Teknologi Malaysia (UTM) and Department of Irrigation and Drainage (DID) of the Ministry of Environment and Water Malaysia. It has been a long journey with a few hiccups overcome during the pre-publication stage (before engaging LNCE). Authors also bared with us throughout those times and have finally led to this publication in LNCE for the 2nd time. The publication of this 2nd volume shows the commitment of everyone involved in making sure the promise of publication from ICWR 2021 is executed. By the time of publication, almost two years have passed since ICWR 2021, and we hope everyone reading the published work here will look forward to the next ICWR 2024 (hopefully, there is one). Johor, Malaysia May 2023

Ilya Khairanis Othman Mohd. Ridza Mohd. Haniffah Mohamad Hidayat Jamal

v

Acknowledgement

The 5th International Conference on Water Resources (ICWR2021), held virtually in November 2021, has attracted over 70 abstract submissions and over 100 participants locally and internationally. We want to thank everyone involved in organising and sponsoring the conference. Special thanks go to the conference committee, Faculty of Civil Engineering, Department of Irrigation and Drainage (DID) Malaysia and UTM SPACE for their time and efforts in funding, managing, organising the conference and communicating with authors and reviewers. The peer-review processes have narrowed the selected papers to 19, presented in this 2nd volume in three chapters. Our sincere appreciation goes to more than 30 reviewers for their expertise demonstrated and commitment to reviewing the selected papers. All authors have cooperated in response to reviewers’ comments and the editorial needs.

vii

About This Book

This 2nd volume proceeding compiles selected high-quality papers presented during the 5th International Conference on Water Resources (ICWR2021), held virtually in November 2021. The fifth edition of this conference series, themed “Innovating for Resilience and Enhanced Preparedness to Water-Related Challenges”, focuses on various issues, novel findings and sustainable developments in water resources and environment, river basin and coastal zones conforming to the SDGs. Consequently, the steadily increasing stress and challenges on future natural resources require multidiscipline approaches. The papers presented in this book are arranged in three chapters. Firstly, Integrated River Basin Management, followed by Hydro-Environment and Coastal Engineering and Management. This book caters to postgraduate students, researchers and practitioners involved in advocating and embedding sustainability and enhanced preparedness in planning, simulation, development, design and management of water resources systems, environmental quality and coastal zones.

ix

Contents

Integrated River Basin Management Assessing the Impact of Climate Change on Flood Characteristics at Langat River Basin Using Rainfall-Runoff Inundation (RRI) Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ummi Hani, Nor Eliza, Kamarul Azlan, Rahmah, and Mohamad Wafiy Adli Analysing Impact of Climate Change on Hydrological Trend in Kelantan River Basin Using HEC-HMS Coupled with SDSM . . . . . . . . Muhammad Zahran Syahmi Armain, Zulkarnain Hassan, Mohd Remy Rozainy Mohd Arif Zainol, Sobri Harun, Ain Nihla Kamarudzaman, and Salwa Mohd Zaini Makhtar

3

17

Machine Learning Algorithms with Hydro-Meteorological Data for Monthly Streamflow Forecasting of Kurau River, Malaysia . . . . . . . . . Muhammad Nasir Mohd Adib and Sobri Harun

29

Prediction of Industrial Water Consumption - Blue Water Footprint in Kuantan River Basin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E. A. Aziz, S. N. Moni, M. J. Letchumy, N. Yusoff, and S. Z. Zabir

43

Validation of Gridded Data Set Over Semi-arid Region of Syria . . . . . . . . Rajab Homsi, Shamsuddin Shahid, Tarmizi Ismail, Jam Shahzaib Khan, Zafar Iqbal, and Atif Muhammad Ali

59

Flood Risk Assessment Considering the Effect of Covid-19 Pandemic in the Municipality of Balayan, Batangas . . . . . . . . . . . . . . . . . . . C. E. F. Monjardin, F. A. R. Cala, D. C. D. Po, and M. A. M. Sy

79

Sustainable Stormwater Management: Developing Stormwater Management and Drainage Master Plan for Serian, Sarawak . . . . . . . . . . R. Salleh, K. Y. Wong, Judy J. K. Kueh, T. Sulaiman, and A. Ainan

93

xi

xii

Contents

Development of Low-Cost Technology for Monitoring of Soil Moisture and Recycling Rainwater for Irrigation . . . . . . . . . . . . . . . . . . . . . 111 Siti Nurhayati Mohd Ali and Nuryazmeen Farhan Haron Hydro-Environment Utilising Aerial Mapping Approach on Dam Disaster Risk Reduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125 Rahsidi Sabri Muda, Izawati Tukiman, Ahmad Fadhli Mamat, Fatin Shahira Abdullah, and Mohamad Hidayat Jamal Short Timescale Riverbank Erosion and Bank Stability of Sg. Bernam Using Bank Stability and Toe Erosion Model (BSTEM) . . . . . . . 141 Azlinda Saadon, Zulkiflee Ibrahim, and Mohamed Fuad Said Khamis 3D Simulation on 90 Degree Off-Take Branching Channel with Separation Zones . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159 Siti Aimi Asyarah Zakaria, Mohd Ridza Mohd Haniffah, Amyrhul Abu Bakar, M Faizal Ahmad, and Iskandar Shah Mohd Zawawi Saline Water and Freshwater Interactions in a Narrow Meandering Channel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173 Mazlin Jumain, Zulkiflee Ibrahim, Wan Nor Afiqa Wan Mustafah Kamal, Sharifah Nurfarain Syed Abdul Jabar, Md.Ridzuan Makhtar, Noorarbania Abd Rani, Nurfarhain Mohamed Rusli, and Mohd Zulkhairi Mat Salleh Understanding Variability of Groundwater Potentials in Western Sokoto Basin: Implications for Sustainable Groundwater Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185 Saadu Umar Wali, Noraliani Binti Alias, and Sobri Bin Harun The Application of Statistical ANOVA, LSD and RSM to Agro-Based Filter Design Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . 199 A. A. Awang Nasrizal, J. Jason Lowell, J. Idris, A. T. Nazaruddin, and N. Bolong Coastal Engineering and Management Sandy Beach Responses to Sea Level Rise: Comparison Potential Coastal Inundation Maps Using Static and Numerical Model for Ibai River, Malaysia Case Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213 Yannie Anak Benson, Lee Hin Lee, Mohamad Hidayat Jamal, Dunstan Anthony Pereira, Ahmad. Khairi Abd. Wahab, Khairul Anuar Mohamad, Ikmalzatul Abdullah, and Ilya Khairanis Othman

Contents

xiii

Assessment of Coastal Vulnerability Index Under Storm Surge and Sea Level Rise Impact on the East Coast of Johor . . . . . . . . . . . . . . . . . 223 Norzana Mohd Anuar, Siti Habibah Shafiai, Hee Min Teh, and Ahmad Mustafa Hashim Hydrodynamic Assessment on the Impacts of Sea Level Rise at the Kelantan Shorelines and Delta, Malaysia . . . . . . . . . . . . . . . . . . . . . . . 239 Lee Hin Lee, Dunstan Anthony, Ikmalzatul Abdullah, Anizawati Ahmad, Aidah Rahim, Mohamad Hidayat Jamal, and Lee Shin Yun Impact of Sea Level Rise and Adaptations for Malaysia Shoreline . . . . . . 249 Hin Lee Lee, Khairul Anuar Mohamad, Dunstan Anthony Pereira, Amri Md. Shah, Yannie Anak Benson, Wan Ahmad Hafiz Wan Mohamed Azhary, Muhammad Rizal Razali, Saiful Bahri Hamzah, and Mohamad Hidayat Jamal Hybrid Current Turbine and Solar Cell Renewable Energy Device . . . . . 261 Azrul Aminur Rahman Yunus, Adi Maimun Abd Malik, Mohammad Hidayat Jamal, and Nursahliza Muhamat Yain

Integrated River Basin Management

The Integrated River Basin Management (IRBM) chapter presents papers on climate change impacts on flood characteristics and river discharge using the Runoff Inundation model and Hydrologic Modeling System. Blue Water footprint assessments and the water consumption trend are presented for Kuantan River Basin. A few Malaysian River basins highlighted in this chapter are the Langat River, Kelantan River, Kurau River and Kuantan River. Validation of Gridded Climate Data from the Global Precipitation Climatology Center and temperature Climate Research Unit using various statistical analyses over a semi-arid region in Syria are also presented. A paper on establishing a flood risk assessment incorporating COVID-19 effects for Balayan, Batangas, Philippines, is presented using Risk index mapping generated using the Analytical Hierarchy Process and ArcGIS. The Department of Irrigation and Drainage Malaysia has also introduced the Pelan Induk Saliran Mesra Alam (PISMA) for Serian, Sarawak. The PISMA Serian provides a comprehensive and long-term solution for controlling the urban stormwater quantity and quality issues. Lastly, a paper presents a functional prototype design and modelling of an automated irrigation system created to water the cocoa plants collected from a Rainwater Harvesting System.

Assessing the Impact of Climate Change on Flood Characteristics at Langat River Basin Using Rainfall-Runoff Inundation (RRI) Model Ummi Hani, Nor Eliza, Kamarul Azlan, Rahmah, and Mohamad Wafiy Adli

Abstract Flood modelling is one of popular tools in flood management. It can simulate flood depth and streamflow at different sections of the studied area enabling hydrologist to have better understanding on the flood especially with the impact of climate change. This study aims to understand the flood impact due to climate change in a basin scale using Rainfall-Runoff Inundation (RRI) model. Langat River Basin in Selangor, Malaysia is selected due to its location and urbanised area. The RRI were set up and calibrated based on observed data obtained from Department of Irrigation and Drainage (DID) Malaysia, topographic data from HydroSHEDS at 15 arc-second and landuse data from PLANMalaysia. The projected climate data (2080–2099) were extracted from Non-Hydrostatic Regional Climate Model (NHRCM) developed by Meteorological Research Institute (MRI) under the worst case scenario, RCP8.5. The maximum 1-day rainfall at 100-year ARI were chosen as the rainfall input in RRI. It is found that the rainfall intensities in projected 100-year ARI shows a decrement by 6.4% to 12.4% at most area in Langat River Basin. This has cause the peak streamflow in Langat River to reduce by 6.8% and peak flood depth to change by ±3% in the future. Thus, it can be inferred that due to climate change, the peak streamflow and flood depth at Langat River Basin is reduce in the future though the impact is low. Keywords climate change · flood modelling · rainfall-runoff inundation model · streamflow · flood depth

U. Hani · K. Azlan · M. W. Adli Faculty of Civil Engineering, Universiti Teknologi Malaysia, 81310 Johor Bahru, Johor, Malaysia N. Eliza (B) Centre for Environmental Sustainability and Water Security (IPASA), Universiti Teknologi Malaysia, 81310 Johor Bahru, Johor, Malaysia e-mail: [email protected] Rahmah Department of Mathematics, Faculty of Science, Universiti Teknologi Malaysia, 81310 Johor Bahru, Johor, Malaysia © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023 I. K. Othman et al. (eds.), Proceedings of the 5th International Conference on Water Resources (ICWR) – Volume 2, Lecture Notes in Civil Engineering 365, https://doi.org/10.1007/978-981-99-3577-2_1

3

4

U. Hani et al.

1 Introduction Sixth Assessment Report [9] has confirmed that climate change is already affecting many weather and climate extremes in every region across the globe. In fact, the evidence of observed changes in extremes such as heatwaves, heavy precipitation, droughts and tropical cyclones has been strengthened since Fifth Assessment Report (AR5). At the global scale, extreme daily precipitation events are projected to intensify by about 7% for each 1 degree of global warming. The heavy precipitation at most continents are projected to become more frequent and more intense with climate change. The associated flooding is bound to follow similar response which translates to an increase in the frequency and magnitude of floods. According to previous studies [1, 8, 16, 17], Malaysia is found to experience increasing precipitation trend which explained the high number of flood occurrences in the 21st century [17]. There are two types of floods occurred in Malaysia, namely monsoon floods and flash floods. Monsoon floods are governed by heavy and long durations of rainfall which give devastating impacts over Malaysia. Two common approaches adopted in reducing the impact of flood problems in Malaysia includes structural and non-structural measures. Structural measures such as river widening, deepening and straightening are usually high in cost. Hence, nonstructural measures such as flood modelling are introduced at preliminary stage in order to have better understanding on the flood behaviour before any structural measures are taken. Flood modelling has been widely used and there are numbers of models available in the market. Rainfall-runoff Inundation (RRI) model is one of them and it has been developed by International Centre for Water Hazard and Risk Management (ICHARM). Commonly, flood modelling requires two models, hydrological model and hydraulic model. However, with RRI the number of model requires can be reduced from two to one as the hydrological and hydraulic calculation is conducted simultaneously. Hence, less error is expected in RRI simulation. Previous studies have shown that RRI results are reasonable and acceptable [3, 12, 13]. It is important to study the impact of changing climate at basin scale as the adaptation strategies required to deal with the impacts will handled locally and regionally. Langat River Basin is selected as study area due to its location and urbanised area. The urbanization and the deforestation processes that take places in Langat River Basin may change the pervious and impervious surface area. These changes may lead to higher peak streamflow and surface runoff during heavy rainfall and eventually triggering flooding condition. In addition to climate change issues such as heavy rainfall, the frequency and severity of flood may increase in Langat River Basin. Previous studies related to flood analysis in Langat River Basin [7, 11] does not includes any flood modelling analysis. Therefore, finding the flood response to the changes in climate using flood modelling is essential in order to minimize the flood hazard in the future. This study is aimed to have better knowledge on the impact of climate change on flood characteristics at Langat River Basin using Rainfall-Runoff Inundation (RRI) model. In order to incorporate climate change in the simulation, data from climate

Assessing the Impact of Climate Change on Flood Characteristics …

5

model is going to be adopted. The rainfall intensities for 100-year flood is going to represent the extreme event and assessment on the climate change impact is made based on the comparison of both observed and projected simulation.

2 Materials 2.1 Study Area Langat River Basin is located in Selangor, Malaysia. It is an urbanised catchment in which capital city of Kuala Lumpur is located (Fig. 1). The total catchment area is approximately 1,815 km2 with the length of main river is 141 km. Langat River Basin can be divided into three categories namely upstream, middle stream and downstream. There are two reservoirs within the study area, which is Langat dam and Semenyih dam. The average annual rainfall at Langat River Basin is around 2500 mm with the highest recorded monthly rainfall is in the month of November, approximately 694 mm. Generally, Langat River Basin receives high amount of rainfall which can lead to extreme streamflow and flooding.

Fig. 1 Study area showing the location of rainfall stations, streamflow station and 5 km × 5 km grid point of NHRCM

6

U. Hani et al.

2.2 Data Collection One of the important parameters required for this study are rainfall and river flowrate. The observed data were provided by Department of Drainage and Irrigation (DID) Malaysia to represent present condition while the projected future climate rainfall data were extracted from Non-Hydrostatic Regional Climate Model (NHRCM). NHRCM was developed by the Meteorological Research Institute (MRI) and has a resolution of 5 km. NHRCM was selected for future projection due to its high resolution in which the topography, coastlines and landmass distribution may be adequate in resolving local climate. Besides, the projected data has been localised for Southeast Asia domain which is suitable for Malaysia. The rainfall data covered from 1980–2002 is considered as present condition and 2080–2099 as future condition. Figure 1 shows the location of five rainfall stations, one streamflow station and the 5 km × 5 km grid point of NHRCM. The NHRCM data were derived at the same location as observed stations and were labelled as GP representing grid point in Fig. 1. These data were then bias-corrected as described in [2]. The maximum daily rainfall intensities for flood with magnitude of 100-year is selected from both observed and projected data to represent extreme event. The topography data needed in RRI model is digital elevation model (DEM), flow accumulation (ACC) and flow direction (DIR). These data can be downloaded from the website of USGS HydroSHEDS, which is a global scale dataset offered by the United States Geological Survey (USGS). In this study, the 15 second resolution (approximately 450 m) of Langat River Basin data were used. Landuse data were obtained from PLANMalaysia. The data were assigned into five categories, namely agriculture, forest, mangrove, urban/developed area and water body as illustrated in Fig. 2. Besides that, surveyed river cross-sections of Langat River Basin were also obtained from DID Malaysia. This information is required for determining the river cross sectional parameters (Cw , Sw , Cd , and Sd ) in RRI model.

3 Methods 3.1 RRI Model Setting RRI model is a two-dimensional (2D) model with the ability to simulate rainfallrunoff and flood inundation simultaneously [14]. It simulates flows on land and in river as well as their interactions at a river basin scale. The flow on the land grid cell is calculated with the 2D diffusive wave model while the flow inside the river channel is calculated using 1D diffusive wave model. In order to represent better rainfall-runoff-inundation processes, RRI model simulates lateral subsurface flow at mountainous area and vertical infiltration at flat areas.

Assessing the Impact of Climate Change on Flood Characteristics …

7

Fig. 2 Landuse map of Langat River Basin on 2010

RRI model is calibrated from the initial ranges of parameters values. The values are set based on suggestion from previous model applications [13–15] and determined after a trial-and-error approach. The model parameters used in this study are summarized in Table 1. The Manning’s roughness in river channel (nr ) was set at 0.04 m−1/3 s. Model calibration was done based on flood event from 28th February 2012 to 13th Mac 2012, while the validation is from 27th April 2012 to 12th May 2012. The rainfall input used in this study is the maximum daily rainfall intensities for 100-year ARI. The calculated maximum daily rainfall intensities of each station for present and future condition is tabulated in Table 2. The 24-h temporal distribution of rainfall designed for Selangor is adopted in this study [5].

8

U. Hani et al.

Table 1 RRI parameters Forest

Mangrove

Urban

Water body

Manning’s roughness on 0.7 slope cells, ns (m−1/3 )

0.7

0.7

0.7

0.7

Soil depths, d (m)

2.0

2.0

2.0

2.0

2.0

Effective porosity, F

0.464

0.475

0.430

0.475

0.475

Parameters

Agriculture

Green Ampt Infiltration Model Parameters Vertical saturated hydraulic conductivity, kv (m/s)

5.56e−7

0

3.33e−7

0

0

Suction at the wetting front, Sf (m/s)

0.2088

0.3163

0.239

0.3163

0.3163

0

0

0

Lateral subsurface and surface model parameters Lateral saturated hydraulic conductivity, ka (m/s)

0

Table 2 Present and future maximum daily rainfall intensities of 100-year ARI

1.67e−3

Station name

Present

Future

% Difference

2,913,001

458.71

270.53

−41.0

2,815,001

132.07

123.67

−6.4

2,818,110

137.54

144.75

5.2

2,917,001

214.83

188.27

−12.4

3,118,102

264.42

234.54

−11.3

4 Results and Discussion 4.1 Calibration and Validation Figure 3 shows the model calibration and validation results of comparative streamflow for 2012 flood event at Station No. 2917401. Referring to the Fig. 3(a), it is clearly seen that the pattern of the simulated streamflow follows the observed streamflow. However, RRI model is found to simulate lower peak flow when compared to the observed. The underestimation of peak flow is a common issue in hydrological modelling [10, 18] especially when it comes to extreme flow [4]. This could due to errors during the estimation of RRI model parameters. In order to improve this, it is advisable to take the peak flow into account during the calibration process. Nevertheless, the performance of calibration is considered good based on the evaluation statistics of NSE = 0.81 and R2 = 0.90. Figure 3(b) shows the streamflow in the validation process. The result is the same as in calibration process. The values of NSE and R2 for validation is 0.64 and 0.82 respectively. Even though the performance of validation is low in NSE, the R2

Assessing the Impact of Climate Change on Flood Characteristics …

Fig. 3 (a) Calibration and (b) validation of RRI model

9

10

U. Hani et al.

shows acceptable values which can be explained that both simulated and observed streamflow correlated to each other.

4.2 Impact of Climate Change on Rainfall Intensities The maximum daily rainfall intensities for 100-year ARI is shown in Table 2 for present and future conditions. Overall, the maximum daily rainfall intensities at Langat River Basin is expected to decrease in the future due to climate change, particularly at downstream area as represent by Station No. 2913001. The amount of maximum daily rainfall intensities was reduced to almost half from the present rainfall intensities with 40% decrement. Nevertheless, the rainfall intensities at other parts are expected to reduce by 6.4% up to 12.4%. Meanwhile, at Station No. 2818110 which located at Semenyih River is found to experience some increment of maximum daily rainfall intensities at 5.2%. Generally, this finding is consistent with previous study conducted by NAHRIM (2006), in which a decrement in mean monthly rainfall is expected in Selangor, includes Langat River Basin. Based on this outcome, the associates streamflow and flood risk are predicted to experience similar response in the future.

4.3 Impact of Climate Change on Peak Streamflow The hydrograph of 100-year ARI for present and future condition is plotted in Fig. 4. The hydrograph was taken at the Station No. 2917401. The peak streamflow for present and future condition were 244.4 and 227.7 m3 /s respectively. [2] did a study on impact of climate change to the streamflow of Langat River Basin and predicted that the peak streamflow can goes up to 183 m3 /s. The reason of such difference is mainly due to the definition of extreme event. In this study, a 100-year ARI is used to represent extremely large event which theoretically to happen only once in every century. Previous studies showed that Langat River Basin is experience an increasing trend of streamflow [6, 18]. However, based on Fig. 4, it is clearly shown that the peak streamflow of future condition is lower than the present which indicates a decreasing trend. The percentage difference is about 6.8%. This small values depicted that climate change actually gives lower impact to the streamflow of Langat River Basin. However, there is uncertainty in the result since only one dataset of RCM is included in this study. Therefore, in order to consider the uncertainty, it is suggested to include more datasets from other RCMs/GCMs for comparison purposes.

Assessing the Impact of Climate Change on Flood Characteristics …

11

Fig. 4 Comparison of present (1980–2002) and future (2080–2099) peak streamflow for 100-year ARI at Station No. 2917401 in Langat River Basin

4.4 Impact of Climate Change on Flood Characteristics Besides streamflow, RRI also simulate flood depth and inundation area. Figure 5 illustrates the distribution of peak flood depth in Langat River Basin at 100-year ARI for present and future condition while Fig. 6 shows the difference between both conditions. Based on Fig. 5, it is found that the maximum flood depth for present and future condition is 1.75 and 1.73 m respectively. In the meantime, the flood depth at Station No. 2917401 is predicted to achieve up to 1.02 m in present time and reduce to 0.84 m in the future. These values show a decrement in future flood depth at about 17.7%. According to Fig. 6, the difference between present and future condition at most part of areas in Langat River Basin is 3% only. The high level of flood risk (peak depth: 1.16 to 1.75 m) area will remain the same in the future except at the downstream part of Langat River Basin. This is because the future rainfall at downstream part is expected to reduce at about 41% in the future. Based on these findings, it can be inferred that Langat River Basin is getting drier in the future due to climate change. However, based on the small percentage difference, the changes are considered low.

12

U. Hani et al.

Fig. 5 Distribution of peak flood depth at 100-year ARI for (a) present (1980–2002) and (b) future (2080–2099) conditions in Langat River Basin

Assessing the Impact of Climate Change on Flood Characteristics …

13

Fig. 6 The difference of peak flood depth at 100-year ARI between present (1980–2002) and future (20,080–2099) conditions in Langat River Basin

5 Conclusion Langat River Basin is an urbanized basin which experienced rapid development in agriculture, urbanization and industrialization. These changes together with climate change will possibly affect the streamflow and flood risk within Langat River Basin. In order to includes climate change, this study has adopt projected rainfall dataset derived from Non-Hydrostatic Regional Climate Model (NHRCM) developed by Meteorological Research Institute (MRI) into RRI model. The maximum daily rainfall intensities for flood with magnitude of 100-year is selected to represent extreme event. RRI model produce outputs in forms of streamflow, flood depth and inundation area. It is found that most of the part at Langat River Basin are expected to getting drier by 6.4% to 12.4% in the future. This has cause the streamflow in Langat River to reduce by 6.8% and peak flood depth changes by ±3%. Nevertheless, an increasing sign of rainfall is expected at Semenyih River by 5.2% and the associates streamflow as well as flood depth are bound to follow. Generally, based on these small number of percentage difference, it can be concluded that climate change gives lower impact to the streamflow and flood depth. However, this study only includes one RCM dataset. In order to reduce the uncertainty of one dataset, it is suggested to include more datasets from other RCMs/GCMs for comparison purposes.

14

U. Hani et al.

Acknowledgements. The authors would like to state their appreciation to the Universiti Teknologi Malaysia and the Malaysian Ministry of Education under the GUP2018 [QJ130000.2522.19H68], UTM Fundamental Research Grant [Q.J130000.2551.20H75] and Fundamental Research Grant Scheme (FRGS) (FRGS/1/2018/WAB05/UTM/02/6) [R.J130000.7851.5F032]. The authors also thank the Department of Irrigation and Drainage Malaysia (DID) and PLANMalaysia for providing data of Langat River Basin. Gratitude is also for JASTIP fund support. The JASTIP research was supported by Japan Science and Technology Agency (JST), Collaboration Hubs for International Research Program (CHIRP) within the framework of the Strategic International Collaborative Research Program (SICORP).

References 1. Amin IMZBM, Ercan A, Ishida K, Kavvas ML, Chen ZQ, Jang S-H (2019) Impacts of climate change on the hydro-climate of peninsular Malaysia. Water 11(9):1798 2. Anuar UHM, Alias NE (2021) Modelling the impact of climate change on the streamflow of Langat River Basin. In: Alias NE, Haniffah MRM, Harun S (eds) Water Management and Sustainability in Asia. Emerald Publishing Limited, pp 65–76. https://doi.org/10.1108/S2040726220210000023013 3. Bhagabati SS, Kawasaki A (2017) Consideration of the rainfall-runoff-inundation (RRI) model for flood mapping in a deltaic area of Myanmar. Hydrol Res Lett 11(3):155–160 4. Chen X, Yang T, Wang X, Xu CY, Yu Z (2013) Uncertainty intercomparison of different hydrological models in simulating extreme flows. Water Resour Manage 27(5):1393–1409 5. DID (2011) Urban stormwater management manual for Malaysia (MSMA 2nd Edition). DID, Kuala Lumpur 6. Ebrahimian M, Nuruddin AA, Soom MAM, Sood AM, Neng LJ, Galavi H (2018) Trend analysis of major hydroclimatic variables in the Langat River basin Malaysia. Singapore J. Tropic. Geography 39(2):192–214 7. Huang Y, Mirzaei M, Yap WK (2016) Flood analysis in langat river basin using stochatic model. Int J Geomate 11:2796–2803 8. Hussain M, Yusof K, Mustafa M, Mahmood R, Shaofeng J (2017) Projected changes in temperature and precipitation in Sarawak state of Malaysia for selected CMIP5 climate scenarios. Int J Sustain Dev Plan 12:1299–1311 9. IPCC (2021) Summary for policymakers. In: Climate change 2021: the physical science basis. contribution of working group I to the sixth assessment report of the intergovernmental panel on climate change. Cambridge University Press. In Press 10. Lane RA, Coxon G, Freer JE, Wagener T, Johnes PJ, Bloomfield JP, Greene S, Macleod CJ, Reaney SM (2019) Benchmarking the predictive capability of hydrological models for river flow and flood peak predictions across over 1000 catchments in Great Britain. Hydrol Earth Syst Sci 23(10):4011–4032 11. Mirzaei M, Faghih M, Ying TP, El-Shafie A, Huang YF, Lee J (2016) Application of a rainfallrunoff model for regional-scale flood inundation mapping for the Langat River Basin. Water Pract Technol 11:373–383 12. Nastiti KD, Kim Y, Jung K, An H (2015) The application of rainfall-runoff-inundation (RRI) model for inundation case in upper Citarum watershed, West Java-Indonesia. Procedia Eng 125:166–172 13. Sayama T, Ozawa G, Kawakami T, Nabesaka S, Fukami K (2012) Rainfall–runoff–inundation analysis of the 2010 Pakistan flood in the Kabul River basin. Hydrol Sci J 57:298–312. https:/ /doi.org/10.1080/02626667.2011.644245 14. Tam TH et al (2021) The effects of climate change on flood hazards in Kelantan River Basin Malaysia. In: IOP conference series: earth and environmental science, vol 880, no 1, p 012016. IOP Publishing

Assessing the Impact of Climate Change on Flood Characteristics …

15

15. Tan ML, Ibrahim AL, Yusop Z, Chua VP, Chan NW (2017) Climate change impacts under CMIP5 RCP scenarios on water resources of the Kelantan River Basin, Malaysia. Atmos Res 189:1–10 16. Tang KHD (2019) Climate change in Malaysia: Trends, contributors, impacts, mitigation and adaptations. Sci Total Environ 650:1858–1871 17. Wijayarathne DB, Coulibaly P (2020) Identification of hydrological models for operational flood forecasting in St. John’s, Newfoundland, Canada. J Hydrol Region Stud. 27: 100646 18. Yusoff SHM, Hamzah FM, Jaafar O, Tajudin H (2021) Long term trend analysis of upstream and middle-stream River in Langat Basin, Selangor Malaysia. Sains Malaysiana 50(3):629–644

Analysing Impact of Climate Change on Hydrological Trend in Kelantan River Basin Using HEC-HMS Coupled with SDSM Muhammad Zahran Syahmi Armain, Zulkarnain Hassan, Mohd Remy Rozainy Mohd Arif Zainol, Sobri Harun, Ain Nihla Kamarudzaman, and Salwa Mohd Zaini Makhtar

Abstract Climate change dramatically alters many hydrologic systems, which affects the availability of water and leads to runoff and river discharge. This study assessed the effects of the future scenario of climate change on the monthly river discharge of the Kelantan River Basin, Malaysia. Statistical DownScaling model (SDSM) was used to downscale the rainfall from large climate variables of the second-generation Canadian Earth System Model (CanESM2) under the Representative Concentration Pathways of 8.5 (RCP 8.5) and project future river discharge using the Hydrologic Modeling System (HEC-HMS). From this study, the monthly rainfall and river discharge over the Kelantan River basin will be significantly reduced in the future by 30 and 50% compared to the current period. Keywords Rainfall · Hydrology · Statistical downscaling model · HEC-HMS · CanESM2 · Kelantan River Basin M. Z. S. Armain · Z. Hassan (B) · A. N. Kamarudzaman · S. M. Z. Makhtar Faculty of Civil Engineering Technology, Universiti Malaysia Perlis, Kompleks Pusat Pengajian Jejawi 3, 02600 Arau, Perlis, Malaysia e-mail: [email protected] A. N. Kamarudzaman e-mail: [email protected] S. M. Z. Makhtar e-mail: [email protected] M. R. R. M. A. Zainol School of Civil Engineering, Universiti Sains Malaysia, 14300 Nibong Tebal, Pulau Pinang, Malaysia e-mail: [email protected] S. Harun School of Civil Engineering, Faculty of Engineering, Universiti Teknologi Malaysia, 81310 Johor Bahru, Johor, Malaysia e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023 I. K. Othman et al. (eds.), Proceedings of the 5th International Conference on Water Resources (ICWR) – Volume 2, Lecture Notes in Civil Engineering 365, https://doi.org/10.1007/978-981-99-3577-2_2

17

18

M. Z. S. Armain et al.

1 Introduction For many decades, human activities have caused continuous greenhouse gas emissions to affect the global atmosphere. Continuous greenhouse gas emissions at or above present levels trigger significant warming and lead to various changes in the global climate system [6]. Thus, climate change may impact regional hydrological processes, long-term water supply, and the frequency of floods or drought [5]. Many studies have looked into the effects of climate change on regional hydrological processes, such as studies by [7] and [8]. They reported that climate change (rainfall) fluctuations influenced runoff. In climate change impact studies, a combination of General Circulation Model (GCM) outputs and hydrologic models has been used over the past decade to assess future hydrologic changes in drainage basins. However, their utility in regional applications, such as river flow simulations in conjunction with hydrologic models, is constrained by their coarse spatial resolution [11]. Nevertheless, the gap scales between these models can be reduced using downscaling approaches. This study aims to assess the trend of the discharge projection that correspond to climate scenarios of the GCM output over the Kelantan basin, Malaysia, using the integration between the Statistical DownScaling Model (SDSM) and Hydrologic Modelling System (HEC-HMS) models. The materials and methods are introduced in the following sections, followed by the findings and discussions of the study. Then, the conclusions are presented.

2 Materials and Methods A 248-km-long basin in Malaysia originating from the mountains Titiwangsa and Tahan, the Kelantan River Basin has been delineated as a study area covering 12,134 km2 of the catchment area, comprising five major sub-basins as shown in Fig. 1. The frequent unpredictable high magnitude of discharge that occurs downstream of this basin provides need for it to be assessed in terms of the climate variability that occurs in the area [2]. Thus, this study can provide insights into relevant water resource considerations. In this study, the second-generation version of the Canadian Earth System Model (CanESM2) was used. The CanESM2 data for Representative Concentration Pathways scenarios of 8.5 (RCP 8.5) was downloaded from the Canadian Climate Impacts Scenarios (CCIS) website, including the CanESM2 which are re-analysed data predictors of the National Center of Environmental Prediction (NCEP). For the rainfall and discharge data (Table 1), the data was obtained from the Department of Irrigation and Drainage Malaysia. The study was performed using two models, which were the SDSM and HECHMS as the downscaling and hydrological models, respectively. The downscaling process of GCM (CanESM2) for current and future periods using SDSM can be

Analysing Impact of Climate Change on Hydrological Trend …

19

Fig. 1 Illustration of the study area with rainfall and discharge stations Table 1 The description of the selected rainfall and discharge stations No

Station type

Station name

1

Rainfall

Brook

Coordinate latitude (° ‘ ‘’ N)

Longitude (° ‘ ‘’ E)

04 40 35

101 29 05

2

Redip

04 49 00

101 59 00

3

Kg. Aring

04 56 15

102 21 10

4

Gob

05 15 05

101 39 45

5

Bertam

05 08 45

102 02 55

6

Kg. Lebir

05 12 45

102 18 15

7

SK Lubok Bungor

05 33 40

101 53 20

8

Ldg. Lepan Kabu

05 27 35

102 13 50

9

Machang

05 47 15

102 13 10

Nenggiri

05 08 55

102 02 45

10

Discharge

11

Galas

05 22 55

102 00 55

12

Lebir

05 16 30

102 16 00

13

Guillemard Bridge

05 45 45

102 09 00

20

M. Z. S. Armain et al.

simplified into three main phases. The first is the screening process between the inputs of the SDSM model. Rainfall data sets were utilised as predictands, and NCEP-reanalysis and RCP 8.5 (extreme high scenario) were used as predictors. Partial correlation and P-value analysis were done between selected predictors to choose the predictor data sets. Secondly, SDSM utilised multi-linear regression, a process known as calibration followed by a validation period to adjust the predictor relationship. The study employed daily rainfall from 1990–2005 for calibration and validation from 2006–2018. Thirdly, rainfall data was generated for each station daily, including the future trend based on selected scenarios. Finally, daily rainfall series were used to compute monthly and yearly rainfall. The details of the SDSM model can be referred to [4]. Meanwhile, the hydrological analysis used in this study also consisted of three main phases by following the study from [3]. The first was the development of the basin model. This phase was used to depict the physical features of the selected basin. This phase was also to measure the parameters using GIS or by observing the basin characteristics in the field. The parameters include sub-basin area (km2 ), flow length (km), slope, and impervious surface (%). Secondly, the basin model must be assigned with the hydrological parameter (inferred parameters) based on the investigated criteria of the basin. These parameters cannot be measured or inferred from field observations as it needs to be calculated using a mathematical model such as the loss model; i.e. initial abstraction (Ia ), SCS CN, percent of impervious surface (Imp ), and transform model; lag time (Lt ). Thirdly is the phase of calibration and validation of the developed HEC-HMS model where the applied parameter values were evaluated. The HEC-HMS model used in this study was calibrated using flood events from 12–18 January 2012 and 08–17 January 2014, and the parameters utilised for both events were averaged to be validated on the 01–08 December 2013 flood event. The validated HEC-HMS model then used synthetic daily rainfall from the SDSM projection to project hydrological variables for two-time slices of the 2050s (2041–2070) and 2080s (2071–2100).

3 Performance of SDSM as a Downscaling Model Table 2 shows how the SDSM model performed while downscaling the GCMs during the model development process. The performance of the model was evaluated by using the most commonly used performance indicator, which was the coefficient of determination, R2 and can be written as follows:  (y − x)2 R = (x − x)2 2

(1)

Analysing Impact of Climate Change on Hydrological Trend … Table 2 The performance of the SDSM model during the calibration and validation periods for monthly rainfall series

21

Station name

Calibration (R2 )

Validation (R2 )

Brook

0.982

0.916

Gua Musang

0.896

0.927

Kg. Aring

0.975

0.966

Gob

0.908

0.951

Bertam

0.947

0.876

Kg. Laloh

0.968

0.919

Kuala Krai

0.979

0.976

Kg. Jeli

0.980

0.960

Machang

0.996

0.972

where, x is the observed rainfall value, x is the mean of observed rainfall value, y is the simulated rainfall value and n is the number of rainfall values. A value that is closer to 1.0 means a perfect agreement. In general, the model performed admirably during the calibration and validation periods, with mean monthly R2 values ranging between 0.896–0.996 and 0.876–0.976, respectively, in each rainfall station distributed along the Kelantan basin. Hence, the results reveal that the SDSM technique successfully replicated the observed historical rainfall during the model development period.

4 Generating Future Rainfall Corresponding to RCP 8.5 using the SDSM Model The rainfall projection in the Kelantan basin was covered for 60 years, from the years 2041 to 2100. The 60 years were then divided into the 2050s (2041–2070) and 2080s (2071–2100) to analyse future patterns. Figure 2 displays the results of future emission downscaling. SDSM generated the ensembles of synthetic daily time series based on the RCP 8.5 scenario. From the figure, there was a remarkable difference in terms of monthly rainfall between the current and projected numbers. The result shows that rainfall decreases in all projected time frames in overall months except February and June to September. The most significant reduction of rainfall can be seen for the months of October to December. These findings are not consistent with the projected monthly rainfall of the basin reported by [9], in which their finding shows that the monthly rainfall increases in January and December.

Monthly Rainfall (mm)

Fig. 2 Average annual rainfall at the a) 2050s and b) 2080s. Current rainfall is from 2006 until 2018

M. Z. S. Armain et al.

350 300 250 200 150 100 50 0

Monthly Rainfall (mm)

22

350 300 250 200 150 100 50 0

a)

Current RCP 8.5

1 2 3 4 5 6 7 8 9 10 11 12 Month b)

Current RCP 8.5

1 2 3 4 5 6 7 8 9 10 11 12 Month 5 Performance of HEC-HMS as Hydrological Model Table 3 shows the calibrated parameter values for each of the components represented in the HEC-HMS model. Aside from adjusting the model basin, the parameters were calibrated simultaneously by adjusting their values until this study attained a good agreement between the observed and simulated hydrographs. Two flood events were chosen, which were in 2012 and 2014, as the HEC-HMS model calibration period. The parameters from these two events were needed to obtain the value of the average parameters, which was then used during the 2013 validation period. This is because the physical characteristics of a watershed or basin will change over time, and thus the measured and inferred parameters also tends to change [1]. The calibration and validation hydrographs for the three flood events at the last discharge station (Guillemard Bridge) is shown in Fig. 3. This station was given focus because the Guillemard Bridge is located at the downstream area which is a hotspot for flooding to occur. R2 and Percent Error in Peak Flow (PEPF) as displayed

Analysing Impact of Climate Change on Hydrological Trend …

23

Table 3 Calibrated and validated parameter values for three flood events Sub-basin

Area (km2 )

Ia (mm)

CN

Imp (%)

Lt (min)

Calibrated 2012 Nenggiri

3708.7

0.66900

21

0.034

1100

Lebir

1243.1

0.49331

32

0.063

780

Galas

2261.1

0.78522

72.113

0.030

500

Kuala Krai

2392.0

0.25066

60

0.910

1320.8

Guillemard Bridge

2323.7

0.57430

74

1.000

1850

Calibrated 2014 Nenggiri

3708.7

0.66900

21

0.034

1100

Lebir

1243.1

0.49331

32

0.063

780

Galas

2261.1

0.78522

72.113

0.030

500

Kuala Krai

2392.0

0.25066

60

0.910

1320.8

Guillemard Bridge

2323.7

0.57430

74

1.000

1850

Validated 2013 Nenggiri

3708.7

0.66900

21

0.034

1100

Lebir

1243.1

0.49331

32

0.063

780

Galas

2261.1

0.78522

72.113

0.030

500

Kuala Krai

2392.0

0.25066

60

0.910

1320.8

Guillemard Bridge

2323.7

0.57430

74

1.000

1850

in Eq. (2) were used to evaluate the performance of the HEC-HMS model, where Q o is the observed discharge and Q m is the modelled discharge.    Q o ( peak) − Q m ( peak)    P E P F = 100  Q ( peak)

(2)

o

In general, the HEC-HMS model obtained a good agreement between the simulated and observed discharge during the calibration and validation periods, as shown in Table 4. During the calibration period in 2012 and 2014, the table shows that the model can capture the observed discharge with R2 and PEPF between 0.87–0.987 and 0.42–20.25%, respectively. However, the performance of the model was slightly reduced during the validation period, in which R2 and PEPF were in the ranges of 0.714–0.915 and 2.19–19.72%. The performance of the HEC-HMS model was also checked by comparing between observed and simulated hydrograph, as illustrated in (Fig. 3). The figure shows that the peak discharge occurred on the same day with a maximum time difference of one hour faster during calibration and three hours faster during validation, which for flood simulation was considered good. The simulated discharge managed to follow exactly the observed discharge pattern with a slight difference in terms of the peak and volume of the discharge. Hence, the results demonstrate

24

Discharge (m3/s)

5000 4000

Observed

3000 2000 1000

4000 Discharge (m3/s)

Calibrated

a)

0 12-1-12

b)

14-1-12 16-1-12 Date

18-1-12

Calibrated Observed

3000 2000 1000

0 9-1-14 11-1-14 13-1-14 15-1-14 Date

7000 6000 Discharge (m3/s)

Fig. 3 Simulated vs. observed discharge in a) 2012 calibration, b) 2014 calibration and c) 2013 validation at Guillemard Bridge.

M. Z. S. Armain et al.

c)

Validated Observed

5000 4000 3000 2000 1000 0 2-12-13 4-12-13 6-12-13 8-12-13 Date

Analysing Impact of Climate Change on Hydrological Trend …

25

Table 4 Statistical model performance during calibration and validation Sub-basin

Calibration event 1

Calibration event 2

Validation

R2

R2

PEPF (%)

R2

PEPF (%)

PEPF (%)

Nenggiri

0.978

2.84

0.949

7.57

0.720

2.19

Galas

0.937

20.25

0.870

1.31

0.714

1.92

Lebir

0.947

5.48

0.960

0.42

0.811

3.45

Guillemard Bridge

0.987

5.90

0.977

3.81

0.915

19.72

that the developed HEC-HMS model successfully approached historically observed discharge and is reliable for further simulation.

6 Hydrological Trend in Kelantan Basin Based on Climate Change Following the developed hydrological model after calibration and validation with HEC-HMS, as previously stated, the discharge simulation based on climate change was carried out. HEC-HMS uses generated future rainfall computed by the SDSM model that has been well validated as described above to predict future discharge. In order to analyse the simulated streamflow, two periods were considered: the 2050s (2041–2070) and 2080s (2071–2100). Figure 4 displays the projected percent change in the average monthly discharge at Guillemard Bridge under the RCP 8.5 relative to baseline discharge. The projected discharge shows a decline in almost all of the months. A significant decline of discharge at the early and end of the month showed above 50% in the 2050s, while there was an increasing tendency in the discharge for September and October for both periods. The decreasing trend continued to decline until the 2080s in the same month. Similarly, Tan et al. [10] also reported discharge reduction in these months according to their discharge projection using the Regional Climate Model (RCM). From this point, the study area will be more prone to the water crisis in the future due to the decreasing discharge trend projection. However, there is insufficient evidence to predict that hydrological drought will happen in the near future.

M. Z. S. Armain et al.

Monthly Discharge (%)

26

2050s

40 20 0 -20 -40 -60 -80

2080s

Month Fig. 4 Projected percent change in monthly discharge at the Guillemard Bridge under RCP 8.5 scenario.

7 Conclusion This study aimed to examine the effects of climate change (rainfall) on the hydrology of the Kelantan River Basin based on the developed hydrological model coupled with the downscaling model. The SDSM and HEC-HMS models successfully simulated the historical monthly rainfall and hourly discharge, respectively. In general, the significant declining trend in monthly rainfall projection resulted in a major change to discharge decrement, which affects the future water balance in the basin area. Hence, the trend of projected discharge in the Kelantan basin is anticipated to cause a water crisis or even drought. This study is a preliminary result of the climate assessment over the Kelantan River Basin. Several GCM data sets are proposed to evaluate the uncertainty and potential impact of climate change on rainfall projection over the basin. Acknowledgements This study is financed by the Fundamental Research Grant Scheme (FRGS) with the grant number of FRGS/1/2019/TK01/UNIMAP/02/4. Providence of data for this study from DID and the Canadian website is greatly appreciated.

References 1. Aboelnour MA, Engel BA, Frisbee MD, Gitau MW, Flanagan DC (2021) Impacts of watershed physical properties and land use on baseflow at regional scales. J Hydrol Region Stud 35:100810 2. Armain MZS, Rozainy MMR, Kamarudzaman AN (2021) Hydrodynamic modelling of historical flood event using one dimensional HEC-RAS in Kelantan basin, Malaysia. In: IOP Conference Series: Earth and Environmental Science, vol 920, no 1, p 012031. IOP Publishing

Analysing Impact of Climate Change on Hydrological Trend …

27

3. Hamdan ANA, Almuktar S, Scholz M (2021) Rainfall-runoff modeling using the HEC-HMS model for the Al-Adhaim river catchment, northern Iraq. Hydrology 8(2):58. https://doi.org/ 10.3390/hydrology8020058 4. Hassan Z, Shamsudin S, Harun S (2014) Application of SDSM and LARS-WG for simulating and downscaling of rainfall and temperature. Theoret Appl Climatol 116(1):243–257 5. Jiménez-Navarro IC, Jimeno-Sáez P, López-Ballesteros A, Pérez-Sánchez J, Senent-Aparicio J (2021) Impact of climate change on the hydrology of the forested watershed that drains to Lake Erken in Sweden: an analysis using SWAT+ and CMIP6 scenarios. Forests 12(12):1803 6. Li Y, Li Z, Zhang Z, Chen L, Kurkute S, Scaff L, Pan X (2019) High-resolution regional climate modeling and projection over western Canada using a weather research forecasting model with a pseudo-global warming approach. Hydrol Earth Syst Sci 23(11):4635–4659 7. Liu Y, Huang Y, Liu Y, Li K, Li M (2021) The impact of rainfall movement direction on urban runoff cannot be ignored in urban hydrologic management. Water 13(20):2923 8. Singh NK, Emanuel RE, McGlynn BL, Miniat CF (2021) Soil moisture responses to rainfall: implications for runoff generation. Water Resourc Res 57(9). https://doi.org/10.1029/2020WR 028827 9. Tan ML, Yusop Z, Chua VP, Chan NW (2017) Climate change impacts under CMIP5 RCP scenarios on water resources of the Kelantan River Basin, Malaysia. Atmos Res 189:1–10 10. Tan ML, Juneng L, Tangang FT, Samat N, Chan NW, Yusop Z, Ngai ST (2020) SouthEast Asia HydrO-meteorological droughT (SEA-HOT) framework: a case study in the Kelantan River Basin Malaysia. Atmosph Res 246(105155):1–12 11. Wang Y, Cao J, Liu Y, Zhu Y, Fang X, Huang Q, Chen J (2022) Spatiotemporal analysis of soil moisture variation in the Jiangsu Water Supply Area of the South-to-North Water Diversion using ESA CCI data. Remote Sens 14(2):256

Machine Learning Algorithms with Hydro-Meteorological Data for Monthly Streamflow Forecasting of Kurau River, Malaysia Muhammad Nasir Mohd Adib and Sobri Harun

Abstract Monthly streamflow forecasting is crucial in water resources management to assess the possible future streamflow patterns. It becomes vital where streamflow of Kurau River is the primary water source to irrigate the large-scale rice scheme of Kerian, Perak, coupled with future climate change uncertainty. In this context, machine learning algorithms have received outstanding attention due to their high accuracy in forecasting through high-speed input–output data processing of selflearning from physical processes. In this study, two machine learning algorithms, support vector regression (SVR) and random forest (RF), were considered to forecast the streamflow of Kurau River in Malaysia using gauged hydro-meteorological dataset for the period from 1976 to 2005. The predictions of monthly streamflows were based on hydro-meteorological data such as rainfall, minimum and maximum temperature, relative humidity, and wind speed. A comparative study is executed to evaluate the efficiency of SVR and RF in performing the streamflow predictions of Kurau River. The results show that RF outperformed the SVR in both the training and testing phases. The results have proven that machine learning algorithms, especially the RF model, can be implemented for forecasting streamflow by using only hydrometeorological data with high accuracy, which will improve future water resources management. Keywords machine learning · support vector regression · random forest · forecasting · streamflow

M. N. M. Adib (B) Faculty of Civil Engineering, Universiti Teknologi Malaysia, 81310 Johor Bahru, Johor, Malaysia e-mail: [email protected] Department of Biological and Agricultural Engineering, Faculty of Engineering, Universiti Putra Malaysia, 43400 UPM Serdang, Selangor, Malaysia S. Harun Faculty of Civil Engineering, Universiti Teknologi Malaysia, 81310 Johor Bahru, Johor, Malaysia e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023 I. K. Othman et al. (eds.), Proceedings of the 5th International Conference on Water Resources (ICWR) – Volume 2, Lecture Notes in Civil Engineering 365, https://doi.org/10.1007/978-981-99-3577-2_3

29

30

M. N. M. Adib and S. Harun

1 Introduction Streamflow forecasting is crucial for water resources management for efficient irrigation practices, flood risk assessment, and hazard management in most parts of the world [2, 7, 12, 27]. In terms of the agriculture sector, especially rice cultivation, forecasting streamflow could reduce the risk of disaster by preserving water availability for use during dry periods and securing food security [2]. In Malaysia, rice cultivation practices double-cropping with constant irrigation supply to ensure the sustainability of rice production. The irrigation supply mainly depends on rivers that consume 80% of water resources associated with randomness and uncertainty resources, giving difficulty in allocating irrigation amount to the rice fields. To have a better irrigation management decision, accurate prediction of streamflow is mandatory [9]. Due to this, massive studies have been done to evaluate and improve various streamflow prediction models. Understanding the complicated phenomena of possible future streamflow could explain how the global climate changes would influence the hydrologic cycle resulting in uncertainty in water resource systems [34]. Interpret different hydrological processes for forecasting can be executed using different types of hydrological models, categorized as physically-based, conceptual, and empirical models classified based on model parameters [1]. These models work by investigating the relationship between water, soil, topography, land use/land cover, geology, groundwater aquifer, precipitation, and climatic parameters [2, 3, 8, 14]. Physically-based models are the most selected models to deal with hydrological processes, followed by empirical and conceptual models. Among the most selected models under physically-based are HEC-HMS (open-source models) [29], SWAT (public domain model) [4], and MIKE-SHE [19]. In the case of empirical models, artificial neural network (ANN), logistic regression (LR), support vector machine (SVM), autoregressive integrated moving average (ARIMA), and decision tree (DT) are among popular models [1]. Conceptual models such as TOPMODEL [26] and IHACRES [11] are the least selected models used in hydrological modeling, especially in Malaysia. Among the model categories, empirical models are inexpensive and incredibly fast data processing, relying only on historical time-series data records and mathematical equations. In contrast, physically-based and conceptual models consume a lot of input data (soil moisture content, initial water depth, topography, topology, and more of the physical features of the catchment). In recent years, machine learning techniques that belong to the empirical category have received great attention in hydrological processes, especially streamflow modeling. The ability to mimic the non-linear and complex mathematical expressions of physical processes with minimum input data has provided superior performance in prediction systems [17]. Various machine learning techniques have been applied for streamflow prediction, such as artificial neural network (ANN), multilayer perceptron (MLP), support vector regression (SVR), random forest (RF), and AdaBoost

Machine Learning Algorithms with Hydro-Meteorological Data …

31

[6, 12, 15, 18, 24, 33, 35]. These models have proven to provide outstanding performance, with SVR and RF as the leading models with a high-efficiency prediction [12, 18, 20, 28]. Kurau River is the output of the Kurau River Basin, with a total area of 322 km2 located in the northern part of Peninsular Malaysia, linked to the Bukit Merah Reservoir catchment [11]. Moreover, the Kurau River is the primary water source for the large-scale rice scheme of Kerian, Perak. Therefore, it is crucial to understand this river’s future streamflow behavior to guarantee its sustainability in providing irrigation supply for the rice scheme throughout the year. In this study, our objectives are to evaluate the performance of SVR and RF models as alternative techniques in forecasting the streamflow of Kurau River using hydro-meteorological data such as rainfall, maximum and minimum temperature, relative humidity, and wind speed.

2 Materials and Methods 2.1 Data Inventory The Kurau River, Perak, is the primary source for irrigation to the Kerian rice irrigation scheme, where its headwork is integrated with a Bukit Merah reservoir to satisfy the irrigation demands and cope with water shortages. It covers an area of 322 km2 located northern part of Peninsular Malaysia within latitude 4°51‘–5°10‘N and longitude 100°38‘–101°55‘E (Fig. 1). It experiences a humid tropical climate with an annual rainfall of around 2500 mm, mainly concentrated between April and October, and an average temperature of 27 °C to 28 °C [31].

2.2 Input Data Sets This study used historical streamflow data from Tanjung Pondok station (5007421), located at the outlet of the Kurau River, and rainfall data from three stations from Kurau River Basin, as shown and listed in Fig. 1 and Table 1. The streamflow and rainfall data were collected from the Department of Irrigation and Drainage (DID). Minimum and maximum temperatures, relative humidity, and wind speed of Ipoh station were collected from the Malaysian Meteorological Department (MMD). The climate data are the main driving factors for streamflow scenario analysis. The selected gauging station had enough data records for our study. A few missing data in the record were estimated as the long-term average for the station. The range of sequence of 30 years of data from 1976 to 2005 was selected due to less missing and reliable data. The data is partitioned into model development (training) and model evaluation (testing) parts, the first 70% of the whole dataset is used for the training phase, and the remaining 30% is used for the testing.

32

M. N. M. Adib and S. Harun

Fig. 1 Kurau River, Kurau River Basin, and hydro-meteorological stations (Adapted from [2])

Table 1 List of rainfall stations Station no

Station name

Latitude

Longitude

4,907,019 (1)

Ladang Norsman

4° 57‘55“

100° 45‘50“

4,908,013 (2)

Ibu Bekalan Sempeneh di Batu Kurau

4° 56‘05“

100° 49‘40“

4,908,018 (3)

Pusat Kesihatan Kecil di Batu Kurau

4° 58‘45“

100° 48‘15“

2.3 Support Vector Regression Support vector machine (SVM) is a recognized machine learning technique for classification and regression [30]. The SVM used for regression is known as support vector regression (SVR). SVR is constructed based on minimizing the structural risk to deal with related problems efficiently. The insensitive loss function (ε–) is the model tolerating training data errors. Hence, the SVR ε pursues a linear function as in Eq. 1 follows F(x) = w[∅(x)] + b

(1)

where w and b represent the coefficient of the weight vector, x is the input vector of the SVR, and ∅(x) is the feature space vector [22]. This can be clarified as in Eq. 2

Machine Learning Algorithms with Hydro-Meteorological Data …

33

⎧ ⎨

F(x) − yi ≤ ε + ξi∗ Subject to (2) yi − F(x) ≤ ε + ξi ⎩ ∗ ξi , ξi ≥ 0, i = 1, 2, . . . , N

  1 ξi + ξi∗ Min w2 + C 2 i=1 N

where C > 0 is a penalty parameter. The constant C can grade the experimental error. ξi and ξi∗ are slack variables, express the distance between actual values and the corresponding boundary value of ε–tube [22]. The function as in Eq. 3 is defined by simplifying Eq. 1 subject to Eq. 2 [21]. f (x) =

N   ∗  αi − αi K (x, xi ) + B

(3)

i=1

where K (x, xi ) is the kernel function, αi, αi∗ ≥ 0 are the Lagrange multipliers, and B is a bias term. Lagrange multipliers solved the optimization problem in dual form by using sequential minimal optimizer (SMO) due to the ability to provide an analytical solution for a subset without invoking a quadratic optimizer [22]. The kernel function maps low dimensional data to high dimensional data in SVR [25]. In this study, the radial basis function (RBF) that converges fast and works well in high-dimensional space is applied to build an optimum SVR model [32]. The RBF function as in Eq. 4   K (x, xi ) = ex p −γ x − xi 2

(4)

where γ is the bandwidth of the kernel function and C, γ , and ε are three predefined parameters that should be attentively regulated. The combination of parameter value sets that maximize the objective function was determined using a trial-and-error technique using a grid-search method on the training data sets. A range for the constant C between 1 and 5 with an increment of 0.1, for the parameter ε between 0.01 and 0.5, and parameter γ between 0.01 and 1 with increment 0.01 were evaluated.

2.4 Random Forest Random forest (RF) algorithm by [5] is a supervised learning model used for classification and regression. It contains multiple non-pruning classification regression trees (CART) known as combined classifiers [5]. The RF prediction can be written as in Eq. 5 by introducing independent and identically distributed random variables θ . The using training set data and θ to generate decision tree h(x, θ ). All the generated decision trees are combined by ensemble learning. For regression problem, the mean of all the predicted output from the decision trees will appear as final prediction result by RF.

34

M. N. M. Adib and S. Harun

h(X ) =

N 1  h(x, θn ) N n=1

(5)

where n = 1, …, N . N is the number of regression trees, x is the input vector. θ is an independent, uniformly distributed random vector, which is used by the prediction trees as a numerical value. In RF modeling, two hyperparameters have to be set, which are the number of trees (Ntree ) and the number of randomly selected features (Mtry ). RF algorithms are more sensitive to Mtry compared to Ntree [13]. Decreasing the Mtry parameter will reduce computation time; but, it also weakens correlations between any two trees and the strength of every tree in the forest, affecting the classification accuracy [23]. The value of Ntree is flexible since the RF classifier does not overfit due to high computational efficiency [10]. Several researchers recommended that the optimum number for the Ntree is 500 due to insignificant improvement in the accuracy for using a number of Ntree higher than 500 [10]. In contrast, the Mtry parameter is recommended to be set to one-third of the number of input features, where it depends on the data at hand [23].

2.5 Accuracy Statistics of Model Performance The performance of tested models are evaluated on training data and testing data. The commonly used evaluation criteria are coefficient correlation (R2 ) (Eq. 6), mean absolute error (MAE) (Eq. 7), and root mean square error (RMSE) (Eq. 8). The evaluation of model prediction can be classified as satisfactory if the R2 value is higher than 0.5 [16]. The various indicators for assessing the model’s performance are expressed as in Eqs. 6–8. ⎞2  M L  Obs mean mean P P − P − P ⎟ ⎜ i=1 i i i i R2 = ⎝

0.5 ⎠    n  Obs  2 2 n Obs − Pimean − PiM L i=1 Pi i=1 Pi ⎛

n 

M AE =

n  1  Obs (Pi − PiM L ) n i=1

 RMSE =

n i=1

(6)

(7) 2

(Q iobs − Q iM L ) n

(8)

In Eqs. 6–8, PiObs is calculated observed streamflow, PiM L is predicted streamflow using machine learning, P mean is mean of the observed streamflow, and n is the total number of data.

Machine Learning Algorithms with Hydro-Meteorological Data …

35

3 Results and Discussion The performance of the SVR and RF model in predicting streamflow based on hydrometeorological data is shown in Fig. 2. The figure compares the output of two generated streamflow models with observed streamflow data of Tanjung Pondok station (5007421). The model was evaluated using 30 years historical record (1976–2005) of three rainfall stations and one meteorological station. Three statistical precision indicators that are R2 , MAE, and RMSE, indicated that SVR and RF models are suitable for predicting the streamflow of the Kurau River based on the classification proposed by [16]. During the training and testing phases, the RF algorithm seems to be outperformed the SVR for all precision statistics. For RF, the R2 , MAE, and RMSE values are 0.95, 2.15 and 2.75 m3 /s for the training phase and 0.72, 5.43 and 7.17 m3 /s for the testing phase, respectively. While for SVR, the R2 , MAE, and RMSE values are 0.66, 4.17 and 6.48 m3 /s for the training phase and 0.71, 5.63, 7.34 m3 /s for the testing phase, respectively. For the training phase, the predicted flows using the RF algorithm are able to capture the peak flows very well due to the high monsoon recorded during September, October, and November, as well as the low flows. While the SVR algorithm predicted the low flows reasonably well but not for high peak flows. For the testing phase, the SVR model seems to capture low flows quite well compared to the RF model; however, in contrast, the RF model is better at predicting high flows than the SVR model. To further analyze the prediction performance, scatter plots were constructed for monthly streamflow predicted by SVR and RF models for the training and testing phases, as shown in Fig. 3a, 3b. The minor deviance from the ideal line (red line) indicates high significant prediction accuracy. The scatter plot of the SVR model falls far from the ideal line for observed and predicted streamflows, with R2 values of 0.66 (Fig. 3(a)) and 0.71 [Fig. 3(b)] for the training and testing phases, respectively. In contrast, the RF model presents minor deviance compared to the SVR model for the training phase with R2 value of 0.95 (Fig. 4(a)) and a testing phase with R2 value of 0.72 (Fig. 4(b)). These results reveal that the RF model appeared to be the best prediction model for this study, with the remarkable ability to establish the non-linear relationship between input (hydro-meteorological data) and output (streamflow data).

Fig. 2 Observed and forecasted streamflow data in training and testing phases using SVR and RF

36

M. N. M. Adib and S. Harun

The superior accuracy of the RF model algorithms is that it consists of the decision trees that continuously grow in parallel until it reaches optimal results. The plot of Cumulative Density Function (CDF) in Fig. 4a, 4b presents the variability of predicted mean monthly streamflow by SVR and RF for training and testing phases. Figure 5(a) illustrates that the RF model successfully captures the monthly

Fig. 3(a) Scatter plot of observed and predicted streamflow using SVR during training phase

Fig. 3(b) Scatter plot of observed and predicted streamflow using SVR during testing phase

Machine Learning Algorithms with Hydro-Meteorological Data …

37

Fig. 4(a) Scatter plot of observed and predicted streamflow using RF during training phase

Fig. 4(b) Scatter plot of observed and predicted streamflow using RF during testing phase

streamflow while SVR underpredicts. Figure 5(b) shows that SVR tends to underpredict, while RF overpredicts for low flow. In terms of peak flow, both models seem to be underpredicted the streamflow with less prediction error under the RF model. Developing separate models between the dry season (low flow period) and wet season (peak flow period) could improve the model in capturing the peak flow patterns [9]. In addition, it is appropriate to add more climate data and consider and combine

38

M. N. M. Adib and S. Harun

relevant input data such as soil, topography, and vegetation, giving a possibility of improving the machine learning models by nearly 30% in RMSE value [36]. Despite that, the performance results of both the training and testing phase are good and acceptable.

Fig. 5(a) Variability of predicted mean monthly streamflow by SVR and RF models during training phase

Fig. 5(b) Variability of predicted mean monthly streamflow by SVR and RF models during testing phase

Machine Learning Algorithms with Hydro-Meteorological Data …

39

4 Conclusion In this study, an investigation of machine learning algorithms between support vector regression (SVR) and random forest (RF) has been carried out to evaluate their performance based on minimum input of hydro-meteorological data for predicting streamflow in Kurau River. These streamflow forecasting models were trained and tested based on 20 (1976-1995) and 10 (1996-2005) years of observed station data, respectively, driven only by rainfall, maximum and minimum temperature, relative humidity, and wind speed data. The study reveals that the RF model outperforms the SVR model in both the training and testing phases. Despite that, all tested models show good agreement between observed and predicted streamflow data in both phases. As a recommendation, including more relevant information that has a specific influence on the high stochasticity and randomness of streamflow as input variables could improve the machine learning model’s prediction ability. Overall, RF and SVR models can confidently be employed for streamflow forecasting for the Kurau River of Kurau River Basin, with a better predictive ability under the RF model. Acknowledgements. This research was supported by the Ministry of Higher Education (MOHE), Universiti Putra Malaysia (UPM), and Universiti Teknologi Malaysia (UTM). The authors are grateful to the Department of Irrigation and Drainage (DID) and the Malaysian Meteorological Department (MMD) for providing gauged hydro-meteorological data for the study.

References 1. Abdulkareem JH, Pradhan B, Sulaiman WNA, Jamil NR (2018) Review of studies on hydrological modelling in Malaysia. Model Earth Syst Environ 4(4):1577–1605. https://doi.org/10. 1007/s40808-018-0509-y 2. Adib MNM, Rowshon MK, Mojid MA, Habibu I (2020) Projected streamflow in the Kurau River Basin of Western Malaysia under future climate scenarios. Sci Rep 10(1):1–15. https:// doi.org/10.1038/s41598-020-65114-w 3. Alam MJ, Meah MA, Noor MS (2011) Numerical modeling of ground water flow and the effect of boundary conditions for the HSIEH aquifer. Asian J. Math. Stat. 4:33–44 4. Arnold JG et al (2012) SWAT: Model use, calibration, and validation. Trans ASABE 55(4):1491–1508. ISSN 2151–0032 5. Breiman L (2001) Random forests. Mach Learn 45:5–32. https://doi.org/10.1201/978042946 9275-8 6. Cai Y, et al (2019) Integrating satellite and climate data to predict wheat yield in Australia using machine learning approaches. Agric For Meteorol 274:144–159. https://doi.org/10.1016/j.agr formet.2019.03.010 7. Cheng M, Fang F, Kinouchi T, Navon IM, Pain CC (2020) Long lead-time daily and monthly streamflow forecasting using machine learning methods. J Hydrol 590:125376. https://doi.org/ 10.1016/j.jhydrol.2020.125376 8. Devia GK, Ganasri BP, Dwarakish GS (2015) A review on hydrological models. Aquatic Procedia 4:1001–1007. https://doi.org/10.1016/j.aqpro.2015.02.126

40

M. N. M. Adib and S. Harun

9. Ghaith M, Siam A, Li Z, El-Dakhakhni W (2020) Hybrid hydrological data-driven approach for daily streamflow forecasting. J Hydrol Eng 25(2):04019063. https://doi.org/10.1061/(asc e)he.1943-5584.0001866 10. Guan H, Li J, Chapman M, Deng F, Ji Z, Yang X (2013) Integration of orthoimagery and lidar data for object-based urban thematic mapping using random forests. Int J Remote Sens 34(14):5166–5186. https://doi.org/10.1080/01431161.2013.788261 11. Hassan Z, Shamsudin S, Harun S, Malek MA, Hamidon N (2015) Suitability of ANN applied as a hydrological model coupled with statistical downscaling model: a case study in the northern area of Peninsular Malaysia. Environ Earth Sci 74(1):463–477. https://doi.org/10.1007/s12 665-015-4054-y 12. Hussain D, Khan AA (2020) Machine learning techniques for monthly river flow forecasting of Hunza River Pakistan. Earth Sci Inform 13(3):939–949. https://doi.org/10.1007/s12145-02000450-z 13. Ishida H, Oishi Y, Morita K, Moriwaki K, Nakajima TY (2018) Development of a support vector machine based cloud detection method for MODIS with the adjustability to various conditions. Remote Sens Environ 205:390–407. https://doi.org/10.1016/j.rse.2017.11.003 14. Ismail H, Kamal MR, Hin LS, Abdullah AF (2020) Performance of HEC-HMS and ArcSWAT models for assessing climate change impacts on streamflow at Bernam River Basin in Malaysia. Pertanika J Sci Technol 28(3):1027–1048 15. Maity R, Bhagwat PP, Bhatnagar A (2010) Potential of support vector regression for prediction of monthly streamflow using endogenous property. Hydrol Proces Int J 24(7):917–923. https:/ /doi.org/10.1002/hyp.7535 16. Moriasi DN, Gitau MW, Pai N, Daggupati P (2015) Hydrologic and water quality models: Performance measures and evaluation criteria. Trans ASABE 58(6):1763–1785. https://doi. org/10.13031/trans.58.10715 17. Mosavi A, Ozturk P, Chau KW (2018) Flood prediction using machine learning models: literature review. Water 10(11):1536. https://doi.org/10.3390/w10111536 18. Pham LT, Luo L, Finley A (2021) Evaluation of random forests for short-term daily streamflow forecasting in rainfall-and snowmelt-driven watersheds. Hydrol Earth Syst Sci 25(6):2997– 3015. https://doi.org/10.5194/hess-2020-305 19. Refshaard JC, Storm B (1995) Computer models of watershed hydrology. In: Singh VP (ed) MIKE SHE. Water Resources Publications, Colorado, USA, pp 809–846 20. Sahoo BB, Jha R, Singh A, Kumar D (2019) Application of support vector regression for modeling low flow time series. KSCE J Civ Eng 23(2):923–934. https://doi.org/10.1007/s12 205-018-0128-1 21. Samadianfard S, Jarhan S, Salwana E, Mosavi A, Shamshirband S, Akib S (2019) Support vector regression integrated with fruit fly optimization algorithm for river flow forecasting in Lake Urmia Basin. Water 11(9):1934. https://doi.org/10.3390/w11091934 22. Shevade SK, Keerthi SS, Bhattacharyya C, Murthy KRK (2000) Improvements to the SMO algorithm for SVM regression. IEEE Trans Neural Networks 11(5):1188–1193. https://doi.org/ 10.1109/72.870050 23. Sheykhmousa M, Mahdianpari M, Ghanbari H, Mohammadimanesh F, Ghamisi P, Homayouni S (2020) Support vector machine versus random forest for remote sensing image classification: a meta-analysis and systematic review. IEEE J Select Top Appl Earth Observ Remote Sens 13:6308–6325. https://doi.org/10.1109/JSTARS.2020.3026724 24. Sidhu RK, Kumar R, Rana PS (2020) Machine learning based crop water demand forecasting using minimum climatological data. Multimedia Tools Appl 79(19):13109–13124. https://doi. org/10.1007/s11042-019-08533-w 25. Smola AJ, Schölkopf B (2004) A tutorial on support vector regression. Stat Comput 14(3):199– 222 26. Suliman AHA, Katimon A, Darus IZM, Shahid S (2016) TOPMODEL for streamflow simulation of a tropical catchment using different resolutions of ASTER DEM: optimization through response surface methodology. Water Resour Manage 30(9):3159–3173. https://doi.org/10. 1007/s11269-016-1338-2

Machine Learning Algorithms with Hydro-Meteorological Data …

41

27. Tan ML, Liang J, Samat N, Chan NW, Haywood JM, Hodges K (2021) Hydrological extremes and responses to climate change in the kelantan river basin, malaysia, based on the CMIP6 highresmip experiments. Water 13(11):1472. https://doi.org/10.3390/w13111472 28. Tongal H, Booij MJ (2018) Simulation and forecasting of streamflows using machine learning models coupled with base flow separation. J Hydrol 564:266–282. https://doi.org/10.1016/j. jhydrol.2018.07.004 29. USACE-HEC (2000) Hydrologic modeling system HEC-HMS: technical reference manual. US Army Corps of Engineerm Hydrologic Engineering Center, California, USA 30. Vapnik V (1995) The Nature of Statistical Learning. Theory. Springer-Velag, New York 31. Wan Zin WZ, Jamaludin S, Deni SM, Jemain AA (2010) Recent changes in extreme rainfall events in Peninsular Malaysia: 1971–2005. Theoret Appl Climatol 99(3):303–314. https://doi. org/10.1007/s00704-009-0141-x 32. Wu CH, Tzeng GH, Lin RH (2009) A Novel hybrid genetic algorithm for kernel function and parameter optimization in support vector regression. Expert Syst Appl 36(3):4725–4735. https://doi.org/10.1016/j.eswa.2008.06.046 33. Wu MC, Lin GF, Lin HY (2014) Improving the forecasts of extreme streamflow by support vector regression with the data extracted by self-organizing map. Hydrol Process 28(2):386– 397. https://doi.org/10.1002/hyp.9584 34. Yaseen ZM, Kisi O, Demir V (2016) Enhancing long-term streamflow forecasting and predicting using periodicity data component: application of artificial intelligence. Water Resour Manage 30(12):4125–4151. https://doi.org/10.1007/s11269-016-1408-5 35. Zewdie GK, Lary DJ, Liu X, Wu D, Levetin E (2019) Estimating the daily pollen concentration in the atmosphere using machine learning and NEXRAD weather radar data. Environ Monit Assess 191(7):1–9. https://doi.org/10.1007/s10661-019-7542-9 36. Zounemat-Kermani M, Batelaan O, Fadaee M, Hinkelmann R (2021) Ensemble machine learning paradigms in hydrology: a review. J Hydrol 598:126266. https://doi.org/10.1016/j. jhydrol.2021.126266

Prediction of Industrial Water Consumption - Blue Water Footprint in Kuantan River Basin E. A. Aziz, S. N. Moni, M. J. Letchumy, N. Yusoff, and S. Z. Zabir

Abstract Industrial and agricultural sectors have been recognised as sectors attributed to the world most water consumption that may lead to the creation of water scarcity in Malaysia. The beginning of water scarcity event may be manifested by repeating water shortages occurrences at many places, owing to poor water management. Water footprint assessment has been accepted as a tool to account the amount of freshwater being consumed for goods and services we used. In this study, appropriation of freshwater of water consumed for industrial sector within Kuantan River Basin is accounted. The accounting of blue water footprint (WFblue) was performed individually on to the Water Treatment Plant (WTP) available in the Kuantan River Basin and the trend was successfully defined; WFblue of Semambu industrial area showed an increased trend from 2015 to 2019 while the Panching industrial area showed a decreasing trend. In attempted to predict the trend by using ANN, both industrial areas showed an increased trend. In addition to that, the significant factor that influenced the accounting of WFblue was able to be defined based on social, economy and environmental. It can be deduced that an increasing light industrial activities contribute to the increasing calculation blue water footprint as well as the prediction trend. Thus, it is important for the industrial activities to well manage the proportion of water resource as industrial activities have been the main contributor to the unsustainability of Kuantan River Basin water resources. Keywords Water footprint · Blue water footprint · Recurrent Neural Network · Water Footprint Prediction

E. A. Aziz · S. N. Moni (B) · N. Yusoff · S. Z. Zabir Department of Civil Engineering, College of Engineering, Universiti Malaysia Pahang, Gambang Campus, 26300 Kuantan, Pahang, Malaysia e-mail: [email protected] M. J. Letchumy Innovative Manufacturing, Mechatronics and Sports Laboratory, Faculty of Manufacturing and Mechatronics Engineering Technology, Universiti Malaysia Pahang, 26600 Pekan, Pahang, Malaysia e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023 I. K. Othman et al. (eds.), Proceedings of the 5th International Conference on Water Resources (ICWR) – Volume 2, Lecture Notes in Civil Engineering 365, https://doi.org/10.1007/978-981-99-3577-2_4

43

44

E. A. Aziz et al.

1 Introduction Water is an important component for industrial activities and human well-being and plays a vital role in improving the efficiency of agriculture and industry [6]. Lots of water bodies are contaminated with all sorts of substances, and stocks of both surface water and groundwater are drained in many places around the world [8]. Recently, the global water scarcity resources and sustainable use of freshwater resources has become a main concern for organisations, governments, water consumers, policy makers, and managers of water resources [19]. Therefore, one of the efficient and sustainable tool to measure total water used is by using water footprint accounting [15]. The water footprint (WF) is the quantity of water used for the manufacture of products and services [3, 4, 11, 13, 15–18, 20] and a measure of the human appropriation of global water resources as determined by the amount of water consumed and or contaminated [13]. Currently, a water shortage problem keeps on occurring in several states of Malaysia due to water pollution mostly by industrial activities especially in Klang Valley [5]. Hence, efficient water management is crucial to ensure the sustainability of water resources. Specifically, in Kuantan, Pahang practising good water management is important as the city is currently having a rapid urbanization. Besides, the population of Kuantan has risen steadily over the past five years [14]. Previous study showed that in 2015 to 2016, the ratio of water footprint to water availability at Semambu WTP was 86.09, 76.39, 83.13, 81.48, 65.06, 73.08 and 73.29%, respectively. Those values were all greater than the Sg. Kuantan water availability (m3 /year) which in 2010 to 2016 was 68.4, 141.7, 95.0, 101.1, 233.7, 156.0, and 154.9 m3 /year, respectively. This eventually implies an unsustainable method of water supply treatment process (WSTP) occurrences at Semambu WTP [2]. Meanwhile, in Panching WSTP, although it has a capacity of 160 million litters per day in Panching WTP, it has to covers not only the residential areas but also to cope with the demand from 5,600 hectares of industrial area [1]. Thus, it is believed that this study would be able to determine the proportion of industrial water consumption by based on WFBLUE accounting and predict the trend in five years’ time. Hence, the factors that influence the blue water footprint (WFBLUE) accounting can be identified.

2 Study Area The study area is located at Kuantan River Basin with the area of the catchment is about 1,630 km2 . The geographical coordinate for the Kuantan River Basin is at 3 º54’28”N and 103 º07’54”E. Within the area there are 5 WTPs but only 2 WTPs supply water to the industrial areas of Semambu and Panching as shown in Figs. 1 and 2. The intake water for all WTPs within Kuantan River Basin is solely from Sungai Kuantan.

Prediction of Industrial Water Consumption - Blue Water Footprint …

45

Fig. 1 Semambu WTP location

Fig. 2 Panching WTP location

2.1 Data Inventory Data inventory is a process to list sets of data that define the source’s licensing, content, and other valuable information sets. In this research, the data that has thoroughly recorded from the Kuantan City Council was inventoried. This inventorying method consists of the monthly water consumption details. It was being inventoried based on industry types from different areas.

46

E. A. Aziz et al.

2.2 Blue Water Footprint Accounting The blue water footprint refers to the consumption of water, which requires the evaporation of surface water and groundwater blue water resources [9]. Below is the equation to calculate the Total WFblue, ind:   T ot al W F bl e,wt p = W at er I nt ake + [ET 0 × Ar ea] + Rai n f all × Ar ea (1)   vol ume T ot al W F bl e,wt p = I ndust r i al W at er C onsum pt i on (2) t i me where: Total WFblue,WTP = Total blue water footprint for WTP. ET0 = Evaporation of each tank. Area = Area if each tank. Total WFblue,ind = Total blue water footprint for industrial.

2.3 Blue Water Footprint Prediction Since the data obtained was from the previous years, the backpropagation method of ANN is being used in order to get the prediction the data’s pattern. The Artificial Neural Network is used to predict the blue water footprint. ANN is an information manager model that is comparable to the work of the human brain’s biological nervous systems [10]. Using the test data collection, the performance of the ANN models obtained after training was evaluated. The precision of this ANN prediction is found to be 92.4 to 99.96% [12]. Due to the introduction of the ANN model, pre-processing was performed by removing the outliers while addressing the missing data through mean imputation (min–max method). Later, to provide an accurate treatment of the functionality used due to the wide collection of data, the data was normalised from 0 to 1. In this study, the loss function that were implemented is mean square error along with Adam as optimizer. The model consists of 3 layers with 50-unit size on each layer. The model was fit with batch normalization size of 10 and epoch value were set 50.

2.4 Factor Influencing the Prediction of Blue Water Footprint Influence factor to the total WF prediction can be classified into three main factors such as environmental, social, and economic. In Environmental aspect; monsoonal activities is the factor that can affect the water footprint accounting. For example, the changes in environmental conditions can be significant to the influence for water

Prediction of Industrial Water Consumption - Blue Water Footprint … Table 1 Number of industries at Semambu WTP

Year

47

Type of industry Light

Medium

Heavy

2015

673

185

46

2016

648

194

52

2017

626

187

54

2018

612

199

52

2019

615

194

56

supply and demand. It is possible that increased precipitation would affect the supply of water. Meanwhile, in social aspect; the use of water involves the availability of water and the demand for water for industrial use. The amount of water used for these activities is also related to the number and spatial distribution of people in the area to some degree. The economic impact’s calculation for water footprint is difficult because not only water to include in the calculation but also the economic analysis that requires a broad analysis including all production factors. There will be no more than a partial analysis of an economic analysis which focuses on water as an input factor, meaning that no general economic conclusions can be drawn from it [7]. In order for the people to use the water they need to pay before the water supplied to their residential.

3 Results and Discussion 3.1 Inventory of Industries According to Water Treatment Plant Supplied The implementation of the inventory method was used to identify the types of industry. Tables 1 and 2 show the number of industries at both Semambu and Panching WTP. The types of the industry will be inventoried into categories as shown in Table 3. The industries available within Kuantan River Basin is divided in accordance with the Water Treatment Plants (WTP) that supply water to them.

3.2 Blue Water Footprint Determination for Industrial Sector Within Kuantan River Basin The production of water to be supplied from both WTPs; Semambu WTP and Panching WTP was calculated by previous researcher [2]. In this study, the portion of the supplied water taken by each industrial area from its WTP was successfully

48 Table 2 Number of industries at Panching WTP

E. A. Aziz et al.

Type of industry

Year 2015

Table 3 Total industries supplied by each WTP

Light

Medium

51

11 12

2016

52

2017

55

8

2018

55

8

2019

40

12

Year

Semambu WTP

Panching WTP

Total no. of Industry

2015

904

62

966

2016

894

64

958

2017

867

63

930

2018

863

63

926

2019

865

52

917

obtained. It shows that, the industrial areas use about half of the supplied water and the remaining percentage goes to other sectors such as domestic and agriculture. The industry makes about 52% of total water consumption and the other 48% is for agricultural and domestic uses such as commercial, public use, and paddy planting, palm plantation, rubber plantation, and etc. Based on Fig. 3, the value of the blue water footprint for Semambu industrial area from January to July has fluctuated. The WFblue reached a peak in August with the value of 3,756,643.00 m3 /month before significantly fall to 1,908,619.00 m3 /month in September 2015. The value of WFblue increased because of the highest number of heavy industries in the area. It is known that heavy industries use more water than other sectors and it is in their nature to require more water to clean up waste. Thus, it would have an impact on the monthly decrement in the amount of WFblue. The value increased slightly in October with the value of 2,039,553.00 m3 /month before decreased towards December. In 2016 from January to April the value of blue water footprint for the industrial area was constantly decreased with the value of 1,986,089.00, 1,805,257.00, 1,697,708.00, and 1,607,752.00 m3 /month, respectively. Then, the value of WFblue slightly increased in May with the value of 1,807,866.00 m3 /month; decreased to 1,625,108.00 m3 /month in June and increased again in July with the value of 2,207,671.00 m3 /month. Later, the value increased to 4,233,728.00 m3 /month in August due to the highest amount of water consumed by heavy industry in the area with the value of 3,658,195.00 m3 /month, before decreased slightly to 3,965,308.00 m3 /month in September and dropped to 2,448,372.00 m3 /month in October. After that, the value reached the lowest point with the value of 2,151,536.00 m3 /month in December.

Prediction of Industrial Water Consumption - Blue Water Footprint …

Total WF 5,000,000.00 4,500,000.00 4,000,000.00 3,500,000.00 3,000,000.00 2,500,000.00 2,000,000.00 1,500,000.00 1,000,000.00 500,000.00 -

49

at Semambu Industrial Area (m³/day) blue

2015 2016 2017 2018 2019

Fig. 3 Total WFblue at Semambu for industrial area from 2015–2019

Figure 3 shows, in 2017 and 2018 the WFblue value was gradually increased and decreased. In November 2017, the value of WFblue reached the highest value which is 3,386,903.00 m3 /month and decreased to 1,789,593.00 m3 /month in December 2017. At the end of the month of 2018, the value of WFblue increased to 2,098,773.00 m3 / month although it is not the highest value. The trend shows decreasing in the number of water consumption by 5.7% from 2018 to 2019. The percentage of water consumes was decreased due to the lessened in the number of industrial at the Kuantan River Basin area. In the year 2019, from January to May the value of blue water footprint decreased; it slightly increased in March with the value of 2,795,427.00 m3 /month. Then, the value of WFblue increased in June with the value of 2,965,600.00 m3 / month and decreased to 2,523,095.00 m3 /month in July before remaining constant from August to September. In October, the value of WFblue increased significantly and reached a peak value which is about 4,416,080.00 m3 /month. Later the value dramatically decreased to 2,219,827.00 m3 /month in November and hit the lowest value of 1,984,017.00 m3 /month. From Fig. 4, from January to February 2015 the values dropped from 21,660.00 to 11,131.00 m3 /month. However, from March to April, the WFblue for Panching industrial area rose steadily. Then, in May the amount of water consumed decreased to 14,724.00 m3 /month until June with the value of 12,828.00 m3 /month; the value increased dramatically reaching the highest amount for the year in July with 33,707.00 m3 /month. In August, the value slumped to 12,853.00 and 10,300.00 m3 / month in September. Later, the value increased in October and fell back in November to 13,346.00 m3 /month and remained constant until the end of the year. The total WFblue for the Panching industrial area which also shows the amount of water consumed increased about 22% from 2015 to 2016 shown in Fig. 4. From January 2016, the amount of WFblue dropped significantly from 68,936.00 to 37,064.00 m3 /month, respectively. However, the value of water consumption rose to 48,665.00 m3 /month in March. From April to June, the graph’s pattern fluctuated

50

E. A. Aziz et al.

TotalWF

at Panching Industrial Area (m³/day)

blue 80,000.00 70,000.00 60,000.00 50,000.00

2015

40,000.00

2016

30,000.00

2017

20,000.00

2018 2019

10,000.00 -

Fig. 4 Total WFblue at Panching for industrial area from 2015–2019

and from July to December the amount of water consumed for the industrial area remained stable. In 2017, the water consumption of Panching industrial area decreased by 24%. The value of WFBLUE climbed gradually from January to May and declined slightly in March with 11,762.00 m3 /month. In June, the value calculated was 12,836.00 m3 / month. In July, the value increased to 25,880.00 m3 /month and decreased to 16,887.00 m3 /month in August. The pattern continued to decrease until October with the values of 12,615.00 m3 /month. In November, the value increased to 14,169.00 m3 /month and in December the value was registered at about 23,495.00 m3 /month. It can be seen that the trend decreased for the amount of total WFblue of industries for all year. However, in 2017 the trend increased by 13% at the end of the year and this may be due to the increasing number of the light industry in the area. Based on Fig. 4, the value of WFblue for the Panching industrial area was constant and in April 2018 the value reached its highest with 15,413.00 m3 /month. The value of WFblue shows the lowest throughout 2019. This might be due to the decreasing number of industries in the area.

3.3 Prediction of Blue Water Footprint for Semambu and Panching Industrial Area in Kuantan River Basin Prediction of WFblue for both Semambu and Panching was done by using ANN. Since the data was the industrial area from the previous years, the backpropagation method is used to obtain a prediction for the next five years of the data’s pattern.

Prediction of Industrial Water Consumption - Blue Water Footprint …

51

Figure 5 visualised the trend of values of WFblue at Semambu industrial area after the training by using ANN. It can be seen that the highest values of WFblue at the end of 2019. There are some significant roses in the values of WFblue around the end of each year. These might be due to the climate changes and monsoonal seasons. There is a long-term fluctuation in the values of WFblue at Semambu industrial area which decreased during 2016, increased in 2017, and decreased again in 2018. The different between the actual WFblue values and the predicted values at Semambu industrial area after the training by using ANN as shown in the Fig. 6. It can be seen that the predicted values follow the actual WFblue values very closely. Therefore, it is possible to apply this model to test the data for the next five years. Figure 7 shows the increasing trend of WFblue at Semambu industrial area after training the data by using ANN. After the training process, the most optimised hyperparameters have been chosen to run the data in order to show the prediction trend whether it is increasing or decreasing. Therefore, increasing trend was predicted during 2020 to 2024 as shown in the Figure. The increasing trend of predictions might be due to the increasing number of industries that contribute to industrial water consumption. Figure 8 shows the trend of values of WFblue E at Panching industrial area after the training by using ANN. A large increased in the values of WFblue at Panching industrial area was in the beginning of 2016. These might be due to the increased number of active industries in that year. Apparently, a constant pattern for values of WFblue from the middle of 2016 until the end of 2018. There is significant dropped in the values of WFblue in 2019. The decreased number of industries can be the effect to the pattern. After the training period by using ANN model, the different between actual WFblue values and the predicted values at Panching industrial area is visualised

Fig. 5 The trend of values of WFblue at Semambu industrial area

52

E. A. Aziz et al.

Fig. 6 Actual WFblue values and predicted values for Semambu industrial Activities

Fig. 7 The prediction of WFblue at Semambu industrial area by using ANN

in the Fig. 9. The results show there is decreased in the actual values of WFblue in 2019. This might me due to the decreased number of industries at Panching industrial area in that year. The training was able to produce an output from the data, however, it could not state that it is possible to apply this model to test the data for the next five years. After undergoing the series of training by using ANN. The most optimised hyperparameters have been chosen to run the data in order to show the prediction trend whether it is increasing or decreasing. There is a slope predicted in 2024 that may be

Prediction of Industrial Water Consumption - Blue Water Footprint …

53

Fig. 8 The trend of values of WFblue at Panching industrial area

Fig. 9 The actual WFblue values and the predicted values at Panching industrial area

affected by the total WFblue of the industrial area. The predicted value at Panching showed an average pattern of change in Fig. 10. Hence, it could not be declaring the number was increased nor decreased. The model predicted peak of increasing value by the end of 2024 can be neglected due to uncertainties of WFblue value. Therefore, it can be concluded that, the trend will be slightly increasing but within the normal range. This is because, Panching WTP area was not gazetted as industrial zone and mostly light and medium industry are operated within that area.

54

E. A. Aziz et al.

Fig. 10 The prediction of WFblue at Panching industrial area by using ANN

3.4 Influence Factor to Blue Water Footprint Determination Social Table 4 shows the water demand of industry in the Kuantan River Basin area from 20,152,019. Industrial activities contribute 52% of blue water footprint calculation. For the social factor, the number of industries that exist is influenced by water demand. The increased population industry (per person) will increase the water demand for the area. As we can see, the estimate of water demand was increased about 1% throughout the year. Although the number of premises decreases, the number of population industry is increasing. The increasing number of populations was due to increasing number of heavy industries that uses more manpower. Light industry are industries that usually are less capital-intensive than heavy industry and are more consumer-oriented than businessoriented, as they typically produce smaller consumer goods. Table 4 Water demand for the industry from 2015–2019 Year

Num. of premises

Population industry (per person)

C, (m3 /head/day)

F

Dn, (m3 )

Wdn, m3

Total WFBLUE (Industry)

2015

966

240,240

0.218

0.95

45

49,798.70

24,078,031

2016

958

244,276

0.219

0.95

45

50,866.59

28,154,784

2017

930

247,200

0.220

0.95

45

51,709.80

25,688,189

2018

926

250,704

0.221

0.96

45

53,234.36

24,837,189

2019

917

254,208

0.222

0.96

45

54,221.81

32,396,745

Prediction of Industrial Water Consumption - Blue Water Footprint …

55

According to Fig. 11, the highest amount of total rainfall in 2015 was in August which is 2611.0 mm, while the lowest was in March which is 405.0 mm. In 2016, the graph trend indicates an increase in total rainfall from August to December. As we can see, the water treatment plant recorded the most rain in January 2017 and 2018, with 5819.5 and 6235.5 mm, respectively. Following that, the total amount of rainfall was received inconsistently in the year 2019. Then, March and April receive the least amount of rain nearly every year. The trend of the graph in Fig. 11 above indicates that it steadily rises and declines, with increases from 2015 to 2017 and decreases from 2018 to 2019. The water treatment plant recorded the most total rainfall for a year in 2017, with 27,459.5 mm. However, it recorded the least amount of total rainfall in 2015, with 16,193.5 mm. We may assume that climate and monsoonal events have an effect on overall rainfall per year. The Fig. 12 shows the number of industries at Semambu industrial area is decreasing throughout the year. The highest number of industries is in 2015 which was 904 with 24,078,031 m3 /year of total blue water footprint. The lowest number of industries is 863 that recorded in 2018 with 24,837,189 m3 /year of total blue water footprint. The highest number of industries that consumed water at Semambu industrial area was light industries, with a total of 3,124 over a five-year period. The year with the most active companies for light industry is 2015 which is 671. While the heavy industry has the lowest number of industries over the years. From 2015 to 2019, the number of heavy industries increased while remaining constant from 2016 to 2017. We can assume that the number of industries was decreased due to increasing number of inactive or closed companies for light industries. The increasing number

Total Rainfall 7000.0 6000.0 5000.0 4000.0 3000.0 2000.0 1000.0 0.0 2015 Jan

Feb

2016 Mar

Apr

May

2017 Jun

Fig. 11 Total rainfall per month for 2015–2019

Jul

2018 Aug

Sep

Oct

2019 Nov

Dec

56

E. A. Aziz et al.

Number of Industries 800 700 600 500 400 300 200 100 0 2015

2016

2017 Light

Medium

2018

2019

Heavy

Fig. 12 Number of industries at Semambu WTP from 2015-2019

of heavy industries might be a factor that slightly influence the increasing of total blue WF in 2019. There are only two industries that consumed water from Panching WTP which are light and medium. Figure 13 shows that light industries consumed the most water from Panching WTP, with a total of 253 over a five-year period. Medium industries have constant value from 2018 to 2019 with the value of 8. In 2019, the total WFblue recorded the highest even though the total number of industries is only 52. Thus, we can assume that the increasing number of total

Number of Industries 60 50 40 30 20 10 0 2015

2016

2017 Light

Medium

Fig. 13 Number of industries at Panching WTP from 2015–2019

2018

2019

Prediction of Industrial Water Consumption - Blue Water Footprint …

57

WFblue is because of the increasing number of medium industries that higher than the light industry which can be a factor that affects the increasing number of total WFblue shown in 2019. Furthermore, the medium industry consumed more water consumption for their production.

4 Conclusion The type of industries at both Semambu and Panching industrial areas were inventoried into three different categories which are light, medium, and heavy. Light industries recorded the highest number of industries at both Semambu and Panching WTPs. The total water consumption of the industrial area and the portion of the supplied water taken by each industrial area from its WTP has been determined. It can be concluded that industries at Semambu consumed more water as compared to Panching industrial area. This might be due to the number of industries at Semambu industrial area more than at Panching industrial area. The predicted trend of blue water footprint for the next five years after the actual data subjected to a series of training using ANN modelling was able to produce successfully in this study. The predicted trend of WFblue at Semambu industrial area showed an increasing of water consumed and the trend also showed the amount of water consumed for the next five years is predicted to be increased. Meanwhile, in Panching industrial area, there is a slope predicted in 2024 that may be affected by the total WFblue of the industry. Hence, it can be said that the trend of WFblue was predicted to be slightly increased with in the normal range. Finally, it can be deduced that an increasing light industrial activities contribute to the increasing calculation of blue water footprint as well as the prediction trend. Thus, it is important for the industrial activities to well manage the proportion of water resource as industrial activities have been the main contributor to the unsustainability of Kuantan River Basin water resources. Acknowledgements The authors would like to acknowledge the support by the Ministry of Education, Malaysia, from the Fundamental Research Grants Scheme (FRGS) FRGS/1/2019/TK01/UMP/ 02/3, and Environmental Research and Sustainability Centre (ERAS), UMP due to the support in accomplished this study.

References 1. Aziz EA, Malek MA, Moni SN, Zulkifli NF, Hadi IH (2018) Water supply treatment sustainability of panching water supply treatment process-water footprint approach. In: IOP conference series: materials science and engineering, vol. 318, no 1, p 012028. IOP Publishing. https:// doi.org/10.1088/1757-899X/318/1/012028

58

E. A. Aziz et al.

2. Aziz EA, Malek MA, Moni SN, Hadi IH, Zulkifli NF (2018b) Water supply treatment sustainability of semambu water supply treatment process-water footprint approach. In: IOP conference series: materials science and engineering, vol 318, no 1, p 012027. IOP Publishing. https:/ /doi.org/10.1088/1757-899X/318/1/012027 3. Bazrafshan O, Zamani H, Etedali HR, Dehghanpir S (2019) Assessment of citrus water footprint components and impact of climatic and non-climatic factors on them. Sci Hortic 250:344–351. https://doi.org/10.1016/j.scienta.2019.02.069 4. Chooyok P, Pumijumnog N, Ussawarujikulchai A (2013) The water footprint assessment of ethanol production from molasses in Kanchanaburi and Supanburi province of Thailand. APCBEE Proc 5:283–287 5. Chung C (2020) ‘Air Selangor’: Over 850 areas in Klang Valley affected by water supply disruption (updated). The Star pp 1–7. https://WFblue.thestar.com.my/news/nation/2020/12/ 08/air-selangor-over-800-areasin-klang-valley-affected-by-unscheduled-water-cuts 6. Fan JL, Wang JD, Zhang X, Kong LS, Song QY (2019) Exploring the changes and driving forces of water footprints in China from 2002 to 2012: a perspective of final demand. Sci Total Environ 650:1101–1111. https://doi.org/10.1016/j.scitotenv.2018.08.426 7. Hoekstra AY, Chapagain AK, Aldaya MM, Mekonnen MM (2009) Water footprint manual: State of the art 2009. Water footprint network, enschede, The Netherlands p 255 8. Hogeboom RJ (2020) The water footprint concept and water’s grand environmental challenges. One Earth 2(3):218–222. https://doi.org/10.1016/j.oneear.2020.02.010 9. Hogeboom RJ, Knook L, Hoekstra AY (2018) The blue water footprint of the world’s artificial reservoirs for hydroelectricity, irrigation, residential and industrial water supply, flood protection, fishing and recreation. Adv Water Resour 113:285–294. https://doi.org/10.1016/j.advwat res.2018.01.028 10. Abiodun OI, Jantan A, Omolara AE, Dada KV, Mohamed NA, Arshad H (2018) State-of-theart in artificial neural network applications: a survey. Heliyon 4(11):e00938. https://doi.org/10. 1016/j.heliyon.2018.e00938 11. del Milagro Jorrat M, Araujo PZ, Mele FD (2018) Sugarcane water footprint in the province of Tucumán, Argentina. Comparison between different management practices. J Clean Prod 188:521–529. https://doi.org/10.1016/j.jclepro.2018.03.242 12. Khuan LY, Hamzah N, Jailani R (2002) Prediction of water quality index (WQI) based on artificial neural network (ANN). In: student conference on research and development, pp. 157– 161. IEEE 13. Liu J, Zhao D, Mao G, Cui W, Chen H, Yang H (2020) Environmental sustainability of water footprint in mainland China. Geography Sustain 1(1):8–17. https://doi.org/10.1016/j.geosus. 2020.02.002 14. Macrotrends (2021) Kuantan, Malaysia Metro Area Population 1950–2020, Macrotrends LLC, 2020–2022. https://www.macrotrends.net/cities/21810/kuantan/population 15. Muratoglu A (2020) Grey water footprint of agricultural production: an assessment based on nitrogen surplus and high-resolution leaching runoff fractions in Turkey. Sci Total Environ 742:140553. https://doi.org/10.1016/j.scitotenv.2020.140553 16. Nezamoleslami R, Hosseinian SM (2020) Data needed for assessing water footprint of steel production. Data Brief 30:105461. https://doi.org/10.1016/j.dib.2020.105461 17. Subramaniam V, Hashim Z (2018) Charting the water footprint for Malaysian crude palm oil. J Clean Prod 178:675–687. https://doi.org/10.1016/j.jclepro.2018.01.061 18. Wu B, Zeng W, Chen H, Zhao Y (2016) Grey water footprint combined with ecological network analysis for assessing regional water quality metabolism. J Clean Prod 112:3138–3151. https:/ /doi.org/10.1016/j.jclepro.2015.11.009 19. Yu J, Long A, Deng X, He X, Zhang P, Wang J, Hai Y (2020) Incorporating the red jujube water footprint and economic water productivity into sustainable integrated management policy. J Environ Manage 269:110828. https://doi.org/10.1016/j.jenvman.2020.110828 20. Zheng J, Wang W, Liu G, Ding Y, Cao X, Chen D, Engel BA (2020) Towards quantification of the national water footprint in rice production of China: a first assessment from the perspectives of single-double rice. Sci Total Environ 739:140032. https://doi.org/10.1016/j.scitotenv.2020. 140032

Validation of Gridded Data Set Over Semi-arid Region of Syria Rajab Homsi, Shamsuddin Shahid, Tarmizi Ismail, Jam Shahzaib Khan, Zafar Iqbal, and Atif Muhammad Ali

Abstract Arid and semi-arid regions are particularly susceptible to climate change. However, precipitation and temperature reliable data for the long term are unavailable in many developing regions mainly in the conflict-affected regions of the world. Gridded Climate Data (GCD), in recent years, has emerged as a reliable data sets especially for the remote regions. The reliability and accuracy of these data sets vary from region to region. In this study, the GCD of precipitation global precipitation climatology center (GPCC) and temperature climate research unit (CRU) is validated using various statistical and spatial analysis over Syria. The quality of the observed data has been checked by the double mass curve and compared with the observed by time-series and residual analysis. CDF and PDF were plotted for the evaluation of GCD along with the comparison of the spatial distribution of precipitation and temperature. The double mass curve results showed that the annual rainfall time series showed no breakpoint in the graph representing the consistency of the observed data of various stations. The PDF and CDF of the observed and GPCC data were plotted to find the relationship of the GPCC with the observed data. The R2, PBIAS, NRMSE, NSE and md of GPCC with observed data shows that at the various station the gridded data is in close relationship with the gridded data, which strengthens the hypothesis that the GPCC can perform well over Syria. Time series analysis of R. Homsi · S. Shahid · T. Ismail · J. S. Khan · Z. Iqbal (B) School of Civil Engineering, Faculty of Engineering, Universiti Teknologi Malaysia (UTM), 81310 Johor Bahru, Malaysia e-mail: [email protected] S. Shahid e-mail: [email protected] T. Ismail e-mail: [email protected] J. S. Khan Department of Civil Engineering Quaid e Awam, University of Science and Technology, Sakrand Road, Nawabshah, Shaheed Benazirabad 67450, Sindh, Pakistan A. M. Ali College of Hydrology and Water Resources, Hohai University, Nanjing 210098, China © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023 I. K. Othman et al. (eds.), Proceedings of the 5th International Conference on Water Resources (ICWR) – Volume 2, Lecture Notes in Civil Engineering 365, https://doi.org/10.1007/978-981-99-3577-2_5

59

60

R. Homsi et al.

50 years (1951–2010) was compared along with the residual of each station. The comparison showed that the GPCC data matched well with the observed data with a slight under or overestimation at some stations. The spatial distribution of the annual mean rainfall over Syria was plotted for the observed as well as GPCC data. The spatial distribution of the GPCC was much more similar in most part of the country which represents the ability of the gridded data to replicate the annual mean precipitation over most part of the country. Keywords Gridded data validation · Climate change · GPCC · CRU · GCD

1 Introduction Precipitation and temperature are the major variable in the global water cycle, having foremost influence on the climate of a region [56]. Variations in climate have attracted the attention of research community in past decades [37, 68]. For studying regional and global climate changes precipitation and temperature are very important variables [49]. However, precipitation and temperature reliable data for long term are unavailable in many developing regions [59], mainly in the conflict-affected regions of the world. The problem of data scarcity is more severe in semi-arid and arid regions located in developing countries [48, 58]. Arid and semi-arid regions, on the other hand, are particularly susceptible to climate change [26]. About 20% of the world’s population lives in arid areas, which are subjected to severe effects of climate change in terms of hydrological and meteorological extremes [7]. Unavailability of good quality climatic data over a longer period remained a major challenge for the atmospheric and hydrological researchers [6, 38]. The irregular distributed observed stations and data scarcity over the undeveloped region makes the observed station data unsuitable for the hydrological and climate related studies [3]. Whereas various questions are raised for regarding the data quality of the observed satiation data. With the development of technologies, various climate data sets are available for the hydro-climate scientist such as reanalysis data, satellite based remote sensing products and gauge-based observations. Gauge-based gridded data, due to their availability over a long temporal and spatial scales, are used commonly by climate experts [29, 43]. In areas where long-term accurate observation data or gauge data are scarce, a number of multi-source climate data products have been produced for hydrological and climatic studies [28, 50]. These products are primarily divided into gage-based measurements, satellite products, re-analytical results and combination of various datasets [44]. Owing to the spatial and temporal continuity these gridded climate datasets (GCD) are often used by hydrologist and researchers working on climate [68]. Several results have shown that GCD can be used for climate analyses, including the characterisation of drought [2, 60], trend analysis [9, 66] climate downscaling [4], as well as hydrological studies like, flood studies [1, 70], etc. GCD can be regional as well as global. The region specified GCD are, climate dataset was developed

Validation of Gridded Data Set Over Semi-arid Region of Syria

61

by; [53] for Switzerland, [10] for Portugal, [22] for Europe, [65] for Asia, [36, 40] for the north and the south America and [24] for Spain. On other hand various organizations have released the global climate datasets, such as, the WorldClim dataset [25], Global Precipitation Climatology Centre (GPCC) [54]), national oceanic and atmospheric administration (NOAA) [12], climatic research unit (CRU) [21], global historical climatology network (GHCN) [61], the University of Delaware research centre (UDel) [39] and Climate Prediction Center (CPC). GCD is usually assessed by comparing it to reliable observed data. Several experiments have been undertaken to evaluate the validity of various global climate datasets in various parts of the world. [14, 17, 18, 29, 41, 46]. Rainfall in arid areas is sparse and sporadic, resulting in large fluctuations over a short distance. Furthermore, due to less human settlements in arid regions, the rainfall stations are often sparsely distributed. The optimization techniques, the algorithms and the quality of the observed data defines the quality of the GCD [15, 69]. The differences in models and observations can cause counterfeit variability and trends into GCD [15]. Furthermore, owing to the huge temporal and spatial variability of rainfall, the GCD often contain large uncertainties, which increases due to unavailability of observed station data over vast arid region [19, 69]. Due to this and various other factors, the reliability of GCD varies with time and geography [5, 19, 34]. This accentuates the prerequisite for an valuation of the applicability of gridded climate product to reconstruct the climate at a certain location. In this study the gridded data of precipitation and temperature is validated using various statistical and spatial analysis over Syria. The Quality of the observed data has been checked by double mass curve. The Gridded data set was compared with the observed by timeseries and residual analysis. CDF and PDF were plotted for the evaluation of GCD and finally the spatial distribution of the gridded and observed precipitation and temperature was analysed.

2 Study Area and Data 2.1 Study Area Syria is situated in the Middle East and covers an area of 185,180 km2 . Its latitude ranges from 32° to 38°N and its longitude ranges from 35° to 43°E. The country is bordered on the west by the Mediterranean Sea and Lebanon, on the north by Turkey, on the south by Jordan, and on the east by Iraq. The country’s topography is defined in the west by a narrow coastal plain, mountains in the west, and a desert plateau in the east (Fig. 1). The country has two distinct seasons: a rainy and cool winter (November to April) and a warm summer (May to October). The average daily maximum temperature varies from about 40 °C in summer to 12 °C in winter. The average daily minimum

62

R. Homsi et al.

Fig. 1 Location in the world and topography of Syria

temperature goes down to 2 ºC in winter. The summer minimum temperature is around 20 °C.The seasonal variation of temperature in Syria is given in Fig. 2. The Mediterranean winds carries moist air. Syria receives much of the precipitation throughout the winter (November to May). The precipitation befalls in the form of ice and snow in the north. Summer months see virtually no rainfall in most parts of the country. The country’s rainfall varies between 75 and 1000 millimeters each year. Syria’s seasonal variation in precipitation is presented in Fig. 3.

Fig. 2 The seasonal variation of temperature of Syria estimated using climate research unit (CRU) temperature for the period 1951–2010.

Validation of Gridded Data Set Over Semi-arid Region of Syria

63

Fig. 3 The seasonal variation of precipitation of Syria estimated using global precipitation climatology centre (GPCC) precipitation for the period 1951–2010

2.2 Gridded Climate Data In this study, globally used GCD, GPCC V.7 (Global Precipitation Climatology Centre) [8] was used while it was already evaluated as suitable data set in previous researches including our own paper [27]. The details of GCD used in this study is given in Table 1. The distribution of gauges per grid which was used for the development of GPCC data over Syria is represented in Fig. 4. According to the last available data product and the time of inception of GPCC, the number of gauges was found to vary over time. Figure 4 shows the rain gauge density during the year 1990 over Syrian region which was used for preparation of GCD. The gauges were found sparsely distributed, which indicates the presence of uncertainty in GCD and the need of validation. Table 1 Details of GPCC product Data

Detail and source

GPCC

Global Precipitation 0.5° × 0.5° Climatology Center, GPCC v.7 (http://www.esrl. noaa.gov/psd/data/gri dded/data.gpcc.html)

Spatial resolution

Temporal resolution

Geographical coverage

Monthly (1901–2010)

Global, Land only

64

R. Homsi et al.

Fig. 4 Distribution of rainfall gauges that were used for the development of GPCC gridded rainfall for the year 1990

3 Methodology 3.1 Quality Assessment Climate Data The observed data quality was first evaluated using homogeneity tests. The quality ensured observed data were then used for the valuation of the suitability of gridded data. The methods used for this purpose are presented in the following sections. Data quality monitoring is needed prior to the use of climate data for hydro-climatic research because inaccurate outliers can have a significant effect on the results [67]. Furthermore, missing data is another major problem for climatic studies in Syria. In the present study, data of the missing years were discarded. The missing values were filled with the Expectation Maximize approach available in SPSS software. Certain quality control tests were performed to evaluate the data quality such as; rainfall magnitudes checks where the classification was done as rainfall magnitude below 0 mm, rainfall days with higher rainfall magnitude of 20 mm and also winter having more than 60 consecutive dry days.

Validation of Gridded Data Set Over Semi-arid Region of Syria

65

3.2 Double Mass Curve This method is used in data quality analysis to check the data consistency specially for precipitation. It was introduced by [55]. The main procedure in this method it to compare a single station data with all other station in that area. The double mass curve principal is that, it assumes that data of the graph of cumulative plot of one variable is straight against the other variable only if the data is relational. Whereas the proportionality constant between the variables is shown by the slope of the line. Pa = Po

Ma Mo

(1)

o a and Ma = P where Mo = P P P Pa is the adjusted precipitation which is equal to the product of change factor calculated using slope of adjusted Ma and slope of observed data Mo with the observed precipitation Po

3.3 Performance Assessment of Gridded Data Typically, the GCD validated against observed data via interpolation or correlation with nearby stations [12]. The inverse distance weighting method to interpolate gridded precipitation data at the observed point was used in this research. Interpolation was performed using precipitation data from four grid points surrounding the observed point. We compared the interpolated GCD to the observed data.

3.4 Probability Density Function In order to compute the correlation between the probability density functions of observed and the GCD, the skill score proposed by [45] was used. The probability that a variable (i.e. precipitation) falls within a particular set is determined by the integral density of the set and the total area of every PDF which is equal to one [13]. The annual average distribution was studied using 40-year (1951–2010) data. The annual total was estimated for each year for this purpose. The PDFs of the observed and gridded datasets were calculated and plotted together for comparison using Weibull’s probability plotting tool. f (x) = γ x(γ − 1)exp(−(x γ ))x > 0; γ > 0 where γ is the shape parameter.

(2)

66

R. Homsi et al.

3.5 Cumulative Distribution Function (CDF) Cumulative Distribution function (CDF) of observed and gridded data was prepared to show the likelihood of observed and gridded data. The CDF shows the probability that the variable (rainfall) takes the vale less than or equal to the x. x in this case is the mean vale of the rainfall at the particular rainfall station. The cdf function f(x) in term of probability can be defined by the equation below γ

f (x) = 1 − e−(x ) x > 0; γ > 0

(3)

where γ is the shape parameter

3.6 Time Series Analysis The performances of different gridded datasets were also assessed by visual inspection of data. The monthly time series of observed and gridded data were compared to show the efficacy of the data products in replicating the observed data. Furthermore, the residual analysis was carried out by plotting errors against time span.

3.7 Performance Evaluation Indices The following terms are used to define statistical indices.: x_(obs,i) and x_(sim,i) are the i-th observed and gridded data, and n is the number of the observations. Statistical metrics used in this study are explained below:

3.7.1

Percentage of Bias (Pbias)

The Pbias metric assesses the proclivity of model data to over- or under-estimate observed data.Pbias values closer to zero mean better model efficiency. Positive Pbias values mean that the model underestimated bias; negative values indicate that the model overestimated bias. The following equations are used to calculate Pbias  n  Pbias = 100 ×

xsim,i − xobs,i n i=1 x sim,i

i=1

 (4)

Validation of Gridded Data Set Over Semi-arid Region of Syria

3.7.2

67

Normalized Root Mean Square Error (NRMSE)

The NRMSE summarises the magnitudes of prediction errors over time and is thus regarded as a good indicator of accuracy [64]. If the NRMSE of a model is close to zero, it is considered more accurate [12, 31]. It can have a value between −∞ and 1, with 1 being the best. It is defined as follows:    n 1 N RMSE =

3.7.3

n

i=1 1 n

xsim,i − xobs,i  n  i=1 x sim,i

2 1/ 2 (5)

Nash–Sutcliffe Efficiency (NSE)

The NSE is a normalized statistic that is used to evaluate predictive ability of hydrological models [42]. The value of NSE ranges from −∞ and 1, while 1 being the optimal value. Following equation is used to calculate NSE in this study: n N SE = 1 −

3.7.4



i=1 n i=1

2 xsim,i − xobs,i 2  xobs,i − xobs

(6)

Modified Index of Agreement (MD)

The index of agreement ’MD’ detects the additive and relative variations in the variance and means of simulated and observed data, but due to the squared differences, the index is excessively susceptible extreme values. It can have a value from 0 to 1, where 1 is a perfect combination of the measured and forecast values [63]. The following equation is used to calculate: n M D = 1 − n i=1

(xobs − xsim ) j − xobs | + |xobs − x0bs |) j

i=1

(|xsim

(7)

68

R. Homsi et al.

4 Results 4.1 Quality Assessment of Observed Data All the data were found fine in terms of qualitative and quantitative assessments. In the case of precipitation, the days with precipitation more than 50 mm in a day were also not found. For temperature, higher values of minimum temperature compared to maximum temperature was not observed. All other qualitative criteria used for quality assessment (mentioned in the method section) were also found to fulfil by the observed data. The double mass curve for the annual rainfall time series of the selected stations is shown in Fig. 5. The mass curves in the figure show no breakpoint in the time series of the precipitation for the selected stations. Similar results were obtained for all the stations. From the theory of the double mass curve it is credence that the observed precipitation data is consistent.

Fig. 5 (a) The GPCC and observe rainfall time series; and (b) residuals between observed and GPCC rainfall at Aleppo station

Validation of Gridded Data Set Over Semi-arid Region of Syria

69

4.2 Assessment and Validation of Gauged Based Gridded Data The performances of gridded rainfall and temperature data were assessed based on their abilities to replicate observed rainfall and temperature at different stations located throughout Syria. In this study, observed station data were compared with the nearest grid point data to assess the performance of gridded data. Obtained results are discussed in the following sections.

4.3 Rainfall Time Series and Residuals Analysis The time series of 50 years of observed and the GPCC data for the period 1961 to 2010 were compared. The residuals for each of the stations were also generated. The time series plots discussed here were chosen from two different locations of the study area namely, Aleppo and Damascus. The yearly time series and the residuals for the two selected stations are presented in Figs. 6 and 7. A visual inspection of the 50 years’ time series of the 7 stations shows that most of the observed and the GPCC data matched well. Aleppo station (Fig. 6)

Fig. 6 (a) The GPCC and observe rainfall time series; and (b) residuals between observed and GPCC rainfall at Damascus station

70

R. Homsi et al.

Fig. 7 The cumulative distribution function of observed and GPCC rainfall at (a) Aleppo and (c) Damascus, and the probability distribution function of observed and GPCC rainfall at (b) Aleppo and (d) Damascus

represents such a station with a very good match, except for a few cases of over or underestimation. Slight overestimation or underestimations of the observed data by the GPCC were seen at some other stations. Figure 6 shows a situation where the GPCC slightly underestimated the observed values. Damascus station also showed some underestimation of the observed data (Fig. 7).

4.4 Rainfall Probability and Cumulative Distribution Functions The cumulative distribution function (CDF) and the probability density function (PDF) were obtained for each of the stations of the study area. Two of the CDF’s and

Validation of Gridded Data Set Over Semi-arid Region of Syria

71

PDF’s (Figs. 7a to 9d) for the earlier discussed time-series stations are presented. Both the CDFs and the PDFs show that the GPCC gridded data is in close relationship with the observed data. However, for some values, the observed data did not fit as good as others. This can be attributed to the missing data in the time series of the variable.

4.5 Spatial Distribution of Precipitation The spatial pattern maps of the mean annual precipitation of the observed and the GPCC were compared to show the ability of GPCC to show the spatial pattern of annual rainfall over Syria. A very close similarity in the spatial distribution of the observed and the GPCC rainfall was observed except in a few locations in the central part of the country (Fig. 8). This indicates that the GPCC gridded data can replicate the observed precipitation pattern.

4.6 Performance Evaluation of GPCC Rainfall Using Statistical Indices The ability of the GPCC rainfall to replicate the observed rainfall was assessed using statistical metrics namely, normalized root mean square error (NRMSE), percentage of bias (Pbias), modified coefficient of agreement (Md), Nash-Sutcliffe efficiency (NSE), and correlation coefficient (R2) (Table 2). The metrics showed near to ideal values for all stations except at two stations which are a bit far away indicating the ability of the GPCC gridded data in replicating the properties of the observed data.

5 Discussion From the extensive review of previous literature, it is credent that, so far, no studies have been conducted to validate the gridded data over Syria. A few studies were found which have evaluated various gridded data in neighboring countries. Some disparity was found in the evaluation of gridded data sets of various kinds with respect to the performance over various adjacent neighbor countries [5]. However, majority of the studies found GPCC data as the best performing gridded data in the Middle Eastern region. The summary of the studies conducted in the region are given in Table 3. To analyze the performance of the GPCC over Syria, various statistical as well as spatial and temporal comparison were made to validate the gridded data with observed station data over a period of 1951–2010. Double mass curve results showed that the annual rainfall time series showed no breakpoint in the graph representing

72

R. Homsi et al.

Fig. 8 Spatial distribution of the annual mean of observed (a) and GPCC (b) precipitation over Syria

Validation of Gridded Data Set Over Semi-arid Region of Syria

73

Table 2 Results of the statistical assessment of the performance of GPCC rainfall in replicating observed rainfall Station no.

Statistical metrics Pbias

NRMSE

NSE

md

R2

40,001

3.34

38.99

0.83

0.74

0.94

40,007

4.64

10.24

0.9

0.97

0.9

40,022

5.78

74.37

0.74

0.72

0.79

40,030

2.39

5.14

0.9

0.97

0.19

40,080

1.73

6.75

0.86

0.97

0.88

40,045

6.09

13.46

0.61

0.71

0.73

40,061

9.39

19.78

0.51

0.86

0.57

the consistency of the observed data of various stations. The PDF and CDF of the observed and GPCC data were plotted to find the relationship of the GPCC with the observed data. It was observed that at various station the gridded data is in close relationship with the gridded data, which strengthens the hypothesis that the GPCC can perform well over the Syria. Time series analysis of 50 years were compared along with the residual of each stations. The comparison showed that the GPCC data matched well with the observed data with a slight under or over estimation at some stations. A spatial distribution of the annual mean rainfall over Syria was plotted for the observed as well as GPCC data. The spatial distribution of the GPCC was much more similar in most part of the country which represents the ability of the gridded data to replicate the annual mean precipitation over most part of the country. The R2 , PBIAS, NRMSE, NSE and md are broadly used to evaluate the performance of gridded data products. The results of statistical analysis of the GPCC, taking observed data as reference, shows that the GPCC performed well in overall statistical evaluation having lower values in error term like PBIAS, NRMSE and higher correlation values in R2 , NSE and md. The that GPCC can be used confidently for the hydrological and water resource management over Syrian region.

74

R. Homsi et al.

Table 3 Review of recent studies on the validation of gridded precipitation data in countries surrounding Syria Reference

Place

Precipitation product

Major conclusion

Black et al. [11]

Middle East

GPCC

GPCC can be used to show precipitation change in the Middle East

Heiblum et al. [23]

East Mediterranean

TRMM

TRMM can be used to identify the area favorable for precipitation formation

Raziei et al. [47]

Iran

APHRODITE

APHRODITE can show the precipitation variation in Iran

Kheimi et al. [35]

Saudi Arabia

TRMM, TRMM 3B42 offers the best CMORPH, possibility for accurate GSMaP_MVK and estimation of precipitation PERSIANN

Javanmard et al. [30]

North Iran

TRMM

Spatial distribution of TRMM able to show main precipitation patterns

Karakoc et al. [32]

Turkey

TRMM

TRMM shows a higher annual precipitation amount than the ground data

Wehbe et al. [62]

United Arab Emirates

GPCC, TRMM, TMPA, WM, and CMORPH

TMPA shows the highest overall agreement with the observational network

Sarmadi et al. [52]

Iran

GPCC

GPCC can be used for climatic classification

Sharifi et al. [57]

Iran

IMERG, TRMM, ECMWF

IMERG is superior to the other products

Derin et al. [16]

West of Turkey

TRMM, TMPA, CMORPH, MPE

TMPA and MPE products underestimate precipitation

Katiraie-Boroujerdy et al. [33]

Iran

GLDAS, MERRA, MERRA is a preferred GPCC, and alternative of observed rainfall APHRODITE for drought analysis

Ghalhari et al. [20]

Iran

APHRODITE

APHRODITE can show the precipitation variation in Iran

El Kenawy et al. [18]

Saudi Arabia

APHRODITE, GPCC, PRINCETON, UDEL, CRU and PREC/L

GPCC is more capable relative to other products in replicating observed precipitation

Salman et al. [51]

Iraq

GPCC, CRU, APHRODITE, UDel, TRMM, ERA, and MERA

GPCC precipitation data was found best in term of replicating observed precipitation at 21 out of 41 stations

Validation of Gridded Data Set Over Semi-arid Region of Syria

75

6 Conclusion The assessment of variability and precipitation data is intuitive prior to the use of gridded data in any region owing to the difference in distribution, quality of stations data, topography and capability of various data products. GPCC performed better in Middle eastern region due to various reasons like, the use of large number of observed stations in development of gauge-based gridded data sets, various research have reported that over 85000 observed stations were used all around the globe (Schneider et al. 2014). Another factor for the outperformance of GPCC over Syria can be the robust procedures applied during the development of the data and the quality control measures. This study validated the GPCC precipitation over the Syrian region using different statistical and spatial procedures. The results show that GPCC over Syria can be used in hydrometeorological studies as well as water resource management keeping in view the limited number of stations in the arid region.

References 1. Ahmed K. et al (2019) Quantitative assessment of precipitation changes under CMIP5 RCP scenarios over the northern sub-Himalayan region of Pakistan. Environ Develop Sustain 1–15 2. Ahmed K, Shahid S, bin Harun S, Wang X-j (2016) Characterization of seasonal droughts in Balochistan Province, Pakistan. Stochast Environ Res Risk Aassess 30(2):747–762 3. Ahmed K, Shahid S, Chung E-S, Ismail T, Wang X-J (2017) Spatial distribution of secular trends in annual and seasonal precipitation over Pakistan. Climate Res 74(2):95–107 4. Ahmed K, Shahid S, Nawaz N, Khan N (2019) Modeling climate change impacts on precipitation in arid regions of Pakistan: a non-local model output statistics downscaling approach. Theoret Appl Climatol 137(1):1347–1364. https://doi.org/10.1007/s00704-018-2672-5 5. Ahmeda K, Shahida S, Alid RO, Haruna SB, Wange X-J (2017) Evaluation of the performance of gridded precipitation products over Balochistan Province Pakistan. Desalination 1:14 6. Al-Mohammdawi JA, Al-Abadi AM, Al-Ali AK, Shahid S, Fryar A, Wang X (2022) Assessing the spatial and temporal variations of terrestrial water storage of Iraq using GRACE satellite data and reliability–resiliency–vulnerability indicators. Arab J Geosci 15(4):342. https://doi. org/10.1007/s12517-022-09626-x 7. Alazard M, Leduc C, Travi Y, Boulet G, Salem AB (2015) Estimating evaporation in semi-arid areas facing data scarcity: example of the El Haouareb dam (Merguellil catchment, Central Tunisia). J Hydrol Region Stud 3:265–284 8. Becker A, Finger P, Meyer-Christoffer A, Rudolf B, Schamm K, Schneider U, Ziese M (2013) A description of the global land-surface precipitation data products of the Global Precipitation Climatology Centre with sample applications including centennial (trend) analysis from 1901present. Earth Syst Sci Data 5(1):71 9. Beharry SL, Clarke RM, Kurmarsingh K (2014) Precipitation trends using in-situ and gridded datasets. Theoret Appl Climatol 115(3–4):599–607 10. Belo-Pereira M, Dutra E, Viterbo P (2011) Evaluation of global precipitation data sets over the Iberian Peninsula. J Geophys Res Atmosph 116(D20) 11. Black E, Brayshaw DJ, Rambeau CM (2010) Past, present and future precipitation in the Middle East: insights from models and observations. Philos Trans Royal Soc Math Phys Eng Sci 368(1931):5173–5184 12. Chen F-W, Liu C-W (2012) Estimation of the spatial rainfall distribution using inverse distance weighting (IDW) in the middle of Taiwan. Paddy Water Environ 10(3):209–222

76

R. Homsi et al.

13. Chen W, Jiang Z, Li L (2011) Probabilistic projections of climate change over China under the SRES A1B scenario using 28 AOGCMs. J Clim 24(17):4741–4756 14. Chokngamwong R, Chiu LS (2008) Thailand daily rainfall and comparison with TRMM products. J Hydrometeorol 9(2):256–266 15. Dee D, Balmaseda M, Balsamo G, Engelen R, Simmons A, Thépaut J-N (2014) Toward a consistent reanalysis of the climate system. Bull Am Meteor Soc 95(8):1235–1248 16. Derin Y, Yilmaz KK (2014) Evaluation of multiple satellite-based precipitation products over complex topography. J Hydrometeorol 15(4):1498–1516 17. Dinku T, Ceccato P, Grover-Kopec E, Lemma M, Connor S, Ropelewski C (2007) Validation of satellite rainfall products over East Africa’s complex topography. Int J Remote Sens 28(7):1503–1526 18. El Kenawy AM, McCabe MF (2016) A multi-decadal assessment of the performance of gaugeand model-based rainfall products over Saudi Arabia: climatology, anomalies and trends. Int J Climatol 36(2):656–674 19. Eum H-I, Gachon P, Laprise R (2014) Developing a likely climate scenario from multiple regional climate model simulations with an optimal weighting factor. Clim Dyn 43(1–2):11–35 20. Ghalhari GF, Roudbari AD, Asadi M (2016) Identifying the spatial and temporal distribution characteristics of precipitation in Iran. Arab J Geosci 9(12):1–12 21. Harris I, Jones PD, Osborn TJ, Lister DH (2014) Updated high-resolution grids of monthly climatic observations–the CRU TS3. 10 Dataset. Int J Climatol 34(3):623–642 22. Haylock M, Hofstra N, Klein Tank A, Klok E, Jones P, New M (2008) A European daily highresolution gridded data set of surface temperature and precipitation for 1950–2006. J Geophys Res Atmosph 113(D20) 23. Heiblum R, Koren I, Altaratz O (2011) Analyzing coastal precipitation using TRMM observations. Atmos Chem Phys 11(24):13201–13217 24. Hereher ME (2016) Recent trends of temperature and precipitation proxies in Saudi Arabia: implications for climate change. Arab J Geosci 9(11):1–9 25. Hijmans RJ, Cameron SE, Parra JL, Jones PG, Jarvis A (2005) Very high resolution interpolated climate surfaces for global land areas. Int J Climatol 25(15):1965–1978 26. Homsi R, Shahid S, Iqbal Z, Ali AM, Ziarh GF (2021) Historical trends in crop water demand over semiarid region of Syria. Theoret Appl Climatol 146(1):555–566. https://doi.org/10.1007/ s00704-021-03751-5 27. Homsi R, Shiru MS, Shahid S, Ismail T, Harun SB, Al-Ansari N, Yaseen ZM (2020) Precipitation projection using a CMIP5 GCM ensemble model: a regional investigation of Syria. Eng Appl Comput Fluid Mech 14(1):90–106. https://doi.org/10.1080/19942060.2019.1683076 28. Iqbal Z, et al (2020) Evaluation of global climate models for precipitation projection in subHimalaya region of Pakistan. Atmosph Res. 245:105061. https://doi.org/10.1016/j.atmosres. 2020.105061 29. Iqbal Z, Shahid S, Ahmed K, Ismail T, Nawaz N (2019) Spatial distribution of the trends in precipitation and precipitation extremes in the sub-Himalayan region of Pakistan. Theoret Appl Climatol 137(3–4):2755–2769 30. Javanmard S, Yatagai A, Nodzu M, BodaghJamali J, Kawamoto H (2010) Comparing highresolution gridded precipitation data with satellite rainfall estimates of TRMM_3B42 over Iran. Adv Geosci 25:119–125 31. Johnston K (2004) ArcGIS 9: using ArcGIS geostatistical analyst: Esri Press 32. Karakoc U, Patil S (2016) Comparison of TRMM satellite and ground-based precipitation data for predicting streamflow in Kucuk Menderes river basin, Turkey. In: EGU general assembly conference abstracts, pp EPSC2016-1467 33. Katiraie-Boroujerdy P-S, Nasrollahi N, Hsu K-L, Sorooshian S (2016) Quantifying the reliability of four global datasets for drought monitoring over a semiarid region. Theoret Appl Climatol 123(1–2):387–398 34. Khan N, Sachindra DA, Shahid S, Ahmed K, Shiru MS, Nawaz N (2020) Prediction of droughts over Pakistan using machine learning algorithms. Adv Water Resour 139:103562. https://doi. org/10.1016/j.advwatres.2020.103562

Validation of Gridded Data Set Over Semi-arid Region of Syria

77

35. Kheimi MM, Gutub S (2014) Assessment of remotely sensed precipitation products across the Saudi Arabia region. In: 6th International conference on water resources and arid environments, p 1617 36. Liebmann B, Allured D (2005) Daily precipitation grids for South America. Bull Am Meteor Soc 86(11):1567–1570 37. Luo M, Feng J, Xu Z, Wang Y, Dan L (2019) Evaluating the performance of five twentiethcentury reanalysis datasets in reproducing the severe drought in northern China during the 1920s–1930s. Theoret Appl Climatol 137(1):187–199. https://doi.org/10.1007/s00704-0182591-5 38. Mallick J, et al. (2022) Recent changes in temperature extremes in subtropical climate region and the role of large-scale atmospheric oscillation patterns. Theoret Appl Climatol. https://doi. org/10.1007/s00704-021-03914-4 39. Matsuura K, Willmott CJ (2012) Terrestrial precipitation: 1900–2010 gridded monthly time series. University of Delaware 40. Maurer EP, Wood AW, Adam JC, Lettenmaier DP, Nijssen B (2002) A long-term hydrologically based dataset of land surface fluxes and states for the conterminous United States. J Clim 15(22):3237–3251 41. Moazami S, Golian S, Kavianpour MR, Hong Y (2013) Comparison of PERSIANN and V7 TRMM Multi-satellite Precipitation Analysis (TMPA) products with rain gauge data over Iran. Int J Remote Sens 34(22):8156–8171 42. Nash JE, Sutcliffe JV (1970) River flow forecasting through conceptual models part I—a discussion of principles. J Hydrol 10(3):282–290 43. Nashwan M, Shahid S, Dewan A, Ismail T, Alias N (2019) Performance of five high resolution satellite-based precipitation products in arid region of Egypt: an evaluation. Atmos Res 236:104809. https://doi.org/10.1016/j.atmosres.2019.104809 44. Nashwan MS, Shahid S, Chung E-S (2019) Development of high-resolution daily gridded temperature datasets for the central north region of Egypt. Scientific data 6(1):1–13 45. Perkins S, Pitman A, Holbrook N, McAneney J (2007) Evaluation of the AR4 climate models’ simulated daily maximum temperature, minimum temperature, and precipitation over Australia using probability density functions. J Clim 20(17):4356–4376 46. Prakash S, Gairola RM, Mitra AK (2015) Comparison of large-scale global land precipitation from multisatellite and reanalysis products with gauge-based GPCC data sets. Theoret Appl Climatol 121(1):303–317. https://doi.org/10.1007/s00704-014-1245-5 47. Raziei T, Bordi I, Santos JA, Mofidi A (2013) Atmospheric circulation types and winter daily precipitation in Iran. Int J Climatol 33(9):2232–2246 48. Sa’adi Z, Yaseen ZM, Muhammad MKI, Iqbal Z (2022) On the prediction of methane fluxes from pristine tropical peatland in Sarawak: application of a denitrification–decomposition (DNDC) model. Environ Sci Pollut Res. https://doi.org/10.1007/s11356-021-17917-1 49. Salehie O, Ismail TB, Shahid S, Sammen SS, Malik A, Wang X (2022) Selection of the gridded temperature dataset for assessment of thermal bioclimatic environmental changes in Amu Darya River basin. Stochastic Environ Res Risk Aassess. https://doi.org/10.1007/s00477022-02172-8 50. Salman SA, Shahid S, Ismail T, Ahmed K, Wang X-J (2018) Selection of climate models for projection of spatiotemporal changes in temperature of Iraq with uncertainties. Atmos Res 213:509–522 51. Salman SA, Shahid S, Ismail T, Al-Abadi AM, Wang X-J, Chung E-S (2019) Selection of gridded precipitation data for Iraq using compromise programming. Measurement 132:87–98 52. Sarmadi F, Shokoohi A (2015) Regionalizing precipitation in Iran using GPCC gridded data via multivariate analysis and L-moment methods. Theoret Appl Climatol 122(1):121–128 53. Schiemann R, Liniger M, Frei C (2010) Reduced space optimal interpolation of daily rain gauge precipitation in Switzerland. J Geophys Res Atmosph 115(D14) 54. Schneider U, Becker A, Finger P, Meyer-Christoffer A, Ziese M, Rudolf B (2014) GPCC’s new land surface precipitation climatology based on quality-controlled in situ data and its role in quantifying the global water cycle. Theoret Appl Climatol 115(1–2):15–40

78

R. Homsi et al.

55. Searcy J, Hardison C (1962) Double-mass curves. Manual of hydrology—part 1—general surfacewater techniques, methods and practices of the geological survey. US Geological Survey water-supply paper. Waterlow and Sons 56. Shahid S, Alamgir M, Wang XJ, Eslamian S (2017) Climate change impacts on and adaptation to groundwater. In: Handbook of Drought and Water Scarcity, pp. 107–124. CRC Press 57. Sharifi E, Steinacker R, Saghafian B (2016) Assessment of GPM-IMERG and other precipitation products against gauge data under different topographic and climatic conditions in Iran: preliminary results. Remote Sens 8(2):135 58. Shiru MS, Chung ES, Shahid S, Wang XJ (2022) Comparison of precipitation projections of CMIP5 and CMIP6 global climate models over Yulin China. Theoret Appl Climatol 147(1):535–548. https://doi.org/10.1007/s00704-021-03823-6 59. Song YH, Chung E-S, Shahid S (2022) Uncertainties in evapotranspiration projections associated with estimation methods and CMIP6 GCMs for South Korea. Sci Total Environ 825:153953. https://doi.org/10.1016/j.scitotenv.2022.153953 60. Varikoden H, Revadekar J, Choudhary Y, Preethi B (2015) Droughts of Indian summer monsoon associated with El Niño and Non-El Niño years. Int J Climatol 35(8):1916–1925 61. Vose RS (1992) The Global Historical Climatology Network: Long-term monthly temperature, precipitation, sea level pressure, and station pressure data 62. Wehbe Y, Ghebreyesus D, Temimi M, Milewski A, Al Mandous A (2017) Assessment of the consistency among global precipitation products over the United Arab Emirates. J Hydrol Region Stud 12:122–135 63. Willmott CJ (1981) On the validation of models. Phys Geogr 2(2):184–194 64. Willmott CJ (1982) Some comments on the evaluation of model performance. Bull Am Meteor Soc 63(11):1309–1313 65. Yatagai A, Arakawa O, Kamiguchi K, Kawamoto H, Nodzu MI, Hamada A (2009) A 44-year daily gridded precipitation dataset for asia based on a dense network of Rain Gauges. SOLA 5:137–140. https://doi.org/10.2151/sola.2009-035 66. Yildirim HA, Altinsoy H (2017) Chaos and trend analysis of monthly precipitation over Arabian Peninsula and Eastern Mediterranean. Arab J Geosci 10(2):1–12 67. You Q, Kang S, Pepin N, Yan Y (2008) Relationship between trends in temperature extremes and elevation in the eastern and central Tibetan Plateau, 1961–2005. Geophys Res Lett 35(4) 68. You Q, Min J, Zhang W, Pepin N, Kang S (2015) Comparison of multiple datasets with gridded precipitation observations over the Tibetan Plateau. Clim Dyn 45(3):791–806. https://doi.org/ 10.1007/s00382-014-2310-6 69. Zhang L, Zhou T (2011) An assessment of monsoon precipitation changes during 1901–2001. Clim Dyn 37(1–2):279–296 70. Ziarh GF, Asaduzzaman M, Dewan A, Nashwan MS, Shahid S (2021) Integration of catastrophe and entropy theories for flood risk mapping in peninsular Malaysia. J Flood Risk Manage 14(1):e12686

Flood Risk Assessment Considering the Effect of Covid-19 Pandemic in the Municipality of Balayan, Batangas C. E. F. Monjardin, F. A. R. Cala, D. C. D. Po, and M. A. M. Sy

Abstract The constant threats brought upon by hydrometeorological hazards alongside the COVID-19 pandemic have rendered traditional flood risk assessments ineffective in aiding developers and policymakers in establishing proper intervention measures. With the numerous large-scale disasters that have occurred in the Philippines during this pandemic, it has been reported that an alarmingly transmissible virus rampant throughout the country oftentimes causes the neglection or mis-prioritization of personal preservation measures in times of disarray. By utilizing the definitive findings established in the preceding study, the Flood Risk Assessment for Mitigation and Effective Response Project, this research furthers literature by defining the different considerations that significantly affect flood risk. The overall study determines related parameters for flooding, which were obtained from the preceding study, and for COVID-19 to be utilized to establish the flood risk assessment. Along with the considerations made for COVID-19, it was observed that both Hospital Bed Capacity and Viral Load addresses the risk stemming from the pandemic. Through this, the researchers established a flood risk assessment for Balayan, Batangas while considering the effects of COVID-19 pandemic. The flood risk assessment utilizes the following tools: 1) the Analytical Hierarchy Process to provide relative numerical weights, and 2) the ArcGIS to generate an overall risk map for the Region of Interest set within Balayan. Furthermore, the relative numerical weights were also necessary to produce an aggregated result of risk levels in generating the overall Risk index map. As a result, the COVID-19 parameters thus validate the conflicting objectives C. E. F. Monjardin (B) · F. A. R. Cala · D. C. D. Po · M. A. M. Sy School of Civil, Environmental, and Geological Engineering, Mapua University, 658 Muralla St, Intramuros, 1002 Manila, Metro Manila, Philippines e-mail: [email protected] F. A. R. Cala e-mail: [email protected] D. C. D. Po e-mail: [email protected] M. A. M. Sy e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023 I. K. Othman et al. (eds.), Proceedings of the 5th International Conference on Water Resources (ICWR) – Volume 2, Lecture Notes in Civil Engineering 365, https://doi.org/10.1007/978-981-99-3577-2_6

79

80

C. E. F. Monjardin et al.

when addressing risk derived from COVID-19 with flood occurrences. This instance was similarly observed to occur between the newly established peri-pandemic risk map as opposed to the pre-pandemic risk map as well. For this reason, a participatory approach to mitigate the intersection of community health resilience and disaster risk reduction justifies the applications of the Sendai Framework for Disaster Risk Reduction which is supported by Health Emergency and Disaster Risk Management. Keywords Flood Risk Assessment · COVID-19 Pandemic · Analytical Hierarchy Process · ArcGIS · Sendai Framework for Disaster Risk Reduction

1 Introduction Flood Risk Assessment (FRA) is a vital tool for Disaster Risk Reduction (DRR) that serves as a guide for policymakers in implementing science-based, well-structured policy interventions and flood evacuation protocols by determining areas highly susceptible to flooding events [4, 5]. FRA requires identifying, understanding, and quantifying the wide range of relevant factors and parameters affecting risk [1]. Moreover, the presence of community variabilities and uncertainties adds to the complexity in implementing flood mitigation strategies. To simplify such complexity, relevant parameters are categorized into Hazard, Vulnerability, and Exposure indices, constituting the formulation of Risk [14, 15], which serves as a basis in producing flood risk maps [2]. Hence, to support developers and policymakers, the main tools utilized are risk maps created by simple overlaying of hazard, vulnerability, and exposure maps to identify risk levels of different areas on which to intervene; thus, essential for implementing flood evacuation plans [3, 18]. However, a secondary challenge developers face is the disarray that the widespread global pandemic caused by the CoronaVirus Disease 2019 (COVID-19) has brought upon World Health Organization (WHO). As natural hazards inevitably continue to wreak havoc despite the current dire circumstances, an ever-increasing challenge must be faced with the complexity that the global pandemic brings; established intervention measures constantly need to be reassessed. With the large-scale flooding in the Philippines, evacuees had to face the threats of COVID-19 during their time in the evacuation centers as well [10]. Accommodating high volumes of displaced people under given parameters raises several health concerns, especially with a highly transmissible disease rampant [13]. As such disease calls for proper social distancing and maintenance of proper hygiene, which can easily be neglected in the disarray of a disaster, it can be hypothesized that COVID-19 compromises flood evacuation protocols and contributes to the risk factor of flooding.

Flood Risk Assessment Considering the Effect of Covid-19 Pandemic …

81

Therefore, the identification of flood mitigation measures based on conventional FRA parameters is considered insufficient in handling multi-hazard assessments. Hence, DRR approaches require an account to reassess and compare the risk derived from flooding with the risk derived from the COVID-19 pandemic to implement evacuation protocol and plans, quantitatively assessing the primary consequences of flood hazard, as well as the secondary effects of the transmissible virus at a community level. The worsening impact of the pandemic on the communities has shaped the perspective of communities in DRR and paradoxically catalyzed the transformations needed for sustainable development and resilience of communities [15, 16]. The concepts of Sendai Framework for Disaster Risk Reduction (SFDRR) causally expound relationships between FRA, risk mitigation, community preparedness, resilience and recovery, and other global agendas, demonstrating how these concepts, at a local-level action, should be integrated for sustainable development from community levels to a global scale [14, 17]. The research setting was in Balayan, it is a coastal municipality situated at the leftmost part of Batangas, placing it alongside Balayan Bay (Fig. 1 and Fig. 2) thus, frequently exposing it to flooding [9].

Fig. 1 Municipal map of Balayan, Batangas

82

C. E. F. Monjardin et al.

Fig. 2 Region of interest (ROI)

2 Methodology 2.1 Phase 1. Data Analysis Utilizing the Analytical Hierarchy Process (AHP) The study utilizes the Analytical Hierarchy Process (AHP) as the main statistical instrument of the present study (Fig. 3). Data analysis revolves around the three (3) indices of Hazard, Vulnerability, and Exposure. The utilization of AHP allows the researchers to analyze multiple parameters, reducing the gravity of complexity in analyzing such judgments in FRA. AHP focuses on the analysis of the numerical weights of the parameters relative to each parameter of the same index classification. A hierarchical tree structure of the AHP of the present study can be established based on a study conducted by [17].

2.2 Phase 2. Flood Risk Assessment (FRA) The numerical weights calculated were synthesized to produce a general equation for risk incorporating all parameters under each index; subsequently, inputted into ArcGIS to generate maps for Hazard, Vulnerability, and Exposure indices. The mapping outputs indicate which areas in Balayan possess different levels of Hazard, Vulnerability, and Exposure, respectively.

Flood Risk Assessment Considering the Effect of Covid-19 Pandemic …

83

Fig. 3 Methodological framework of the present study

2.3 Phase 3. Post-Flood Risk Assessment (FRA) The established mapping from the previous phase suffices the basis for the researchers on suggestions and recommendations for potential measures and revised protocols on flood evacuation within Balayan, Batangas during the COVID-19 pandemic.

3 Results and Discussion The data gathered in the study can be divided into two (2) components: 1) floodrelated risk, and 2) COVID-related risk. The concepts of risk and resilience, as the main integrant of SFDRR, were adduced in classifying the different risk parameters under three (3) distinct indices depending on their nature and theme: 1) Hazard, 2) Vulnerability, and 3) Exposure. As such, it was similarly addressed in the present study to further incorporate the COVID-19 pandemic-related parameters.

3.1 Flood-Related Risk Parameters The flood-related risk parameters, identified from the preceding study, act as crucial components in assessing and establishing the flood risk levels within Balayan, constituting the overall risk assessment; thus, serving as the base structure of the present study. Figures 4, 5 and 6 present the map of each parameter, under each index,

84

C. E. F. Monjardin et al.

according to its respective normalized scores, ranging from 0.0 to 1.0, from the lowest to the highest index level, respectively.

Fig. 4. Flood-related individual data map for parameters under hazard index.

Fig. 5. Flood-related individual data map for parameters under vulnerability index.

Flood Risk Assessment Considering the Effect of Covid-19 Pandemic …

85

Fig. 6. Flood-related individual data map for parameters under exposure index.

3.2 COVID-19 Pandemic-Related Parameters The COVID-19 pandemic-related parameters considered for the study focus more on those deemed to influence the objectives of the usual flood evacuation protocols, which are: 1) Hospital Bed Capacity, and 2) Viral Loads (Fig. 7). Other parameters and factors derived from the COVID-19 pandemic were not considered due to certain limitations such as time restrictions and the incapabilities of the researchers to obtain the appropriate data due to pandemic restrictions.

Fig. 7. COVID-19 pandemic-related individual data maps a hospital bed capacity (HBC), b viral loads (VL).

86

C. E. F. Monjardin et al.

a) Hospital Bed Capacity (HBC): It has been derived that the public health capacity and capability of Balayan in accommodating COVID-related patients shall be measured through quantifying the number of hospital beds available in each health facility, known to accept COVID-related patients, with respect to the population it has been designed to cater towards [6]. The parameter HBC was deemed necessary and classified under the Vulnerability index [8]. The HBC determines the capacity and resources availability of a community to cope with the disturbances derived from the COVID-19 pandemic [8, 12]. b) Viral Loads (VL): The parameter VL was deemed necessary and classified under the Exposure index [11]. To examine the causal link between flood risk and COVID-19 case evolution, the study cites a time-reliant parameter, wherein a progression of COVID-19 cases is tracked. However, as this cannot be adequately represented through a static map, a solution was adopted wherein it was observed that VL can be directly correlated with the amount of locally recorded transmissions within a Barangay; similar with how population density and spatial distribution are treated [6, 8].

3.3 Relative Numerical Weights or Priorities of the Parameters As the output obtained from the research responses, the importance factors served as preliminaries to derive the numerical weights for each parameter relative to their corresponding index classifications (Table 1). As further validation of the responses was deemed necessary, the researchers employed a Consistency Ratio (CR) as a part of the consideration process. By allotting a ten percent (10%) tolerance on the threshold for the CR, any response valued above this was disregarded. Table 1 Numerical weights of hazard parameters derived from importance factors

Hazard Index Distance from River Network (DR)

10.686%

Land Cover (LC)

10.686%

Elevation (EL)

9.575%

Slope (SL)

9.575%

Flood Height (FH)

27.689%

Arrival Time of Flood (AT)

31.790%

Flood Risk Assessment Considering the Effect of Covid-19 Pandemic …

87

3.4 Flood Risk Assessment (AHP-GIS Scheme) a) Hazard, Vulnerability, and Exposure Index Maps: The hazard map shown in Fig. 8a was the yielding result of the integration of all identified parameters attributed under the Hazard index within the ROI. No additional COVID-19 pandemic-related parameters were included for the index. The vulnerability map shown in Fig. 8b was the yielding result of integrating all identified parameters attributed under the Vulnerability index within the ROI. As observed in Fig. 5, the individual data maps of flood-related risk parameters have shown identical patterns of the areas posing higher index levels. As such, varying vulnerability levels are scattered, partially covering the floodplains near the Binambang River and the Balayan Bay. However, due to the parameter HBC established in the present study, it was deemed the defining factor that consolidates the overall contribution of the Vulnerability index towards risk. Outlying areas of ROI not covered by the proximity of the health facilities have posed intensified vulnerability levels on the respective areas, as visually represented in Fig. 8b. Moreover, as shown in Table 2, the derived numerical weight of HBC has been observed to influence its high relative contribution towards risk computation in the area. By visually examining the encompassing exposure map in Fig. 8c, it can be discerned that a scattered-like result is represented. This occurrence has developed primarily due to the individual data maps not having yielded any substantially wide range of vast clusters of index within a particular area. As such, when overlaying the individual data maps attributed to the parameters, initial results may suggest that no apparent exposure level from the individually mapped parameters defines the cells

Fig. 8 Index map of Balayan, Batangas (a) hazard, (b) vulnerability, (c) exposure

88 Table 2 Numerical weights of vulnerability parameters derived from importance factors

Table 3 Numerical weights of exposure parameters derived from importance factors

C. E. F. Monjardin et al.

Vulnerability Index Gender Ratio (GR)

2.579%

Age of Individual (AI)

3.335%

Emergency Preparedness (EP)

19.374%

Amenities near the Area / Necessities (NA)

13.374%

Type of Build-Up Structure (TB)

6.636%

Physical Health of Individuals (PHI)

19.374%

Distance of the Nearest Evacuation Area (DEA)

13.836%

Highest Educational Attainment (EA)

2.828%

Average Income (AI)

2.828%

Hospital Bed Capacity (HBC)

15.374%

Exposure Index Population Density (PD)

18.505%

Number of Households (NH)

19.382%

Type of Building/Type of Occupancy (TBO)

21.467%

Viral Load (Active COVID-19 Cases) (VL)

40.647%

enough to form large clusters of uniform exposure levels correspondingly. In the small portion of the map shown in Fig. 8c, wherein “high” to “very high” levels of risk can be observed, such data can be utilized to draw initial assumptions on whether VL emerges as the most prevailing factor in the Exposure index. As VL was established to yield an equal distribution of risk within a Barangay’s boundary, shows that Brgy. Caloocan represents the highest level of risk throughout the municipality. b) Flood Risk Incorporating the Effect of the COVID-19 Pandemic: The flood risk map is the overall risk evaluation for Balayan, Batangas, which consolidates the results drawn from the three (3) indices onto a single map. As a result, shown in Fig. 9b, the risk map displays the overall results for the FRA of the study considering risk drivers derived from both flooding hazards and the COVID-19. The risk map visually presents significant differences when compared to the individual index maps, as these differences result from Risk being a product of Hazard, Vulnerability, and Exposure; thus, incurring the represented variation in risk levels. Attributing from the previously presented results, the final FRA, incorporating the effects of the COVID-19 pandemic, presents the following overlying notions. First, as risk is established to be a function of the three indices, the parameters belonging to the Hazard index was observed as the prevailing factor which contributes to a larger value of risk. Second, it can be observed that the low-lying areas near the shoreline and river networks generally pose a higher level of risk due to the greater threat of

Flood Risk Assessment Considering the Effect of Covid-19 Pandemic …

89

Fig. 9 Flood risk map of Balayan a pre-pandemic, b peri-pandemic

flooding. As observed in the results, the hazard map validates the said alleged notion that the areas near the main river network and the coastlines, specifically the mouth of Binambang River which flushes out into the Balayan Bay, poses a substantially higher level of contribution towards risk compared to areas further from both water bodies. By examining the effect of VL on the Exposure index map, it can be observed that the highest risk levels remain and coincide with a majority of the pre-pandemic results. The highest level of risk can be found where VL is most prevalent, but data also shows an altered prioritization of risk levels in areas further from the typical flood-prone areas. With VL holding the greatest numerical weight of 40.647% in the Exposure index, an observation can subsequently be drawn that the typically flood-associated parameters under the Exposure index are considered less significant. Moreover, by examining the effect of HBC on the Vulnerability index map, it can also be observed that the “moderate” to “high risk” level areas are located similar to the pre-pandemic delineation. Outside the cluster of higher risk levels, however, the remainder of the index map presents a circular coverage area that yields the two lowest levels of risk. With this, a conclusive assessment can then be established that an individual’s vulnerability increases the further they are situated away from the group of hospital facilities in the municipality. Moreover, the HBC, with a numerical weight of 15.374%, holds the third highest within the Vulnerability index, and its importance can be accordingly observed as reflected on its corresponding index map.

90

C. E. F. Monjardin et al.

4 Conclusion and Recommendation 4.1 Conclusion For the FRA of the present study, secondary considerations were addressed incorporating the effects of COVID-19 by establishing the additional relevant parameters derived from COVID-19. The study was solely able to include the VL and HBC parameters, underlain by the indices of Vulnerability, and Exposure, respectively; thus, disregarding other derived parameters due to limitations. The data adduced in the present study were specifically within Balayan, deemed pertinent to further support its corresponding identified parameters. These functioned as preliminaries for spatial analysis, constituting the primary components of FRA; thus, serving as the preparatory for AHP. The numerical weights derived from the implementation of AHP determine the distinctiveness of the relative risk contribution of flood-risk parameters of the present study as opposed to the FRAMER Project. Such that, a different set of experts and professionals were called upon to generate qualitative judgments, translated to quantifiable outputs for FRA. The derived numerical weights for each established flood risk and COVID-19 parameters determine their individual priorities towards its corresponding index. The following parameters were observed that significantly affected the overall risk index: 1) For Hazards: Arrival Time of Flood (31.790%), and Flood Height (27.689%). 2) For Vulnerability: Emergency Preparedness (19.374%), Physical Health of Individuals (19.374%), and Hospital Bed Capacity (15.374%). 3) For Exposure: Viral Loads (40.647%), and Type of Building/ Type of Occupancy (21.467%). Upon aggregating all established parameters through mathematical processes and spatial analysis, the overall risk map delineates the consolidation of the collective effects of Hazard, Vulnerability, and Exposure indices towards risk. The map visually represents the varying risk levels within Balayan, explicitly covering the ROI as defined in the study. Accordingly, it can be observed that the following Barangays presented the highest risk due to both flooding and COVID-19: Brgy. 1, Brgy. 2, Brgy. 3, Brgy. 4, Brgy. 5, Brgy. 8, Brgy. 10, Brgy. 11, Canda, Caloocan, Dalig, Magabe, Navotas, Sambat, and Santol which generates a risk factor ranging from 0.134121 0.286501.

4.2 Recommendations Further research warrants the citation of the SFDRR principle towards overall FRA considering the COVID-19 pandemic. The intersection of community health and DRR has emerged through the years as a major concern; such that, health is commonly recognized as both an objective and an outcome of DRR. The SFDRR aims to address

Flood Risk Assessment Considering the Effect of Covid-19 Pandemic …

91

community health resilience, and the integration of health and DRR ensures that the concepts of SFDRR are implemented in the study.

References 1. Brauch HG (2005) Threats, challenges, vulnerabilities and risks in environmental and human security. UNU-EHS 2. Cabrera JS, Lee HS (2020) Flood risk assessment for Davao Oriental in the Philippines using geographic information system-based multi-criteria analysis and the maximum entropy model. J Flood Risk Manag 13(2):e12607. https://doi.org/10.1111/jfr3.12607 3. Danumah JH, Odai SN, Saley BM, Szarzynski J, Thiel M, Kwaku A, Akpa LY (2016) Flood risk assessment and mapping in Abidjan district using multi-criteria analysis (AHP) model and geoinformation techniques, (cote d’ivoire). Geoenvironmental Disasters 3(1):1–13. https://doi. org/10.1186/s40677-016-0044-y 4. Franta B, Roa-Quiaoit H, Lo D, Narisma G (2016) Climate disasters in the Philippines: a case study of immediate causes and root drivers from Cagayan de Oro, Mindanao and Tropical Storm Sendong/Washi. Belfer Center for Science and International Affairs. Harvard University, Cambridge. https://www.xu.edu.ph/images/2017/doc/february/climate-change.pdf 5. Manfreda S, Di Leo M, Sole A (2011) Detection of flood-prone areas using digital elevation models. J Hydrol Eng 16(10):781–790. https://doi.org/10.1061/(ASCE)HE.1943-5584. 0000367 6. Mavroulis S, Mavrouli M, Lekkas E (2021) Geological and hydrometeorological hazards and related disasters amid COVID-19 pandemic in Greece: post-disaster trends and factors affecting the COVID-19 evolution in affected areas. Saf Sci 138:105236. https://doi.org/10.1016/j.ssci. 2021.105236 7. Ouma YO, Tateishi R (2014) Urban flood vulnerability and risk mapping using integrated multi-parametric AHP and GIS: methodological overview and case study assessment. Water 6(6):1515–1545. https://doi.org/10.3390/w6061515 8. Ozkan A, Ozkan G, Yalaman A, Yildiz Y (2021) Climate risk, culture and the Covid-19 mortality: a cross-country analysis. World Dev 141:105412. https://doi.org/10.1016/j.worlddev. 2021.105412 9. PhilAtlas (2015) Balayan, Batangas Profile. https://www.philatlas.com/luzon/r04a/batangas/ balayan.html 10. Rappler Philippines (2020) Flash floods due to Typhoon Rolly sweep away several homes in Batangas. https://www.rappler.com/moveph/photos-batangas-homes-washed-awaycovered-mud-super-typhoon-rolly-flash-floods 11. Sakamoto M, Sasaki D, Ono Y, Makino Y, Kodama EN (2020) Implementation of evacuation measures during natural disasters under conditions of the novel coronavirus (COVID-19) pandemic based on a review of previous responses to complex disasters in Japan. Progress Disaster Sci 8:100127. https://doi.org/10.1016/j.pdisas.2020.100127 12. Sussman N (2020) Time for bed(s): hospital capacity and mortality from COVID-19. COVID Econ 11:116–129. https://ssrn.com/abstract=3588408 13. The World Bank (2020) Philippines: While battling the pandemic, strengthening disaster risk reduction and management needed to aid recovery. https://www.worldbank.org/en/news/pressrelease/2020/12/08/philippines-while-battling-the-pandemic-strengthening-disaster-risk-red uction-and-management-needed-to-aid-recovery 14. UCLG, UN-Habitat, UNDRR (2020) Localizing the sendai framework to achieve resilient cities and territories resilience learning module i: Fundamentals of resilient governance and development UCLG learning. Center for Urban Disaster Risk Reduction and Resilience (CUDRR). https://learning.uclg.org/localizing-the-sendai-framework-for-drr/

92

C. E. F. Monjardin et al.

15. UN (2015) Sendai framework for disaster risk reduction. UNISDR: Geneva. https://www.undrr. org/publication/sendai-framework-disaster-risk-reduction-2015-2030 16. UN-HABITAT (2018) City resilience profiling tool (CRPT) 17. UNDRR (2019) Disaster risk reduction in the Philippines: Status report 2019. Bangkok, Thailand, United Nations Office for Disaster Risk Reduction (UNDRR), Regional Office for Asia and the Pacific. https://www.undrr.org/publication/disaster-risk-reduction-philippines 18. UNISDR (2017) Works into action guidelines: National disaster risk assessment. https://www. unisdr.org/files/52828_04floodhazardandriskassessment.pdf

Sustainable Stormwater Management: Developing Stormwater Management and Drainage Master Plan for Serian, Sarawak R. Salleh, K. Y. Wong, Judy J. K. Kueh, T. Sulaiman, and A. Ainan

Abstract The Department of Irrigation and Drainage Malaysia has completed a study in 2021 to develop the Stormwater Management and Drainage Master Plan or Pelan Induk Saliran Mesra Alam (PISMA) for Serian, Sarawak. The PISMA Serian provides a comprehensive and long-term solution for the control of urban stormwater quantity and quality issues. It is expected to mitigate the flooding and water pollution problem due to urban development. In this study, extensive data have been collected at-site to analyze the root cause of the problems. Hydrologic analysis and hydraulic modelling were conducted for 2020 land use scenario to assess the existing drainage capacity. Structural and non-structural measures were designed based on the 2050 land use scenario to control the stormwater runoff and water quality. The study has also developed a GIS-based Stormwater Asset Inventory to facilitate long-term operation and maintenance. Keywords Sustainable stormwater management · Stormwater Management and Drainage Master Plan · Pelan Induk Saliran Mesra Alam · PISMA · Design of structural and non-structural measures

R. Salleh Division of River Basin Management, Department of Irrigation and Drainage Malaysia, Aras 3, Blok C7, Parcel C, Pentadbiran Kerajaan Persekutuan, 62000 Serian, Sarawak, Malaysia e-mail: [email protected] K. Y. Wong (B) · T. Sulaiman · A. Ainan Division of Stormwater Management, Department of Irrigation and Drainage Malaysia, Jalan Sultan Salahuddin, 50626 Serian, Sarawak, Malaysia e-mail: [email protected] T. Sulaiman e-mail: [email protected] A. Ainan e-mail: [email protected] J. J. K. Kueh G&P Professional (Sarawak) Sdn. Bhd., 1St Floor, Lot 1035 (Sublot 3 & 4) Riveria 3A Kuching Samarahan Expressway, 94300 Serian, Sarawak, Malaysia © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023 I. K. Othman et al. (eds.), Proceedings of the 5th International Conference on Water Resources (ICWR) – Volume 2, Lecture Notes in Civil Engineering 365, https://doi.org/10.1007/978-981-99-3577-2_7

93

94

R. Salleh et al.

1 Introduction Malaysia receives an average annual rainfall of 2,940 mm, which is about 3.5 times the average global rainfall of 830 mm. Due to the high rainfall, many parts of Malaysia are facing flooding problem especially in areas undergoing rapid urban development. In the 1970s to 1990s, rapid flood disposal method was generally adopted in Malaysia. This method was found to be not sustainable, as the flood problem would be transferred to the downstream. Planning of the flood conveyance system was becoming more difficult as urbanisation taking place. It has been recorded more than 500 flood events occurred each year for the past 5 years. In contrast, some of the areas have experienced reduction in rainfall. These reduction of rainfall at a certain period of each year has caused shortage of water in dams and causing the shortage of water supply especially in the urban area. This condition is worsened when El-Nino occurred where it could cause rainfall reduction up to 50%. Another issue deal with water is the water quality. It is found that approximately 300,000 tonnes/year of rubbish has been thrown into the river and drainage system, and this situation has caused pollution of water bodies. Several major flood events had been recorded since 1926, therefore Department of Drainage and Irrigation, Malaysia (DID) had been assigned by the Government to lead the flood mitigation to overcome the flooding problems. Flood can be categorized into three types i.e., monsoonal flood, flash flood and coastal flood, where each of this flood is referring to different type of causes and situations. Monsoonal flood is defined as the seasonal flood due to extreme rainfall for a longer duration during the Northeast Monsoon, typically starts from November until March each year. On the other hand, flash flood is due to high intensity short duration rainfall that occurs mostly at urban areas coupled with inadequate drainage and storage system. Meanwhile, coastal flood is mainly due to backwater effect from tidal influence affecting lower reaches. This study will focus on flash flood issues where it happened on all over Malaysia especially in the city center. With flash flood events that had been recorded from 2018 up to 2021 approximately between 400 to 600 times each year, it is necessary in overcoming this issue holistically. Flash flood are caused by many factors. Very high intensity short duration rainfall seems to be the main factor that cause flash flood. Besides that, rapid and high density of urban development whilst not upgrading the existing drainage system which is old, improper and under capacity will also cause flash flood. In some cases, the existing drainage system is not well maintained or improved due to budget constraint, thus will also cause backflows of water from downstream. Besides, severe erosion and sediment, rubbish together with debris are being washed off from the construction site and soon will flow into the water bodies, hence will clog the drainage system. Another main cause of flash flood is due to shortcoming of Manual Saliran Mesra Alam (MSMA) implementation, besides lack of public awareness and enforcement capacity and capability. In addition to this, lack of understandings and data for urban drainage design and implementation as well as lack of integration, interaction and cooperation in implementation of infrastructures and developments also contribute to flash flood.

Sustainable Stormwater Management: Developing Stormwater …

95

2 Water Management Strategies Before we focus on how to manage and overcome the flash flood issues, DID has brought the concept of water management strategies which includes Integrated Water Resources Management (IWRM), Integrated River Basin Management (IRBM) and Stormwater Management and Drainage Master Plan or Pelan Induk Saliran Mesra Alam (PISMA). IWRM is an approach that managing water holistically at the planning and management of water supply, wastewater, and stormwater systems. With its regional scale management, IWRM could help to protect the world’s environment, foster economic growth and sustainable agricultural development, promote democratic participation in governance, and improve human health. IRBM, on the other hand, is a basin scale management where multiple agencies working within the water boundary for optimum benefits to achieve IRBM objectives that includes ensure sufficient water, ensure clean water, reduce flood risk and enhance environmental conservation. When it comes to the city scale management, PISMA is used where city level agencies will focus on localized urban issues with long-term solution. PISMA is basically a master plan that comply with the MSMA concept and it will try to identify issues and weaknesses related to the existing drainage and stormwater systems and water quality. Proposed works to be implemented based on long term solutions to overcome flash floods and water quality issues in urban areas. As a result, it will give a guidance and reference for more systematic development in future.

3 Objectives of PISMA There are three main objectives of PISMA, which includes (a) formulate long term Stormwater Quantity Management and Master Plan to reduce flood risk by applying the stormwater quantity control; (b) formulate the long term Stormwater Quality Management and Master Plan using best available technology or best management practices as to reduce water pollution impact on waterbody, by applying the stormwater quality control; and (c) develop Stormwater Asset Inventory System (SAIS), where it will later identify optimum maintenance plan to provide continuous benefits through mapping data up to secondary drain. Up to now, DID has completed PISMA in 26 locations, 6 PISMA is in the progress and 16 are in the planning for future implementations, which brings a total of 48 PISMA will be developed by the end of 12th Malaysian Plan (RMKe-12). Overall, the selection of PISMA is based on the localized issues at the city level, which some of it are funded by the Federal Government and some are done with the initiatives of state government. It is hope that PISMA will be referred to in development planning in coming near future.

96

R. Salleh et al.

4 Case Study: Stormwater Management and Drainage Master Plan in Serian, Sarawak (Kajian Pelan Induk Saliran Mesra Alam Serian, Sarawak) 4.1 Background Serian Division is the twelve division in Sarawak since its separation from Samarahan Division on 11th April 2015. Serian Division is located at the South-Western part of Sarawak and borders with Indonesia in the south and Samarahan Division in the north. Serian Division consists of Serian district, Tebedu district and Siburan subdistrict. It has a total area of 2,405 km2 with a population of about 105,700 (2020). Serian town is the administrative centre of the division and it is located about 64 km southeast of Kuching City. The Study Area including Serian Town is located within the Serian Division, Sarawak and covers an area of about 64 km2 (see Fig. 1). The terrain elevation of the Study Area varies between 5 to 550 m above mean sea level (see Fig. 2). The Study Area is located within the Batang Sadong river basin with a total basin area of 3,533 km2 and Batang Sadong is flowing along the southeast portion of the Study Area (see Fig. 3 and 4).

Fig. 1 Location of Study Area in Serian District

Sustainable Stormwater Management: Developing Stormwater …

Fig. 2 Terrain Elevation Map of the Study Area

Fig. 3 Study Area within Batang Sadong River Basin

97

98

R. Salleh et al.

Fig. 4 River System in Study Area

4.2 Problem Statements Every year during the rainy season from December to January, the area is prone to floods because of heavy rain, the poor drainage system and backflow from Batang Sadong into the drainage system. Among the flood-prone areas located within the Study Area are Kampung Hulu Serian, Kampung Hilir Serian, Kuching – Serian Road, Kampung Munggu Limo, Kampung Slabi Empurung, Taman Pasir, and the access road leading to Resident Office and Serian Fire and Rescue Station. According to [1], Serian will be strengthened as a key contributor for food security by transitioning away from oil palm plantation and develop towards the general agriculture that produce food crops such as pineapple and durian. Therefore, it is projected to have more than 50% changes in land use from forest to urban area/ settlements and other agriculture by year 2050. Urbanisation is known to have adverse impacts on the environment as increased stormwater runoff increases the risk of flooding as well as degradation of water quality. Rainfall in urbanized areas wash down contaminants accumulated on land surfaces into stormwater facilities before being transported to receiving waters. Pollutants such as litter, oil and grease, particulates, nutrients, toxic metals, and bacteria could be found in the drainage system which leads to degradation of water quality. Most of the sources of pollution have been caused by human activities, although some of them come from natural sources. Thus, it is important to develop and prepare a Stormwater Management and Drainage Master Plan or Pelan Induk Saliran Mesra Alam (PISMA) for Serian in

Sustainable Stormwater Management: Developing Stormwater …

99

order to provide a comprehensive and long-term solution for the urban drainage to control runoff quantity, to mitigate the flooding at flood-prone areas, and to ensure good quality stormwater runoff discharging into downstream water bodies.

4.3 Stormwater Quantity Analysis and Master Plan The analysis of the study started with river hydrological analysis and the design flood hydrographs with reasonable accuracy are required as input to the river hydraulic modelling to establish the corresponding water levels to serve as the downstream boundary condition for the town drainage design for present and future land use scenarios. Based on the study of the existing river system and urban drainage and to meet the study requirement, the coverage of the river hydrological analysis includes: (i) Btg Sadong main river and its main tributaries including Sg Serian, Sg Tanggak, Sg Pasir, Sg Saroban and Sg Terusan; and (ii) Sg Rayang and its tributaries including Sg Munjin and Sg Tarat, as part of the drainage system in the Project boundary, found to be discharging into Sg Rayang, which itself is a tributary of Btg Sadong. The design flood modelling took into account both the present (at the year 2020) and future (at the year 2050) land use scenarios. The climate change factor (CCF) was applied under future (at the year 2050) land use scenario to reflect the future uncertainties in the rainfall intensity due to climate change. The main purpose of the river hydraulic modelling under this Study is to provide design flood levels for future developments and tailwater conditions for urban hydraulic modelling. For drainage hydraulic analysis, the Study Area is divided into eight (8) main drainage zones; 112 sub-catchments under present land use condition and additional 81 sub-catchments under future land use condition (see Fig. 5). Hydrological analysis was carried out to estimate the peak flow for each drainage sub-catchment based on present and future land use using the prescribed hydrological methods. Rational method and time area method were used in the estimation of design flows for drainage sub-catchments in this Study. Hydraulic analysis and modelling were carried out in order to evaluate the capacity of the present drainage network under both the present (2020) and future (2050) land use scenarios and to identify deficiencies in the present drainage system and subsequently recommend improvements. The capacity of each drain was checked to determine if the drain is sufficient or insufficient to cater for the flows under various ARIs. The drain capacity assessment was checked by comparing design capacity with the maximum flow of each drain for different ARIs. The drain capacity assessment was carried out in two scenarios: (i) The drainage system is discharged and coincided with high river tailwater level, which is usually a more critical flood condition; and (ii) The drainage system is discharged with normal river tailwater level.

100

R. Salleh et al.

Fig. 5 Drainage Catchment for the Study Area (Future (2050) land use)

Based on the hydraulic simulations, the drain capacity assessment for the alldrainage systems in the Study Area under the present (2020) land use and future (2050) land use scenario under 100-year ARI is as shown in Figs. 6, 7, 8 and 9. Under existing condition, regardless of the drainage system is discharged and coincided with high river tailwater level or the drainage system is discharged with normal river tailwater level, it was found that more than 46% of the drainage system is insufficient to accommodate flows under 100-year ARI. With the future development to take place, increase in insufficiency of the drainage system is expected as the drains will not be able to accommodate the increased flow. It is to be highlighted that the analysis is performed with assumption that the flow capacity of drains at any location does not change from 2020 to 2050. What changes are the design flood flows which increase from 2020 to 2050 corresponding to the land use change. Figure 10 and 11 show the maximum flood extent and depth within the Study Area under the various ARIs of present (2020) land use and future (2050) land use scenario respectively. These are the maximum flood extent and depth when the existing drainage system coincide with high river tailwater level. The flood extent in the Study Area of future (2050) land use scenario is considering that no improvements are proposed to increase the drain conveyance to cater for the future urbanization and the developments are taken place based on current topography. It is noted that flooded area is increased due to increase in impervious area. In order to develop a sustainable and effective urban stormwater management plan, structural and non-structural measures are proposed to reduce flooding in the Study Area. The structural measures, mainly comprise of conveyance system, source control and regional/community control. Besides, flood mitigation measures

Sustainable Stormwater Management: Developing Stormwater …

101

Fig. 6 Drain Capacity Assessment for Main Drainage in the Study Area with High Tailwater Level under Present (2020) land use Scenario (100-year ARI)

Fig. 7 Drain Capacity Assessment for Main Drainage in the Study Area with Normal Tailwater Level under Present (2020) land use Scenario (100-year ARI)

102

R. Salleh et al.

Fig. 8 Drain Capacity Assessment for Main Drainage in the Study Area with High Tailwater Level under Future (2050) land use Scenario (100-year ARI)

Fig. 9 Drain Capacity Assessment for Main Drainage in the Study Area with Normal Tailwater Level under Future (2050) land use Scenario (100-year ARI)

Sustainable Stormwater Management: Developing Stormwater …

103

Fig. 10 100-year ARI Flood Inundation Map for the Study Area under Present (2020) land use Scenario

Fig. 11 100-year ARI Flood Inundation Map for the Study Area under Future (2050) land use Scenario

104

R. Salleh et al.

Fig. 12 Overall Preferred Improvement Option in the Study Area

shall be implemented along Batang Sadong to solve river flooding. In this study, a combination of flood bund and wall providing 20-year ARI level of protection is proposed; however, this is subjected to further analysis and detailed design. The Stormwater Quantity Master Plan (see Fig. 12) for the Study Area recommends measures including provision of detention storage (online and offline), flood protection through bunding and pumping; and improvement of main drain conveyance. The flood extent in the Study Area is expected to reduce to 6.96 km2 or 10.9% of the area upon implementation of the Stormwater Quantity Master Plan, where urban area and settlements is expected to be flood-free (see Fig. 13). Non-structural measures involve mainly the management of stormwater system such as urban planning, low impact development, institutional planning, public outreach and education, flood management planning and continuous improvement planning. Non-structural measures aim to promote sustainability in stormwater management by encouraging continuous improvement and optimum planning, operation and maintenance, etc. For future development, the habitable floor levels of the building shall be set at least above the 100-year ARI flood level with 0.5 m freeboard (see Fig. 14). It is also important to ensure drainage reserves are gazetted to preserve flood plains.

Sustainable Stormwater Management: Developing Stormwater …

105

Fig. 13 100-year ARI Flood Inundation Map under Future (2050) Land use Scenario after Improvement

Fig. 14 Proposed Future Minimum Platform Level for the Study Area

106

R. Salleh et al.

4.4 Stormwater Quality Analysis and Master Plan Batang Sadong is the main river that flows through the southern part of the Study Area. Based on the water quality data obtained and monitored by Department of Environment (DOE) at Serian from 2013 to 2018, the overall water quality status for Batang Sadong at Study Area was categorised as good, where the average WQI is 85.4 and 93.5% of the samples are within the Class II range. Natural Resources and Environment Board (NREB) Sarawak also operate two stations at Batang Sadong namely Selabi Empurung (SA1) and Serian (SA2). The overall water quality of Batang Sadong monitored by both DOE and NREB depicted a similar trend. Batang Sadong achieved good water quality status in general, however, there are times where the main river is observed to be polluted from point source and nonpoint source. Point sources (PS) are discrete locations mainly from industries, sewage treatment plants (STP), workshops and restaurants which are relatively easy to be identified and controlled. An inventory of possible water-polluting sources in the Study Area was documented for the Study to provide a relation to the water quality results from both primary and secondary data. Pollution sources were identified from information collected from agencies and field investigations by the Study team. The overall pollution inventory in the Study Area is as shown in Fig. 15. In contrast, non-point sources (NPS) are in less discrete sites such as farms, forests, urban areas and agricultural areas, etc. where pollutants are washed into receiving waters by rain, contributing to the pollution loads in the rivers. NPS could be a significant source of contamination to the rivers. NPS pollutant can come from both

Fig. 15 Overall Pollution Inventory in the Study Area

Sustainable Stormwater Management: Developing Stormwater …

107

rural and urbanized areas. Urban contributions to NPS pollution may come from careless disposal of oils and other household waste into the drainage. Rural areas contribute to NPS pollution by way of poor farming practices and use of fertilizers, construction sites and other exposed areas. In order to identify and quantify the NPS pollution in the Study Area, water quality data at the main drain was acquired. Water quality samplings using Event Mean Concentration (EMC) approach were carried out for parameters of TSS, TN, TP, BOD and Oil & Grease (O&G). The EMC samplings were conducted at five locations with different land use category during the storm events. Similarly, sampling was also conducted during the normal time with no storms at these five locations. The selected land use categories are agriculture, residential, open space, industrial and kampung. In this Study, MUSIC (Model for Urban Stormwater Improvement Conceptualisation) is used to estimate stormwater pollutant loads in a catchment, predict the performance of stormwater treatment measures, develop a stormwater management strategy for a catchment and also estimate the life cycle cost for stormwater treatment systems. The water quality modelling is simulated under present (2020) land use scenario and future (2050) land use scenarios for all the drainage zones. It is noted that each different land use produces different concentrations and composition of pollutants. Urbanization is almost certain to change the present pollutant loadings. These changes should be taken into consideration during catchment planning to anticipate additional pollutant loads in future developments. The comparison of annual loads between the present (2020) and future (2050) land use scenarios are shown in Fig. 16. Comparing present and future pollutant loads of the same area could give a rough estimation of the trends of pollutant loadings and prediction of potential water quality issues. The comparison could be used for planning so as not to worsen water quality of drainage and river in the future. Based on water quality analysis, sustainable and effective urban stormwater management plan was formulated to encompass both the present and future development in the Study Area. Each sub-catchment in the Study Area possesses its own characteristics, each generates different magnitude of pollutants even though we assumed hypothetically all the catchments experience the same amount of rainfall. Consequently, stormwater quality management and planning for each catchment is different from the other. Non-structural measures can be in many forms, such as community awareness, source identification, land use planning and control, street cleaning, isolation of high pollutant source area, etc. In this Study, erosion and sediment analysis is performed to identify erosion risk area within the Study Area. The Soil Loss Map of future (2050) land use (see Fig. 17) is produced for future development references. Structural measures are techniques that aim to reduce the quantity and improve the quality of the stormwater at or near its source by using infrastructure or natural physical resources. MSMA has specified required pollution reduction targets as a reference for future development. However, optimum numbers of BMPs are proposed in this Study as shown in Fig. 18 to ensure water quality at drainage and river will not worsen in the future due to the urbanization. The BMPs are proposed at drainage

Fig. 16. Comparison of Pollutant Loads for Various Parameters for all the Drainage Zones under Present (2020) and Future (2050) Land use Scenarios

108 R. Salleh et al.

Sustainable Stormwater Management: Developing Stormwater …

109

Fig. 17 Soil Loss Map of Future (2050) Land use

sub-catchments which have exceeded the Class II limit. This approach is to ensure water quality will not worsen in the future with minimal implementation cost and future maintenance cost.

Fig. 18 Overall BMPs Proposed for the Study Area

110

R. Salleh et al.

5 Conclusion The content and materials of this paper is extracted from the Stormwater Management and Drainage Master Plan or Pelan Induk Saliran Mesra Alam (PISMA) Serian, Sarawak. The Master Plan in general identify issues like flood and water quality and weaknesses related to the existing drainage and stormwater systems, propose improvements to be implemented based on long term solutions to overcome flash floods and water quality issues in urban areas and give a guidance and reference for more systematic development for future. Acknowledgements We thank all personnel that have involved directly and indirectly in this Study for their constructive comments and suggestions. This study is fully supported by the Ministry of Environment and Water (KASA), Malaysia and Department of Irrigation and Drainage Malaysia.

References 1. Sarawak Land Use Master Plan Study (2020) 2. Department of Irrigation and Drainage Malaysia/G&P Professional (Sarawak) Sdn. Bhd (2021) Kajian Pelan Induk Saliran Mesra Alam Serian, Sarawak

Development of Low-Cost Technology for Monitoring of Soil Moisture and Recycling Rainwater for Irrigation Siti Nurhayati Mohd Ali and Nuryazmeen Farhan Haron

Abstract In a disaster-prone area, the preparation and readiness of the population at risk are important for disaster risk reduction. Some countries prepare for disaster by proposing a periodic evacuation exercise to establish community preparedness for the potential disaster threats in the area. Still, the extent of the practice dictates the level of community readiness and understanding of the disaster. There are numerous ways of planning an emergency evacuation for a disaster ranging from a simple desktop review to the detailed mathematical algorithm of evacuation components. This paper share the experience of applying aerial mapping with the Geographic Information System (GIS) overlay method to support the simulation exercise and evacuation planning for dam disaster risk reduction (DRR). Aerial mapping provide current site conditions for better site assessment. Once the flood hazard map is produced, the evacuation planning could materialise through an actual field exercise to gauge its effectiveness, complementing and improving ground observation methods. A theoretical framework of this approach is significant as guidance for evacuation planning during an emergency. Keywords Rainwater Harvesting (RWH) · Automatic Irrigation System · Arduino UNO · Climate Change · Cocoa Production

1 Introduction In 1950, cocoa was introduced to Malaysia for commercial planting [4]. As a result, the cocoa industry rose to become Malaysia’s third-largest commodity crop after oil palm and rubber [8]. However, in Sabak Bernam, Selangor, smallholder cocoa

S. N. M. Ali · N. F. Haron (B) School of Civil Engineering, College of Engineering, Universiti Teknologi MARA, 40450 Shah Alam, Selangor, Malaysia e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023 I. K. Othman et al. (eds.), Proceedings of the 5th International Conference on Water Resources (ICWR) – Volume 2, Lecture Notes in Civil Engineering 365, https://doi.org/10.1007/978-981-99-3577-2_8

111

112

S. N. M. Ali and N. F. Haron

production is less evident. It was found that cocoa productivity is lower in the smallholder sector than in the plantation sector. Recently, climate change has been identified as one of the most significant factors that affect cocoa production, especially during low annual rainfall [5]. Water security has become a risk as climate change has considerably changed the wet and dry periods, resulting in an imbalance in water availability throughout the year. During the process of planting cocoa, water requirements significantly help the growth of these plants. As a result, sufficient irrigation makes the cocoa plant flourish. However, sometimes there is a delay in irrigation or watering, particularly during the summer, which results in poorly growing plants or dead. [3] revealed that delay in irrigation and insufficient water supply to the cocoa would affect the planting of the cocoa. According to [1], cocoa needs a rainfall range between 1,015 mm and 2,538 mm and a temperature of 21 °C to 32 °C. Too much rain received for this plant will cause diseases, while too little will cause the soil to become too dry for the plant. As a result, one of the approaches that will address water shortage and dependency on the major supply of water is the Rainwater Harvesting System (RWHS) [7]. Rainwater collecting is an efficient way for smallholder cocoa producers to regulate water availability for irrigation. The conventional irrigation method uses potable water through manual control by turning the watering pumps, resulting in more water. Thus, to make it more efficient, research has been made to design an automatic irrigation system that can monitor the soil moisture of the entire plant without much effort. The automatic irrigation system can also control the lack of water for crop growth and produce a health hazard-free crop [10], which might reduce excess water used and save crops from damage [6]. Therefore, monitoring the changes in soil moisture condition is crucial to estimate the quantity of water required for cocoa plant growth. Furthermore, effective and efficient irrigation water usage in crop production is crucial to the long-term economic and environmental sustainability of irrigated farming operations. This system is a low-cost technology that includes a soil moisture sensor interfacing with Arduino. Arduino sensor is a technology that generates data changes of soil moisture. Therefore, it makes effective use of rainwater for irrigation purposes. With this, the water amount will be provided precisely when moisture in the soil goes underneath the edge [9]. Furthermore, according to [2], Arduino sensors use simple obtained components, reducing production and maintenance costs. Thus, this makes the device more cost-effective, suitable, and low- maintenance for applications, especially for smallholder cocoa farmers. It is therefore essential to implement this technology for agriculture in understanding the current agricultural issues and problems. This research aims to determine the potential of a rainwater harvesting (RWH) system for small-scale holders specifically in Sabak Bernam, Selangor. Next, to develop an automatic irrigation watering system to detect soil moisture using the Arduino UNO model.

Development of Low-Cost Technology for Monitoring of Soil Moisture … Table 1 The selected hydrological station at Sabak Bernam, Selangor

Station No.

113

Station Name

RF 3,609,012

Parit 1 at Sg. Besar

RF 3,610,001

S.R.K Parit 4 at Sg. Haji Dorani

RF 3,610,014

Pintu Kawalan P/S at Sg. Nipah

RF 3,710,006

Rumah PAM JPS Bagan Terap

RF 3,710,011

Parit 6. Sg. Besar

Fig. 1 Five selected hydrological stations in Sabak Bernam, Selangor (Source: DID website)

2 Methodology 2.1 Hydrological Stations Rainfall data from five hydrological stations are used to analyse the rainfall pattern for this study as shown in Table 1. The data was obtained from the Department of Irrigation and Drainage (DID) Selangor from the year 2019 to 2020. The number of rainfall data is required to identify the availability of water for the rainwater harvesting system. Meanwhile, Fig. 1 depicts five selected hydrological stations for this research.

2.2 System Blok Diagram The flowchart of the automatic irrigation system is shown in Fig. 2. Water collected from rainwater harvesting is used for watering the cocoa plant. When the system is run, the soil moisture sensor detects the moisture level from the soil. The measured

114

S. N. M. Ali and N. F. Haron

data is forwarded to the Arduino to be processed. Next, Arduino sends the moisture level and the status of the pump to be displayed on the LCD. At the same time, the microcontroller sends the signal to the relay module, which then runs a pump and lights up the LED. If the soil moisture level lower than 60%, the relay driver can be switched ON, so the water pump could function which is to deliver a certain amount of water to the plants. Once enough water is delivered, and the soil moisture sensor detect thats the soil moisture condition is more than 60%, the relay driver can be switched OFF so that the water pump can shut down.

2.3 Principle Operation A block diagram system of the proposed automatic irrigation system is shown in Fig. 3. It consists of a soil moisture sensor as the input that collects data from the soil to the microcontroller. This automatic irrigation system includes functional components which are a relay module, light-emitting diode (LED), and DC watering pump. The relay module is used to turn on and off the component and passes the electric signal to the water pump. When the moisture level falls to a certain value as programmed, Arduino sends a signal to the relay module to automatically turn ON/ OFF the water pump and automatically water the cocoa plant.

3 Results and Discussion 3.1 Hydrological Rainfall Results The data obtained from the Department of Irrigation and Drainage (DID) is summarised in total monthly rainfall as shown in Figs. 4 and 5. The variability of potential rainfall for rainwater harvesting in the selected area is obtained based on the maximum monthly rainfall. During the rainfall analyses, few problems occurred. Notice that there are missing data for both years 2019 and 2020 which had affected the amount of rainfall and the result obtained. In the year 2019, all the stations had the greatest number of rainfalls in November with the highest rainfall of 328.5 mm recorded at station Pintu Kawalan P/S at Sg. Nipah. Meanwhile, all stations recorded the least number of rainfalls in April except for station Pintu Kawalan P/S at Sg. Nipah recorded the least number of rainfalls in February. However, the driest month of the year is in April, notice that all stations received an amount of rainfall in the range of 12 mm to 15 mm. The same pattern is observed in the year 2020 which all of the stations having the greatest number of rainfall in November. The highest rainfall of 490.5 mm at Parit 1 at Sg. Besar. However, the driest month of the year is in March except for station at S.R.K Parit 4 at Sg. Haji Dorani recorded the least number of rainfall in

Development of Low-Cost Technology for Monitoring of Soil Moisture …

Fig. 2 Flowchart of automatic irrigation system

115

116

Soil

Power Supply

S. N. M. Ali and N. F. Haron Soil Moisture Sensor

LCD display Arduino UNO Relay Module

Fig. 3 Block diagram system

Fig. 4 Graph of total rainfall amount versus month for 2019

Fig. 5 Graph of total rainfall amount versus month for 2020

LED

Water Pump

Development of Low-Cost Technology for Monitoring of Soil Moisture …

117

December. There is a slightly different pattern is observed in October and December which the amount number of rainfalls is lower than the previous year. Overall, the results show that Sabak Bernam river basin receives a high number of rainfalls, and it will be a potential and suitable site for a rainwater harvesting system to mitigate the water shortage problem faced by the small-scale farmers planting cocoa plants during an extreme weather events. By obtaining real-time control collected rainwater harvesting to the flexible distribution of irrigation water quantity, we could meet the demand of cocoa plants irrigation water. With the availability of data rainfall, the rainwater harvesting can be successfully implemented at the selected area for agriculture especially, cocoa plant. Thus, this approach will have a greater function for the small-scale farmers to mitigate water crisis scarcity and help control climate change.

3.2 Simulation Testing For the simulation result, the automatic irrigation system has been successfully designed. The components used in the automatic irrigation system are simulated by Proteus software, where each device element is connected, and the moisture sensor produces an output voltage. The output voltage is applied to the Arduino as input which will read the value of voltage which then expresses soil moisture percentage ratio and display it on the LCD screen. After uploaded the irrigation program on the Arduino chip DC pump will operate (irrigation) or will stop (no irrigation) depending on the moisture ratio limit on the program. In this study, the selected percentage for soil moisture conditions of 20%, 40%, 60%, 80% and 100%. The result for soil moisture percentage and voltage were extracted from Proteus Software as shown in Fig. 6a–6e.

3.3 Hardware Testing Figures 7 and 9 shows the output result when soil moisture percentage (60%). As it can be concluded from the picture below, the system has been designed and tested successfully in a successful manner. Also, functionality of the system and the result were as expected and desired. As for the result, whenever a need for water is recognized by the sensor, microcontroller will send a signal to the relay to switch ON/OFF the pump to start watering the plant until enough quantity of sufficient water is delivered. When soil moisture percentage is lower than 60% green LED will light up to describe that the water is delivers to the soil. Otherwise, when soil moisture percentage is more than 60% red LED will light up to describe that the water is not deliver to the soil as the water pump is not switch ON (Fig. 8).

118

S. N. M. Ali and N. F. Haron

Fig. 6 Circuit diagram when soil moisture reading is (a) 20% with voltage 0.92 V, (b) 40% with voltage 1.72 V, (c) 60% with voltage 2.50 V, (d) 80% with voltage 3.39 V, and (e) 100% with voltage 4.20 V

OUTPUT: Water pump switch ON

Dry soil

OUTPUT: Soil Moisture percentage, voltage, and pump status

OUTPUT: Green LED light up Fig. 7 Output when soil moisture sensor (60%)

3.4 Simulation and Hardware Results Under the assumption of successful calibration of the hardware, the result can be presented and analysed to properly calibrate the linear relationship between the sensor voltage and the soil moisture percentage. The soil moisture sensor performance meets good agreement between the Proteus software and hardware result as the percentage error is less than 10% as illustrated in Table 2 and Fig. 11. From the tabulated result, it can be analysed that the soil moisture sensor detects changes in the soil. The amount of water in the soil plays important role in controlling

120

S. N. M. Ali and N. F. Haron

Table 2 The relationship between the voltmeter reading and the percentage ratio of moisture sensor reading

Soil Moisture (%)

Proteus Simulation Result

Hardware Result

Percentage Error (%)

0.93

1.09

Voltage (V) 20

0.92

40

1.72

1.74

1.16

60

2.50

2.58

3.20

80

3.39

3.35

1.18

100

4.20

4.15

1.19

the current flow through the whole system. From Fig. 11, the voltage shows an increasing trend with the soil moisture percentage. Besides, the result of percentage error showed a range between 1.09% to 3.20% lesser than 10% (Fig. 10).

Fig. 10 Output on LCD display when soil moisture reading is (a) 80% with voltage 3.35 V, (b) 100% with voltage 4.15 V

Voltage (V) vs Soil Moisture (%) 4.5 4 3.5 3 2.5 2 1.5 1 0.5 0

Proteus Simulation Result Hardware Result

0

20

40

60

Soil Moisture (%) Fig. 11 The soil moisture sensor performance

80

100

Development of Low-Cost Technology for Monitoring of Soil Moisture …

121

4 Conclusion The following conclusions can be drawn as follows: 1. The proposed systems deal with rainwater harvesting which solved the current problems related to farming such as the availability of rainwater for irrigation to reduce farmers’ effort and save a vast amount of water. 2. With this development of an automatic irrigation system, a farmer can harvest water to great extent and use it afterwards. The development of an automatic irrigation system is to sort out a solution there is a necessity of continuous monitoring of the soil water levels and to conserve water to prevent a future water crisis. 3. Based on the result from simulation using Proteus Software and hardware testing, both showed percentage error lower than 10%. Thus, considering the soil moisture sensor and the whole system could be used for monitoring soil moisture for agriculture. 4. The simulation design from this research can be used to generate accurate results for real-time monitoring from Proteus software. Acknowledgements We thank Universiti Teknologi MARA, Shah Alam, Malaysia for providing facilities to achieve our objectives in our research. We also thank the Department of Irrigation and Drainage, Malaysia for providing particular data.

References 1. Abdul AA (1990) The cocoa industry in Malaysia. Kiel Working Papers, no 449 2. Athani S, Tejeshwar CH, Patil MM, Patil P, Kulkarni R (2017) Soil moisture monitoring using IoT enabled arduino sensors with neural networks for improving soil management for farmers and predict seasonal rainfall for planning future harvest in North Karnataka—India. In: 2017 International conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud)(I-SMAC). IEEE, pp 43–48 3. Carr MKV, Lockwood G (2011) The water relations and irrigation requirements of cocoa (Theobroma cacao L.): a review. Exp Agric 47(4):653–676. https://doi.org/10.1017/S00144 79711000421 4. Chizari A, Mohamed Z, Shamsudin MN, Seng KWK (2017) The effects of climate change phenomena on cocoa production in Malaysia. Int J Environ Agric Biotechnol 2(5):238944. https://doi.org/10.22161/ijeab/2.5.42 5. Gateau-Rey L, Tanner EV, Rapidel B, Marelli JP, Royaert S (2018) Climate change could threaten cocoa production: effects of 2015–16 El Niño-related drought on cocoa agroforests in Bahia. Brazil PloS one 13(7):1–17. https://doi.org/10.1371/journal.pone.0200454 6. Kumar NK, Vigneswari D, Rogith C (2019) An effective moisture control based modern irrigation system (MIS) with Arduino Nano. In: 2019 5th international conference on advanced computing & communication systems (ICACCS). IEEE, pp. 70–72 7. Lani HM, N., Yusop, Z., & Syafiuddin, A. (2018) A review of rainwater harvesting in Malaysia: prospects and challenges. Water 10(4):506. https://doi.org/10.3390/w10040506

122

S. N. M. Ali and N. F. Haron

8. Noor WNWM, Nawi NM, Seng KWK, Buda M (2021) Supply response analysis on the impact of climate change on oil palm production in Malaysia. In: IOP conference series: earth and environmental science, vol 757, no 1. IOP Publishing, p 012009. https://doi.org/10.1088/17551315/757/1/012009 9. Roselin AR, Jawahar A (2017) Smart agro system using wireless sensor networks. In: 2017 International conference on intelligent computing and control systems (ICICCS). IEEE, pp 400–403. https://doi.org/10.1109/ICCONS.2017.8250751 10. Sukhadeve V, Roy S (2016) Advance agro farm design with smart farming, irrigation and rain water harvesting using internet of things. Int J Adv Eng Manag 1(1):33–45

Hydro-Environment

The Hydro-Environment section presents papers related to disaster mitigation, such as dam failures, river bank erosions and floods and also for clean water supply, such as determining the limit of saline water intrusion in a river, and alternative resources, such as groundwater and efficient treatment system through a water filter. These two issues were investigated experimentally, computationally and also through the analysis of a collection of data. The first issue deals with the mitigation of disasters by using an aerial mapping approach for dam disaster risk reduction, modelling of river bank erosion and bank stability and also modelling of branching channels for river diversion purposes to mitigate flood downstream. The first paper applies Geographic Information System (GIS) overlay method to support the simulation exercise and evacuation planning. The second paper did fieldwork measurements of riverbank profile, soil sample, hydraulic parameters, and vegetation properties to quantify the magnitude of erosion and bank stability. The third paper applies threedimensional modelling of flow separation, which occurs in river diversion and is influenced by several factors such as off-take angles, discharge ratios, and channel widths. The second issue deals with sustainable water resources by understanding saline and freshwater interactions, analysis of hydrogeological data for groundwater potential and applying the statistical method to agro-based filter design optimisation. Laboratory experiments are also presented to elucidate the hydrodynamic interactions between saline water and freshwater in a narrow meandering channel to determine the spatio-temporal salinity profiles along the river. The chapter concludes with two papers on data analysis from 113 boreholes of hydrogeological data such as static water level, pumping test and design, including capital costs, electricity consumption, labour force, and technical operations to ensure optimal performance of the filter for water treatment.

Utilising Aerial Mapping Approach on Dam Disaster Risk Reduction Rahsidi Sabri Muda, Izawati Tukiman, Ahmad Fadhli Mamat, Fatin Shahira Abdullah, and Mohamad Hidayat Jamal

Abstract In a disaster-prone area, the preparation and readiness of the population at risk are important for disaster risk reduction. Some countries prepare for disaster by proposing a periodic evacuation exercise to establish community preparedness for the potential disaster threats in the area. Still, the extent of the practice dictates the level of community readiness and understanding of the disaster. There are numerous ways of planning an emergency evacuation for a disaster ranging from a simple desktop review to the detailed mathematical algorithm of evacuation components. This paper share the experience of applying aerial mapping with the Geographic Information System (GIS) overlay method to support the simulation exercise and evacuation planning for dam disaster risk reduction (DRR). Aerial mapping provide current site conditions for better site assessment. Once the flood hazard map is produced, the evacuation planning could materialise through an actual field exercise to gauge its effectiveness, complementing and improving ground observation methods. A theoretical framework of this approach is significant as guidance for evacuation planning during an emergency. Keywords dam failure · disaster risk reduction · aerial mapping · flood hazard map · dam related disaster R. S. Muda (B) · A. F. Mamat · F. S. Abdullah Civil Engineering and Geoinformatics Unit, Generation and Environment Department, TNB Research Sdn Bhd, Kajang, Selangor, Malaysia e-mail: [email protected] A. F. Mamat e-mail: [email protected] I. Tukiman Kulliyyah of Architecture and Environmental Design, International Islamic University Malaysia, Gombak, Selangor, Malaysia e-mail: [email protected] M. H. Jamal School of Civil Engineering, Faculty of Engineering, Universiti Teknologi Malaysia, Johor Bahru, Johor, Malaysia e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023 I. K. Othman et al. (eds.), Proceedings of the 5th International Conference on Water Resources (ICWR) – Volume 2, Lecture Notes in Civil Engineering 365, https://doi.org/10.1007/978-981-99-3577-2_9

125

126

R. S. Muda et al.

1 Introduction Nowadays, earthquakes, floods, hurricanes, thunderstorms, and infectious diseases are the most common natural disasters caused by natural forces [32]. Meanwhile, a man-made disaster such as a nuclear explosion, civil disorder, terrorism, war, biological, chemical threat, cyber-attacks and dam failure are caused by human negligence or error involving a failure of a man-made system [11, 41]. Due to unexpected events, community preparedness toward disasters and emergency responses are needed [13, 25]. Community preparedness is an integral part of disaster management efforts in increasing the community’s knowledge to respond to a disaster. Recently, Dam Related Disaster (DRD) occurrences have become very alarming. Dam failure results in a catastrophic break followed by a flood wave at high speed often with considerable loss of life and catastrophic damage to infrastructure and environment [9]. Community preparedness for DRR is essential in strengthening disaster response strategies and helping the community understand the situations to face disaster and interact with present conditions with efficient manners [10]. The objective of this study is to utilise an aerial mapping approach to quantify the disaster impact from occurrence of dam failure. The study will assist dam owners and local agencies to manage an efficient disaster risk reduction (DRR) approach to increase preparedness and reduce the impact.

2 Literature Review 2.1 Dams Development The dam’s development is as old as early human civilisation. Ancient dams were built to prevent flooding and provide water for irrigation, while most modern dams were built for hydropower generation. The Quatinah Barrage or Lake Homs Dam, located in Syria, is the oldest operational dam in the world; this dam was constructed during the reign of the Egyptian Pharaoh Sethi between 1319-1304 BC, and was expanded during the Roman period and between 1934 and 1938 [43]. The Proserpina Dam located approximately 10km north of Merida in Spain, is the world’s second-oldest dam currently in use. Subsequently, the Romans constructed the earthen dam between the late 1st century AD and early 2nd century AD [2]. Dam is one of the most notable hydraulic structures in size and built area and built over the lake, river or water retention estuary [36]. Before 1000 AD, dams were built using local materials to store water [15]. Dams are constructed significantly for power generation, irrigation, water supply, and flood control [9]. [28] explained dams store huge volumes of water behind their main structure, also supported by auxiliary structures such as a spillway, diversion tunnel and outlet and; the dams are

Utilising Aerial Mapping Approach on Dam Disaster Risk Reduction

127

built for economic growth, and its construction involves major investment in money, natural resources and human resources. As recorded, there are over 36,000 large dams listed in the World Register of Dams [14]. The definition of large dams has changed over the years. Huge reservoirs were constructed in the water main to satisfy irrigational needs, drinking purposes, industrial and domestic use [36]. The concept of large dams has varied, whereby the International Commission on Large Dams (ICOLD) describes a large dam as one over 15 m high, measured from the bottom of the foundations to the peak of a dam, and store more than 3 Mm3 of water [14].Whereby the Texas Administrative Code (TAC) defines a large dam as a dam with a height of 100 feet or more, or a maximum capacity of 50 000 acre-ft (approximately 60 Mm3 ) or more [27]. The large dam has been classified depending on dam owners, authorities, and organisations in the particular region. The average life expectancy of a dam is 50 to 100 years. After 50 years, the maintenance costs and chances of failures start to rise dramatically [18]. There are nearly 104 reported dams in Malaysia with more than 15 m and a storage capacity of at least 3million m3 [35]. 81 out of 104 dams in Malaysia are listed as large dams, while the remaining 23 are small dams. Forty-one dams are classified as high-risk dams, and more than 12% are ageing dams in Malaysia. In Malaysia, Bakun Dam is the largest dam that own by Sarawak Energy Berhad (SEB), followed by Kenyir dam and Temenggor dam owned by Tenaga Nasional Berhad (TNB) as the second and third largest dam, respectively [21].

2.2 Disaster Risk Reduction (DRR) on Dam Related Issues Natural disasters are beyond the control of human being and difficult to be predicted when it occurs. Major natural disasters like floods, earthquake, landslides and droughts when they happen, they result in the threat of human life, loss of property, affect structure, agriculture and environment. The impacts of the disaster are different in intensities and coverage areas. Natural disasters occur every year, and recently the incidence and frequency seem to have significantly increased. Primarily due to environmental degradation, such as deforestation, intensified land use and the increasing population. Dam disaster is considered one of the major occurrences that hold extreme consequences in the environment, economy, and the worst case is loss of life [24]. According to the author, the loss of life becomes severe when the dam is high and the capacity of the water is huge. If the dam disaster occurs, people in the downstream area near the dam will have a more significant effect than others [24]. The effect of the catastrophic dam failure differs based on the extent of the flood area, the size of the population at risk and the available warning time [9]. Any regulated dam failure can have a major impact, generally range from loss of life or injury to economic loss and property and environmental damage. The effect of dam failure will result in a rapid release of water. Thus a high level of energy storage in the body of the water reservoir will undoubtedly have adverse effects on the

128

R. S. Muda et al.

downstream area. The consequence of dam break is unexpected and thus requires a greater understanding and predictive system to avoid fatality and mitigate harm and loss [28]. According to [8], dam failure is an event that could lead to loss or disruption to an organisation’s operations, services or functions. Whereby dam incident is a sudden and excessive controlled or uncontrolled release of water from an impounding structure that could be caused by damages or failures of the structure or any condition that may affect the safe operation of the dam. If not managed properly, they will lead to crisis or dam related disasters.

2.3 Potential Dam Related Disaster and Flood Hazard Besides the benefit, the dam failure occurrences would cause a disastrous impact. Dam failure would release a large amount of impounded water in the reservoir at high speed and extreme downpour downstream, resulting in loss of life and catastrophic damage to assets and the environment [17]. Dam failure could cause by overtopping, piping defects, and foundation and structural defects. Overtopping is one of the leading causes of dam failure; it was estimated that 34 per cent of earth-rock dam failures are due to overtopping [45]. Studies by Schuster et al., (1998) claimed that landslide dam failure could cause devastating outbursts that could inundate downstream areas, resulting in loss of life and property destruction [34]. Dam failure is become alarming and treating human life and substantial economic losses; therefore, effective mitigation measures and impact assessment are needed to manage the risk effectively [37]. The most current dam failure was the Bout dam in Sudan in 2020. A sudden collapse of the Bout dam has destroyed 600 houses. However, residents were successfully evacuated, and no fatality has been reported [6]. In Brazil, the tailings dam of the Brumadinho was failed in January 2019 and caused 259 deaths, and 11 people were missing [38]. One of the famous dam failures that happened in history was Teton Dam in 1976. The failure of the Teton Dam and subsequent draining of the reservoir caused the deaths of 11 people and approximately four hundred million US dollars in damages [42]. It has flooded at least six communities and tens of thousands of acres [7]. According to [42], the investigation has discovered that the proper treatment of foundation material was not fulfilled [1, 24, 35, 42] In February 1883, the [35] old Kuala Kubu town was destroyed by a great flood caused by a broken dam near the town [39]. Many civilians were drowned and lost their homes, and 33 people were killed. The town was severely flooded, and most of the facilities were damaged beyond repair [26]. According to [39], the old town has known as Ampang Pechah or Broken Dam in the Malay language. The tragedy was probably known as “Tragedi Kuala Kubu” by local people [39]. The recent incident of water release from Sultan Abu Bakar Dam Cameron Highlands in 2013 caused three people killed, nearly 100 houses destroyed [1]. The release

Utilising Aerial Mapping Approach on Dam Disaster Risk Reduction

129

of water from the dam is a part of standard operating procedure during dam emergencies by the dam owner. Therefore, there is the need to keep river reserves and corridors downstream of the dam free of human settlements and economic activities [35]. As dams continue to degrade, it is crucial to have an emergency plan to reduce loss of life and sustain the available resources. People are unable to wait until failure occurs that could threaten community safety but have to face the fact of taking notice of the conditions of those dams from the safety point of view. In a potential risk area, the preparation and readiness of the population at risk are essential for disaster risk reduction of dam failures. Some countries prepare for disaster by implementing periodic evacuation exercises to enhance community preparedness and understanding of the disaster.

3 Methodology 3.1 Aerial Mapping Approach Aerial mapping with drones, also known as Unmanned Aerial Vehicles (UAVs), is the most popular mapping technique since it is inexpensive and accurate [5]. UAVs have a lot of potential in mapping projects and can deliver very accurate data [3]. There are many different types and designs of UAVs in the market ready to take aerial mapping works. Depending on the type, UAVs can take images from a wide range of flight altitudes ranging from a minimum of 60m up to 150m [31]. In many countries, aviation regulations limit the altitude at which UAVs can operate, and visibility line of sight is necessary [16]. The same picture acquisition concept as the human flight is also available with UAV’s but without an onboard pilot throughout the flying mission. Google earth is a reliance free source of aerial images. It is a good start, but due to the low frequency of updating, the images provided are not the most accurate for the current condition. In order to capture the most updated features on-site, aerial mapping works is carried out at the area with specific identified boundary. The UAV mapping could provide an aerial map with high positional accuracy when carried out with a ground control point (GCP). GCP are points in the region of interest with established coordinates. These coordinates were derived from other credible sources or measured using classic surveying methods. For greater accuracy, ground control is necessary to calculate the aerial map’s scale, orientation, and absolute position information. It is feasible to receive geo-referenced products without GCP because the images are geo-tagged by the drone’s onboard GPS unit. However, having a large quantity of GCP is highly advised to produce reliable results [19]. GCP improves a map’s absolute accuracy by precisely positioning the model on the earth. Even for large projects, a minimum of 5 GCP is suggested, and 5 to 10 GCP is usually sufficient [29]. More GCP than the optimised number does not result in a significant

130 Table 1 Accuracy with different methods of survey [12, 20, 22, 33]

R. S. Muda et al.

Survey Method

Planimetric Height Accuracy, m Accuracy, m (X&Y (Z-axis) axis)

Image-based UAV 0.20 m with GNSS-RTK

0.3 m

SRTM

10 m

10 m

IFSAR

2.0 m

2m

LIDAR

0.05 m

0.11 m

increase in accuracy; however, in circumstances when the topography of the area is complex, more GCP will contribute to a better and more accurate reconstruction. Studies by [19] and [22] indicate that with a sufficient number of GCPs, the accuracy of the map can be improved up to the marginal error of less than 20 cm in planimetric accuracy (X & Y axis) and less than 30 cm in height accuracy (Z-axis). Other researchers have identified the positioning accuracy of image-based UAVs with Global Navigation Satellite System (GNSS)-RTK in their studies [12, 20, 22, 33]. Table 1 tabulated the accuracy of each method. Thus, the aerial mapping approach has the potential to be used in surveillance missions, planning and aerial tracking for a range of functions and applications [23], including the study of risk reduction for dam-related disaster, which is the emphasise of this paper [23].

3.2 Case Study of Aerial Mapping Analysis The case study approach is widely used because it allows in-depth, multi-faceted explorations of complex issues in real-life settings [4, 30]. The case study method enables a researcher to examine the data within a specific context closely. In most cases, a case study method selects a small geographical area or a minimal number of individuals as study subjects [44]. In this research, Sultan Abu Bakar dams (SAB) in Cameron Highland have been selected for the site study. The study is focused on communities in the surrounding areas of the SAB Dam. The dam is located at Ringlet-Lembah Bertam road with a man-made lake on Ringlet Reservoir’s upstream. The downstream communities along the Bertam River consist of a few settlements that make up the small township of Lembah Bertam and a few Orang Asli villages such as Kg. Sg. Tiang, Kg. Menson and Kg. Leryar. These settlements are particularly vulnerable to flood-prone as some of the houses, structures and farms apparently are within the river reserves and dangerously close to the dam, located within the water flow movement if the released from the dam is operated. In this study, six (6) GCPs have been utilised. Each GCP has been accurately positioned with high accuracy GNSS instruments using GNSS-RTK (Real Time

Utilising Aerial Mapping Approach on Dam Disaster Risk Reduction

131

Kinematics). UAVs has been deployed and provide high-resolution photographs in regions with limited space. An aerial survey has been conducted as the first step of the aerial mapping method to determine the field conditions. Aerial survey field conditions allow for the development of final aerial survey products within the required tolerances. Aerial surveys were conducted between the hours of 10 a.m. and 2 p.m. when deciduous trees are bare (when the sun angle is not less than 30 degrees). It was to avoid the ground is hidden by clouds, haze, fog and dust. All critical structures such as evacuation centres, emergency sirens, buildings and access roads are crucial elements that must be captured in the mapping and left out during the aerial survey. This is also to ensure that residential areas are not overlooked, particularly those not appearing on Google Earth. The next step is the designation of the area of interest for aerial mapping operations. There are various ways to prepare for this. This study utilised Google Earth, which is the most readily available resource. The area has been demarcated and exported in the format of KML (Keyhole Markup Language), which is a file format used to display geographic data in an Earth browser such as Google Earth. KML uses a tag-based structure with nested elements and attributes. The entire aerial mapping was captured using three consecutive flights of DJI Phantom 4 Pro at the height of 150 m from the departure sites. The Pix4D Mapper Pro was used to process the aerial images with accurate GCPs. The aerial images produced are then laid out in the same format KML as a new layer in Google Earth to update the current condition of the selected location. Any houses and structures that were not captured in the current view of Google Earth were placed in the new mapping. Most importantly, the higher resolution provided by the new aerial maps enables the critical features to be easily recognised and assessed for the potential risk of dam-break flood hazards. The results as shown in Fig. 1.

4 Results and Discussion The mapping analyses have generated flood boundaries, flood hazard maps and risk classification maps with the current site conditions in the SAB Dam vicinity and the Susu Dam vicinity. These maps were integrated with the topographic data, spatial data from ArcGIS and aerial images to estimate flood risk to people downstream of the dam.

4.1 Dam Break Flood Boundary Map The primary data from aerial images were used to forecast the flood boundary and the impacted area. The PMF (probable maximum flood) scenario simulated from

132

R. S. Muda et al.

Fig. 1 Aerial Image Overlaid with Google Earth at the selected location

the hydrodynamic model has been selected based on the most critical or worstcase scenario caused by a dam-break disaster [40]. People compounded in the flood boundary are considered as a population at risk (PAR). Figure 2 shows flood boundary results for the selected location in the study area. The results from flood boundary maps (Figure 1) show that a dam-break flood will inundate downstream of the Dam. Lembah Bertam and Kg. Leryar, which is located downstream of the SAB Dam will be severely affected. The estimated inundated areas were determined based on the grid cell (100 x 100 m) developed in the maps. In which dam break occurs, the total area that will be inundated at these two locations is approximately about 0.75 km2 . This area is considered a flood-prone area and highly exposed to the impact of dam-break floods from SAB Dam. The results show in Table 2. Obviously, most settlements are located along the Bertam River because the local economy is related to agricultural-based activities. Besides that, hilly topography has limited them to find a suitable location for settlement but close to a river’s flat area. This condition has put the communities in this area in danger. Based on flood boundary maps, it was estimated that 71% of the populated area at these two locations would be affected if a dam break occurred. More than 65 % of the populated area at each location in study areas would be affected. Hence, the best way to mitigate the risk and save people’s lives is to evacuate people at risk as soon as possible before the flood arrives at this location. Therefore, community preparedness and fast response are crucial to face the dam emergency assurances.

Utilising Aerial Mapping Approach on Dam Disaster Risk Reduction

Fig. 2 Flood boundary maps at the selected location

133

134

R. S. Muda et al.

Table 2 Affected areas due to dam break Location

Inundated area (km2 )

Populated area (km2 ) (A)

Inundated at populated area (km2 ) (B)

% Inundated at populated area (B/A)

Lembah Bertam

0.42

0.54

0.36

67%

Kg. Leryar

0.33

0.16

0.14

88%

Total

0.75

0.7

0.5

71%

4.2 Flood Hazard Map Flood hazard maps determined the boundary and extent of flood depth and flood velocity. The depth map and velocity map were overlaid with the processed aerial image from UAV and satellite image from Google Earth. Flood hazard severity is indicated by the various flood flow zones and depth with various colour layers. The two selected locations of mapping analyses are shown in Fig. 3. Maximum flood depth is extracted from model simulations to quantify flood impact assessment. The maximum flood depth values represent specific locations and are not generalised for the whole town and villages. The average maximum flood depth gives an overall picture of impact as a guide to establishing the flood severity risk and emergency planning. The risk classification has been applied on the flood hazard map to determine the hazard severity level at the selected locations. The maps are marked with square coloured based on their risk category. Red colour represents a high risk, green represents a medium risk, and yellow represents a low risk.

4.3 Population at Risk (PAR) The quantification of impact on people at risk was assessed using the mapping analyses technique, which gave a probability of risk profile. The Population at Risk (PAR) is referred to the local community potentially exposed and will be affected by flood disaster due to the dam break of SAB Dam. The estimation PAR has been identified at the selected location using grid system analyses (100 m x 100 m). The population in-grid was further multiplied by the proportion of land in the grid flooded to attain the PAR for the whole populated area. It found that almost 93% of people at Bertam Valley falls under the high-risk category, while the remaining 7% is identified to be in the medium-risk category. Meanwhile, at Kampung Leryar, the result shows that 100% of PAR is under the high-risk category. The summary of the results is tabulated in a pie chart in Fig. 4. There are multiple strategies to reduce and minimise the impact of dam hazards and increase the community’s readiness, which many scholars have discussed for DRR

Utilising Aerial Mapping Approach on Dam Disaster Risk Reduction

135

Fig. 3 Flood hazard maps at the selected location (Color figure online)

for dam related disasters. In this situation, the aerial mapping and dam-break flood analysis help to predict the impact and identify the risk such as flood boundary, flood depth and PAR at an affected location, and more importantly, this information is very important as guidance for dam owners and responder agencies to develop an emergency plan to avoid any possible harm to the peoples. Other than that, the mapping analysis supports the assessment of the potential hazard at such dam becomes another

136

R. S. Muda et al.

Fig. 4 Percentage of the population at risk (PAR) based on flood severity category.

indicator, particularly for the dam’s owner to alert the communities in advance when there is a potential of flood risks and other emergency calls. Thus, the mapping result helps the process of community preparedness for dam related disaster. The information would be shared directly with the community involved and also the local agencies and authorities.

4.4 Evacuation Planning and Emergency Exercises The findings of the flood mapping analysis supported by the current aerial mapping allow an initial assessment of the potential hazard at a specific location. The results have shown the impact of dam-break flood at the selected area and assessed the affected community within the vicinity of the SAB dam. This information is very important as it gives the exact boundary of the affected areas, which can help the respected agencies to plan the DRR programs accordingly. Proper Evacuation planning and emergency exercises are needed to ensure that the community at risk receives sufficient training and awareness programs as part of continuous mitigation programs. On the other hand, sufficient numbers of responders’ teams can be managed wisely for safe and rescue operations during an emergency. Besides, the facilities required in the evacuation centre could be sufficiently planned to accommodate the victim’s facilities. The risk map generated is useful as a medium of risk information shared with responder agencies and communities as a part of knowledge transfer in a DRR program. The maps could provide additional information to help responders’ agencies such as police and BOMBA (Fire and Rescue Department of Malaysia) to timely alert the communities at risk. The appropriate response time is able to reduce the likelihood of injury and death caused by dam related disaster.

Utilising Aerial Mapping Approach on Dam Disaster Risk Reduction

137

Fig. 5 Siren system (EWS) and signage system at the selected location.

4.5 Disaster Risk Reduction Initiative The dam owner and local agencies are fully aware of the importance of having good disaster risk reduction initiatives such as community emergency preparedness, standard operating procedure (SOP) and an effective warning system to safeguard the downstream community and reducing the risk of loss of life. Dam owners and local agencies had implemented emergency warnings, and they conducted drill exercise programs to prepare the community if a dam disaster happens. Dam owners have installed the siren system at the potential affected location (Fig. 5). The flood hazard map shows that the entire selected village downstream was exposed to disastrous floods under dam break event. Therefore, the drill exercise should be planned properly to make it engrossing, attracting more people to participate. Thus, it was suggested to provide various activities during the event, such as briefing on the dam safety program, talking on save and rescues procedures, social events and feast with locals. Hence, to make the drill exercise more attractive and practical, it is good to have some real situation simulation such as having injured people on-site, ambulance and rescue team on actions; thus, these scenarios will allow agencies and communities to get the idea of the actual situations. Dam owners and local agencies support the evacuation drill exercises conducted in Cameron Highlands. Figure 6 shows Drill exercises conducted in Lembah Bertam and Kg Leryah Cameron Highlands.

138

R. S. Muda et al.

Fig. 6 Evacuation drill with the affected community in Cameron Highlands

5 Conclusion The combination of the aerial mapping approach and dam-break mapping analyses provide the most recent and precise data and information on potential hazards at specific locations. It would help local authorities and emergency’s responders in early warning and evacuation plans to rescue people if the dam failure is imminent. Mapping analysis provides important information such as the potentially affected area, identified population at risk, and appropriate evacuation planning needed. This information is essential to emergency responder’s agencies in supporting an emergency operation that could reduce the loss of life and injury during disaster occurrences. Acknowledgements This study was conducted with the assistance of TNB Research Sdn. Bhd. As part of their support to Malaysia government initiative towards Sendai Framework and SDGs 2015-2030. The research is also part of academic research conducted with International Islamic University Malaysia. The authors would like to forward their acknowledgement to TNB Research Sdn. Bhd. and IIUM for this research collaboration.

References 1. Bahri S (2013) Ringlet dam disaster: Death and destruction on highlands. The Star. https:// www.thestar.com.my/news/nation/2013/10/24/death-and-destruction-on-highlands-three-kil led-in-bertam-valley-dam-disaster/. Accessed 24 Oct 2013 2. Bilewu S, Ayanshola A, Salami A (2016) A rapid health impact assessment of the university of ilorin dam. Niger J Technol 36(1):235–240. https://doi.org/10.4314/njt.v36i1.28 3. Colomina I, Molina P (2014) Unmanned aerial systems for photogrammetry and remote sensing: a review. ISPRS J Photogramm Remote Sens 92:79–97. https://doi.org/10.1016/j. isprsjprs.2014.02.013 4. Crowe S, Cresswell K, Robertson A, Huby G, Avery A, Sheikh A (2011) The case study approach. BMC Med Res Methodol 11(1):1–9 5. Everaerts J (2008) The use of unmanned aerial vehicles (uavs) for remote sensing and mapping. Int Arch Photogram Remote Sens Spat Inf Sci - ISPRS Arch 37:1187–1191 6. Fayez (2020) Dam collapse destroys 600 houses in Sudan. The Star. https://www.thestar.com. my/news/world/2020/08/01/dam-collapse-destroys-600-houses-in-sudan. Accessed 1 Aug 2020

Utilising Aerial Mapping Approach on Dam Disaster Risk Reduction

139

7. FEMA (2013) Living with dams. Know your risks. Dams and Dikes in Development, pp 25–41. https://doi.org/10.1201/b12889-5 8. Fema M (2015) Federal guidelines for dam safety risk management. Fed Emerg Manag Agency, Washington, DC, USA 9. George AC, Nair BT (2015) Dam break analysis using BOSS DAMBRK. Aquatic Procedia 4(Icwrcoe):853–860. https://doi.org/10.1016/j.aqpro.2015.02.107 10. Guo C, Sim T, Ho HC (2020) Impact of information seeking, disaster preparedness and typhoon emergency response on perceived community resilience in Hong Kong. Int J Disaster Risk Reduct 50:101744. https://doi.org/10.1016/j.ijdrr.2020.101744 11. Gupta SK, Rao ORS, Kumar S (2018) Strategy framework for risk management of man made. Disasters 2(2):1–8 12. Hashim S, Adnan NA (2016) Regional conference on science, technology and social sciences (RCSTSS 2014). Regional conference on science, technology and social sciences (RCSTSS 2014). https://doi.org/10.1007/978-981-10-0534-3 13. Hosseini SM, Bahadori M, Raadabadi M, Ravangard R (2019) Ranking hospitals based on the disasters preparedness using the TOPSIS technique in western Iran. Hosp Top 97(1):23–31. https://doi.org/10.1080/00185868.2018.1556571 14. ICOLD (2011) Constitution of ICOLD. Int Comm Large Dams (2):1–21 15. Jia J, Lino M, Jin F, Zheng C (2016) The cemented material dam: a new, environmentally friendly type of dam. Engineering 2(4):490–497. https://doi.org/10.1016/J.ENG.2016.04.003 16. Koeva M, Muneza M, Gevaert C, Gerke M, Nex F (2018) Using UAVs for map creation and updating. a case study in Rwanda. Surv Rev 50(361):312–325. https://doi.org/10.1080/003 96265.2016.1268756 17. Souza L, Chanekar TP, Pandit GS (2019) Case study and forensic investigation of failure of dam above Kedarnath. Int Soc Soil 18. Maclin E, Sicchio M (1999) Dam removal success stories: restoring rivers through selective removal of dams that dont make sense 19. Madawalagama SL, Munasinghe N, Dampegama SDPJ, Samarakoon L (2016) Low cost aerial mapping with consumer-grade drones. In: 37th Asian conference on remote sensing, ACRS 2016, vol 3, pp 2309–2316 20. Mukul M, Srivastava V, Mukul M (2015) Analysis of the accuracy of shuttle radar topography mission (SRTM) height models using international global navigation satellite system service (IGS) network. J Earth Syst Sci 124(6):1343–1357. https://doi.org/10.1007/s12040015-0597-2 21. MyCold (2019) Dams of Malaysia. Malaysian national committee on large dams (MYCOLD) 22. Nagendran SK, Tung WY, Mohamad Ismail MA (2018) Accuracy assessment on low altitude UAV-borne photogrammetry outputs influenced by ground control point at different altitude. In: IOP conference series: earth and environmental science, vol 169. https://doi.org/10.1088/ 1755-1315/169/1/012031 23. Nex F, Remondino F (2014) UAV for 3D mapping applications: a review. Appl Geomat 6:1–15. https://doi.org/10.1007/s12518-013-0120-x 24. Norkhairi FF, Thiruchelvam S, Hasini H (2018) Review methods for estimating loss of life from floods due to dam failure. Int J Eng Technol (UAE) 7:93–97. https://doi.org/10.14419/ ijet.v7i4.35.22334. 25. Ortiz-Barrios M, Gul M, López-Meza P, Yucesan M, Navarro-Jiménez E (2020) Evaluation of hospital disaster preparedness by a multi-criteria decision making approach: the case of Turkish hospitals. Int J Disast Risk Reduct 49:101748. https://doi.org/10.1016/j.ijdrr.2020.101748 26. PeKhabar (2015) Empangan Kuala Kubu Pecah. https://pekhabar.com/h-i-d-s-empangankuala-kubu-pecah/. Accessed 29 Oct 2015 27. Program DS, Division CI (2012) Guidelines for developing emergency action plans for dams in Texas 28. Rahsidi Sabri M, Khidzir ABM, Jamal MH, Samsudin SH, Razad AZA (2018) In flood risk assessment for dam break using GIS application: a case study in Bertam Valley, Cameron highlands. In: 4th international conference on water resources (ICWR 2018). Springer

140

R. S. Muda et al.

29. Ridolfi E, Manciola P (2018) Water level measurements from drones: a pilot case study at a dam site. Water (Switzerland) 10(3). https://doi.org/10.3390/w10030297 30. Rowley J (2002) Using case studies in research. Manag Res News 25(1):16–27 31. Sadeghi S, Sohrabi H (2019) The effect of UAV flight altitude on the accuracy of individual tree height extraction in a broad-leaved forest. In: GeoSpatial conference 2019 – joint conferences of SMPR and GI Research, pp 1168–1173. https://doi.org/10.5194/isprs-archives-xlii-4-w181168-2019 32. Said NB, Chiang VC (2020) The knowledge, skill competencies, and psychological preparedness of nurses for disasters: a systematic review. Int Emerg Nurs 48:100806. https://doi.org/ 10.1016/j.ienj.2019.100806 33. Salach A, Bakuła K, Pilarska M, Ostrowski W, Górski K, Kurczynski Z (2018) Accuracy assessment of point clouds from LiDAR and dense image matching acquired using the UAV platform for DTM creation. Can Hist Rev 7(9). https://doi.org/10.3390/ijgi7090342 34. Schuster RL, Wieczorek GF, Hope DG, Keefer DK (1998) Landslide dams in Santa Cruz County, California, resulting from the earthquake. US Geol Surv Prof Pap 1551:51-70 35. Shahabudin SM (2015) Exploring the viability of dams is key to malaysian water resources development of the future. J Inst Eng, Malaysia 76(1) 36. Shinde SE, Pathan TS, Raut KS, More PR, Sonawane DL (2010) Seasonal variations PhysicoChemical Characteristics of Harsool - Savangi Dam, District Aurangabad, India. The Ecoscan 4(1):37–44 37. Shrestha BB, Kawasaki A (2020) Quantitative assessment of flood risk with evaluation of the effectiveness of dam operation for flood control: a case of the Bago River Basin of Myanmar. Int J Disaster Risk Reduct 50:101707. https://doi.org/10.1016/j.ijdrr.2020.101707 38. Rotta LHS, Alcantara E, Park E, Negri RG, Lin YN, Bernardo N, Souza Filho R (2020) The 2019 Brumadinho tailings dam collapse: Possible cause and impacts of the worst human and environmental disaster in Brazil. Int J Appl Earth Obs Geoinf 90:102119. https://doi.org/10. 1016/j.jag.2020.102119 39. Tam S (2012) Kuala Kubu Dam Tragedy 1883. The Star. https://www.thestar.com.my/travel/ malaysia/2012/06/29/kuala-kubu-baru. Accessed 29 June (2012) 40. TNBR (2017) Study on the potential impact Sultan Abu Bakar dam operational release into Susu Reservoir. Project report, Kuala Lumpur. 41. UNISDR(e) (2018) Man-made and Technological Hazards; Words into Action Guidelines. Geneva - Switzerland. https://www.unisdr.org/files/54012_manmadetechhazards.pdf 42. Wayne JGP (2008) The Teton dam failure- an effective warning and evacuation. In: Association of state dam safety officials 25th anniversary conference, Indian Wells, California 43. Yannopoulos PC, Demetracopoulos AC (2018) A novel methodology for multiple-year regulation of reservoir active storage capacity. Water (Switzerland) 10(9). https://doi.org/10.3390/ w10091254 44. Yin RK (2014) Case study research : design and methods/Robert K. Yin. Appl Soc Res Methods Ser 5 45. Zhong Q, Chen S, Deng Z (2018) A simplified physically-based model for core dam overtopping breach. Eng Fail Anal 90:141–155. https://doi.org/10.1016/j.engfailanal.2018.03.032

Short Timescale Riverbank Erosion and Bank Stability of Sg. Bernam Using Bank Stability and Toe Erosion Model (BSTEM) Azlinda Saadon, Zulkiflee Ibrahim, and Mohamed Fuad Said Khamis

Abstract Riverbank erosion is significantly affected by the integration between hydraulic action of flowing water and geotechnical stability of the bank. This soil– water interaction is a complex process, requires in-depth understanding and integration between both factors that lead to fluvial erosion and geotechnical failures. This study was carried out to evaluate the short timescale riverbank stability using the Bank Stability and Toe Erosion Model (BSTEM) by integrating both factors. The model was applied to the selected reach susceptible to bank erosion along Sg. Bernam, between state of Selangor and Perak. Fieldwork measurements were conducted for two consecutive months at the selected sampling point, namely riverbank profile, soil sample, hydraulic parameters, and vegetation properties. Various scenarios applied to the model in quantifying the magnitude of erosion and the bank stability. The first scenario was on the effect on vegetative cover to the eroded bank, second scenario involved the effect of water table with the eroded bank under dry condition, partially saturated, and fully saturated comprising the effect of tension cracks. The bank stability analysis quantifies the Factor of Safety (FoS) of the bank based on three modes of stability, which is stable, conditionally stable, and unstable. Results showed that the FoS of both right and left banks improved with vegetative cover for the scenario without and with tension crack. On average, the right bank indicates unstable compared to left bank, with FoS ranging from 0.25 to 0.72, with lateral retreat between 28 to 77 cm. Although the left bank shows more stable in terms of the FoS, the erosion length ranging between 28 to 72 cm. Results from A. Saadon (B) School of Civil Engineering, College of Engineering, Universiti Teknologi MARA, 40450 Shah Alam, Selangor, Malaysia e-mail: [email protected] Z. Ibrahim School of Civil Engineering, Faculty of Engineering, Universiti Teknologi Malaysia, 81310 Johor Bahru, Johor, Malaysia e-mail: [email protected] M. F. S. Khamis Faculty of Engineering, Built Environment, and Information Technology, SEGi University, Kota Damansara PJU 5, 47810 Petaling Jaya, Selangor Darul Ehsan, Malaysia © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023 I. K. Othman et al. (eds.), Proceedings of the 5th International Conference on Water Resources (ICWR) – Volume 2, Lecture Notes in Civil Engineering 365, https://doi.org/10.1007/978-981-99-3577-2_10

141

142

A. Saadon et al.

various scenarios indicated that BSTEM model successfully analyzed and quantified the short timescale riverbank toe erosion and the factor of safety of the eroded banks. Keywords Riverbank erosion · bank stability and toe erosion model · BSTEM · Sg. Bernam

1 Introduction Riverbank erosion is a complex natural process; it involves water and soil interactions in greater magnitude. Human activities within the catchment area may contribute to the riverbank erosion rate that will promote bank scouring and infrastructure collapse adjacent to the riverbank. The catastrophe impacts due to the scouring of the riverbank can lead to infrastructure damages and it can be avoided if the severity of the erosion is predicted and assessed. Consequently, research on this topic is highly important to plan and find suitable mitigation methods in minimizing the negative impacts in river systems and catchments that are highly dynamic and constantly changing. Study on riverbank erosion and lateral changes has been emerged with substantial progress over the last 40 years. Among the latest and significant work related to riverbank erosion studies include [21, 25, 30] emphasized on the prediction, technique and measurement of riverbank erosion, evolution of channel changes [9], meander or channel migration [7, 19, 20], and riverbank stability [13, 15]. Studies show that up to 80% of total sediment in river watersheds was due to riverbank erosion [1, 6, 28, 29]. These efforts commonly focus on the use of empirical models based on factors such as measured erosion rates, soil types, flow velocity and water level. However, prediction works using existing equations and methods can be difficult especially dealing with different soil properties and river characteristics. The riverbank hydraulic characteristics and soil properties often influence the selection of best predictor. The developed equations and methods cannot be simply applied to any stream due to some uncertainties and limitations. Most existing equations and methods were developed using data obtained from the selected river resulting unfavourable solutions toward the local river data in Malaysia. Studies of riverbank erosion are limited in Malaysia due to its complexity in riverbank erosion measurement, soil properties and high flow condition. In Malaysia, the data pertaining riverbank erosion are limited to be found. The riverbank erosion process caused the river profile to change, leading to a significant migration of the profile and stream planform. Further consequences relate to the possible collapse of adjacent infrastructures such as bridges, roads, buildings and could lead to loss of land. Previous studies focused on studies of erosion involving cohesive and non-cohesive materials with homogenous soil. Studies on erosion profiles on stratified soils are however limited and further investigation is required. Furthermore, the use of empirical approaches in quantifying erosion rates did not include the riverbank stability prediction [26, 27]. This prediction leads to the Factor of Safety (FoS), an advantageous analysis in the planning and managing

Short Timescale Riverbank Erosion and Bank Stability of Sg. Bernam …

143

the potential erosion in a long run. One of the prominent case studies highlighted is the prolong riverbank erosion that were subjected to both fluvial and mass failure at Sg. Bernam, Tanjung Malim, Selangor. The eroded riverbanks along the outer bank have increased significantly in the past 6 years, affecting the river profile in terms of bank collapsed and loss of land area. The prediction of bank stability can assist the authority and community within the river in proposing appropriate solutions in mitigating the impact of erosion. Knowing the importance of the quantification and prediction of the erosion rate, this study aims to: (1) to quantify riverbank erosion data inclusive of erosion parameters, such as flow resistance, bank geometry and bank properties by fieldwork investigation for natural river channels; (2) to evaluate the effect of various scenarios on riverbank stability, in terms of Factor of Safety (FoS) of the bank, with the effect of tension crack and without tension crack, incorporating of dry condition, partially saturated condition and fully saturated condition (3) to quantify the amount of lateral retreat of the bank and total eroded materials using Bank Stability and Toe Erosion Model (BSTEM).

2 Short-Term Temporal Scale Riverbank Erosion Process-based models aim at simulating physical processes liable for bank retreat. Simpler process-based approaches generally only consider one among the main driving processes resulting in streambank retreat, typically fluvial erosion processes. Bank erosion models of this sort, and especially most meander migration models, commonly utilize a bank erodibility coefficient calibrated to historical retreat rates during a stream usually multiplied by an element associated with streamflow, power, or shear stress [7, 19]. All those studies incorporated geotechnical failure processes and highlighted the importance of near-bank pore-water pressures on streambank migration resulting in the present versions of more advanced process-based models used today [23, 24]. Latest models include numerous versions of one-dimensional or two-dimensional models for hydrodynamics and/or river meander migration with physically based bank evolution [10, 18, 22, 23]. A stability model referred to Bank Stability and Toe Erosion Model (BSTEM) was developed by National Sedimentation Laboratory in Oxford [28]. BSTEM is one of the most used and most advanced erosion process-based model especially in assessing a short-term temporal scale riverbank erosion. Recent research still improved and modified BSTEM. A dynamic version 5.4 of BSTEM is the most current version and has ability to predict BSTEMs with many new features. The model can model a continuous hydrograph by chronologically applying the various model components for a stream depth defined by a hydrograph, redrawing the bank profile, and then moving to the next step of the hydrograph [8]. [16] performed eight BSTEM simulations of the Barren Fork Creek streambank, Oklahoma [16]. [16] found critical issues for improving BSTEM including distributions of the shear stress, pore-water pressure dynamics of riverbank, and methods for predicting non-cohesive erodibility parameters [16]. [8] incorporated root cohesion on top and bottom of bank for ten

144

A. Saadon et al.

composite riverbanks along the Barren Fork Creek in eastern Oklahoma. [8] calibrated the BSTEM to match the observed areal imagery retreat for these composite streambanks. [8] found that soil erodibility parameter was significant predictor of bank retreat. [2–4, 14] found that erodibility parameter decreased as water content increased until reached optimum soil moisture level and maximum dry density. [8] proved that BSTEM is an advantageous estimating tool for predicting bank retreat rates compared to in situ bank retreat measurements. [30] investigated major technical issues in BSTEM that need to be addressed. Spatial and temporal variability in geotechnical failure and fluvial erodibility parameters, accounting for the effect of riparian vegetation and roots on bank shear stress distributions, accounting erodibility soil parameters, and incorporating sub-aerial and other site-specific processes into the model were some of these issues. Other studies estimated bank stability and erosion using other erosion models [5, 11, 12]. [17] modified and compared the soil and water assessment tool (SWAT) on composite riverbanks to investigate the influence of channel parameters on prediction of streambank erodibility. [11, 12] simulated CONCEPTS model on five-mile creek in Western Oklahoma. A framework was developed to evaluate streambank stability, landowner partialities, building costs, and effectiveness. [5] predicted river erosion of Chenab River, Pakistan, using Landsat images. Good agreements were obtained of river erosion between Landsat images and excess shear stress approach. Limited studies were investigated with regards to the riverbank characteristics of the Sg. Bernam, Tanjung Malim. It is also important to predict the amount of sediment that obtained at upstream due to land clearing within the catchment area. Fieldwork measurements were conducted by [25, 26], however the studies focused on the development of empirical model in quantifying the erosion rates. Statistical approach and machine learning approach have been utilized in the model development. Although the developed empirical model successfully achieved model accuracy of more than 90%, most empirical approaches are unable to accurately imitate the natural behaviour of the riverbank erosion, which requires an interaction between several parameters, such as hydraulics, channel geometry and bank properties. The fieldwork observations indicated that severe bank erosion due to undercutting and cantilever failure occurred on the left bank (outer bank) of Sg. Bernam. All previous studies did not count the BSTEM prediction of riverbank stability at different bank moisture content. The aim of this research is to compute the riverbank stability and the magnitude of toe erosion along Sg. Bernam reach using BSTEM at three different scenarios. The first scenario was accomplished by assuming that the layers of bank were varied by three different moisture contents (dry condition, optimum condition, and wet condition). This is to mimic the normal field conditions where the upper layers were usually dried, while the lower layers were either optimum or wet conditions. The second scenario was assumed that the whole bank was varied by three different moisture contents (dry bank, optimum bank, and wet bank). This was to mimic the field condition for dry bank when water level is very low, optimum bank moisture content when water level is high (flood situation), and wet bank when flow hydrograph was at recession limb. The third scenario is to evaluate the riverbank stability using various vegetative cover and non-vegetation scenario. The proposed

Short Timescale Riverbank Erosion and Bank Stability of Sg. Bernam …

145

simulation will quantify the stability of the riverbank based on three states (stable, conditionally stable, and not stable), and the magnitude of toe erosion quantified as total eroded area.

3 Method This research focuses on the quantification of natural channels stability using Bank Stability and Toe Erosion Model (BSTEM). The overall methodology consists of three phases. Phase 1 will focus on fieldwork investigation that includes the evaluation of erosion susceptibility at the selected reach, measurement of the magnitude of erosion (in unit length over time), flow resistance or hydraulic characteristics of the channel, bank geometry and bank properties. Phase 2 focuses on the evaluation of various scenarios on riverbank stability using BSTEM that includes the effect of various scenarios on riverbank stability, such as the effect of moisture content on the riverbank and riverbed and the effect of vegetation on the channel bank face. Phase 3 will focus on the model evaluation in terms of riverbank stability based on Factor of Safety (FoS) and toe erosion using Bank Stability and Toe Erosion Model (BSTEM). This includes model validation using comparative study between the predicted erosion and measured erosion. The detail methodology of this research described in the following subsections.

3.1 Selection of Study Area This study focuses on quantifying riverbank stability and toe erosion for natural channel using BSTEM. For this purpose, the proposed site located at Sg Bernam, Tanjung Malim. This river reach is located between state of Selangor and Perak, flows 216 km in length from Mount Liang Timur (Mount Liang East) in the east on the Titiwangsa Mountains to the Straits of Malacca in the west. Two locations have been selected to represent the upstream of Sg. Bernam at the latitude of 3o40’51.31”N and longitude of 101o31’44.40”E (Point 1), and downstream of Sg. Bernam at the latitude of 3o40’45.48”N and longitude of 101o31’22.07”E (Point 2). The upstream location consist of a bend and focus will be given at the outer bend and the downstream location represent a straight reach, both highlighted in Fig. 1.

3.2 Data Fieldwork investigation and data collection for this study was conducted for two consecutive months, consist of erosion length, bank geometry, vegetation properties, bank properties and hydraulic characteristics of the bank. Physical observation shows

146

A. Saadon et al.

Fig. 1 Location of selected study area along Sg. Bernam, Tanjung Malim

that the bank erosion at the left bank was more severe than the right bank. This is mainly because left bank comprises of the outer bank of the channel bend, where most of the ban particles being eroded, especially during hight flow or rain event. The right bank comprises of the inner bank, where most of the eroded particles deposited and formed aggregation in the form of sand dunes on the channel bed. The profile along with Sg. Bernam is homogeneous in the category of coarse sand. Evidence of sand, clay and silt was observed along both the left and right side of the river. Right bank evidenced bed with shallow ripples and some point bar. The flow in dry season can be relatively low and some portions are exposed to weathering and fluvial entrainment. Potential of basal cleanout is observed along the transcend vertically and horizontally. The left bank evidenced major parts of the bank are exposed to weathering fluvial entrainment. Slab or block failures are the most typical observed within the transect. Bank material is silty with varying clay with no vegetation and fallen blocks of silty material are visible at bank foot. Figure 2 and Fig. 3 show the physical observation of right bank and left bank along the study area. Table 1 shows the fieldwork data categories used in this study, variables, and unit.

3.3 Application of BSTEM in Quantifying Riverbank Erosion and Bank Stability BSTEM is a spreadsheet tool used to stimulate stream bank erosion in a mechanistic framework. It has been successfully used in a range of alluvial environments in both statistic modes to stimulate bank stability conditions and design of streambank static modes to stimulate bank stability conditions and design of streambank stabilization

Short Timescale Riverbank Erosion and Bank Stability of Sg. Bernam …

147

Fig. 2 Physical observation on the right bank of Sg. Bernam

Fig. 3 Physical observation on the left bank of Sg. Bernam

Table 1 Fieldwork data categories used in this study

No Data Categories

Variables

Unit

Length of erosion

mm

1

Riverbank erosion

2

Hydraulic characteristics Water depth Near-bank velocity Boundary shear stress

3

Bank geometry and bank Height of the bank m o profile Bank angle Bankfull width m Bank profile (elevation) m

4

Soil properties

Average particle size

mm

5

Vegetation properties

Type of vegetation Coverage percentage Root depth

% mm

m m/s N/m2

148

A. Saadon et al.

measures and iteratively over a series of hydrographs to evaluate surficial hydraulic erosion, bank failure frequency, and the volume of the sediments eroded from a bank. The model is useful in testing the effect of potential mitigation measures that might be used to reduce the frequency of bank stability and decrease sediment loadings from stream banks. To use the model, the user needs to begin with Input and proceeds through the Bank Material and Bank Vegetation and Protection sheets. The order you use the components is user selectable. However, if you choose to use the Toe erosion component you will be routed to Toe Model output to calculate the amount of bank toe erosion. If the user decides to use the Bank Stability component it will be routed to Bank Model Output to calculate Factor of Safety (FOS). The calculated bank failure profile may be viewed in Bank Model Output. If the user inserts tension cracks the user needs to return to the Bank Geometry macro on Input Geometry. Results can be transferred back into the model for further iterations using the Export New Profile into Model. The Factor of Safety depends on Bank Material, water table (dry condition, partially saturated or fully saturated) and the type of vegetation used. If FoS is lower than 1.0 the results will be shown as unstable. The results displayed will be conditionally stable when the FoS is between 1.0 and 1.3, while the FoS is greater than 1.3, the results displayed will be stable. Twelve BSTEM models were performed to quantify the magnitude of erosion and evaluate the bank stability and hydraulic erosion for critical banks of cross-section profile of the left and right banks along Sg. Bernam for three different scenarios. The first scenario was on the effect on vegetative cover to the eroded bank. This was to mimic the normal field conditions and the effect it has with and without vegetation cover. The second scenario involved the effect of water table where the eroded bank under dry condition, partially saturated, and fully saturated comprising the effect of tension cracks. The bank stability analysis quantifies the Factor of Safety (FoS) of the bank based on three modes of stability, which is stable, conditionally stable, and unstable. Results showed that the FoS of both right and left banks improved with vegetative cover for the scenario without and with tension crack. In this study, the input geometry represents the critical banks of cross-section profiles of the left and right banks. The bank profiles are plots of the riverbank profiles based on the surveyed data using the auto level. The elevations of the respective banks were surveyed at 1-m interval. The y-axis represents the bank elevation in-unit meter and the x-axis represent the bank distance in-unit meter. Figure 4 shows the original and eroded banks for the right and Fig. 5 shows the original and eroded banks for left bank of Sg. Bernam. The bank geometry coordinate must follow the bank profile from top left to bottom right. Each point must be unique, points that lie beyond the shear surface base are ignored by the simulation. The BSTEM model was simulated for two cases. For case 1, the banks were assigned with no tension crack (H = 0 m), with and without vegetation, incorporating additional three sub-conditions to account for the eeffect of groundwater level, namely, dry condition, partially saturated condition, and fuller saturated condition. Similar effects were applied to model case 2, were the tension crack effect (H = 0.84 m) was added to the banks. The models were simulated with and without vegetation, incorporating additional three sub-conditions to account for

Short Timescale Riverbank Erosion and Bank Stability of Sg. Bernam …

149

Fig. 4 Original bank and eroded bank for the right bank of Sg. Bernam

Fig. 5 Original bank and eroded bank for the left bank of Sg. Bernam

the eeffect of groundwater level, namely, dry condition, partially saturated condition, and fuller saturated condition. The bank is divided through entering bank layer thickness, up to five stratigraphic layers can be defined. Layers below the shear surface are ignored in the bank stability stimulation. The nature of the material for each layer will be assigned in Bank Material and Bank Vegetation and Protection. Even if the bank material is homogeneous it is worth using several layers with identical soil properties because the bank’s stability component calculates pore-water pressure and unsaturated soil weight as an average for the midpoint of each layer and slice. The layers may be viewed in Toe Model

150

A. Saadon et al.

Output and Bank Protection Output. The root reinforcement model requires the user to first select either a species from a drop-down box which then activates root tensile strength-diameter curves, or the user may enter their roots tensile strength-diameter relation. The user must then decide whether to use growth curves that use the age of the plant to predict the total number of roots combined with independently-derived woody vegetation and grass-root diameter histograms to enter their root-diameter data. Finally, the user needs to enter the percentage of the study react that is composed of the selected species. The nature of the material, in terms of erodibility, must be assigned for each stratigraphic layer in the bank and the bank toe material. Note that in this version of the model the bed is fixed. Should the critical shear stress and the erodibility coefficient of a given material be known, these may be entered into appropriate boxes. Note that in this case, enter own data must be selected for a given layer from the drop-down list in Bank Material. If only the non-cohesive particle diameter is known the critical shear stress may be estimated in Bank material. Alternatively, should no data be available values may be used by selecting a material type from the drop-down box Protection may be applied to the bank and the bank toe material type from the drop boxes in Bank Vegetation and Protection.

4 Results and Discussion The results were obtained after running the right Bank Model Output in the BSTEM software which was characterized under three conditions which are dry condition, partially saturated and fully saturated. The percentage of the vegetation used was (Without vegetation 40% + Perennial Ryegrass 60%) whereas the age of the plant adopted as ten years, producing a cohesion due to roots of 0.9 kPa, while the proposed bank protection on the bank toe model was (Rip Rap D50 0.256 m). The input duration used was twenty-four hours. The following results are presented in comparison between the case Without Tension Crack (With and Without Vegetation), incorporating Dry, Partially Saturated and Fully Saturated for both right bank left banks, and the case With Tension Crack (With and Without Vegetation), incorporating Dry, Partially Saturated and Fully Saturated for both right bank left banks.

4.1 Stability Evaluation (FoS) for the Case of With and Without Tension Crack (With Vegetation and Without Vegetation) The BSTEM simulation results represented the comparison between with and without the presence of tension crack on the banks. The tension crack denoted as H, refers to the depth of tension crack. For the case of without tension crack, H denotes as

Short Timescale Riverbank Erosion and Bank Stability of Sg. Bernam …

151

0 m, and with tension crack, a depth of 0.84 m is used in the simulation. Based on BSTEM software, the total height of the tension crack is assumed to be 1.68m, therefore the FoS calculation in the software using the tension crack should be half the total height of the tension crack. Table 2 shows the stability evaluation in terms of Factor of Safety (FoS) for the case without tension crack (with vegetation and without vegetation), incorporating dry, partially saturated, and fully saturated for both right bank left banks for two consecutive months (month 1 and 2). Table 3 shows the stability evaluation in terms of Factor of Safety (FoS) for the case with tension crack (H = 0.84 m), with vegetation and without vegetation), incorporating dry, partially saturated, and fully saturated for both right bank left banks for two consecutive months (month 1 and 2). Table 2 BSTEM model results for Case 1: Without Tension Crack (With Vegetation and Without Vegetation) Case 1: Without tension crack (H = 0m) Bank

Bank Condition Dry

Right bank

Left bank

Partially saturated Fully saturated Dry Partially saturated Fully saturated

Month 1

Month 2

Without Vegetation

With Vegetation

Without Vegetation

With Vegetation

FOS

FOS

FOS

Stability

FOS

Stability

1.24

Conditionall Stable

Stability

Stability

2.59

Stable

2.61

Stable

1.24

Conditionally Stable

1.82

Stable

1.85

Stable

0.71

Unstable

0.72

Unstable

0.28

Unstable

0.29

Unstable

5.33

Stable

5.37

Stable

3.89

Conditionally Stable Stable

2.50 1.72

1.01

3.92

Conditionally Stable Stable

Stable

2.53

Stable

3.19

Stable

3.23

Stable

Stable

1.75

Stable

2.19

Stable

2.23

Stable

1.04

Table 3 BSTEM model results for Case 2: With Tension Crack (With Vegetation and Without Vegetation) Case 2: With tension crack (H = 0.84m) Bank

Bank Condition Dry

Right bank

Left bank

Partially saturated Fully saturated Dry Partially saturated Fully saturated

Month 1

Month 2

Without Vegetation

With Vegetation

Without Vegetation

With Vegetation

FOS

FOS

FOS

Stability

FOS

Stability

1.17

Conditionally Stable

Stability

Stability

2.31

Stable

2.61

Stable

1.16

Conditionally Stable

1.67

Stable

1.69

Stable

0.65

Unstable

0.65

Unstable

0.25

Unstable

0.25

Unstable

3.15

Stable

3.17

Stable

3.13

Conditionally 1.17 Stable Stable 3.15

Conditionally Stable Stable

2.06

Stable

2.08

Stable

2.09

Stable

2.11

Stable

1.72

Stable

1.75

Stable

2.19

Stable

2.23

Stable

1.15

152

A. Saadon et al.

For Case 1, the BSTEM model was simulated against the case without tension crack (H = 0 m), with vegetation and without vegetation. The model was simulated under dry, partially saturated, and fully saturated condition. The FoS for the right bank yielded high value (FoS of 2.59 for the right bank in month 1 and 3.89 for the left bank in month 1) when the tension crack was not incorporated in the simulation. The FoS for both banks under Dry condition indicated stable without tension crack. The FoS for the second month for right bank slightly dropped to 1.24, indicating conditionally stable, while FoS of the left bank indicated as 5.33 (stable). When water table is induced as partially saturated, the FoS reduced to 1.82 for right bank (month 1) and 2.5 for left bank (month 1). For the second month, Fos for the right bank dropped further to 0.71 (unstable) and 3.19 (stable) for the left bank. When water table is induced as fully saturated, the FoS for the right bank in month 1 decreased to 1.01 (conditionally stable) and 1.72 for the left bank (stable). Result for the right bank in month 2 indicated that the bank is unstable with FoS of 0.28, however, the left bank remains stable with FoS of 2.19. For the simulated BSTEM model Case 1 with vegetation under the dry condition, the stability increases from 2.59 to 2.61 (stable) for the right bank (month 1). The same trend can be seen for left bank with an increase in the FoS from 3.89 to 3.92 (month 1). Results for month 2 indicated no change on FoS for the right bank, however, the FoS for the left bank in month 2 increased to 5.37. Under partially saturated condition, similar trends are observed for both right and left banks. The FoS increased with vegetation. Fully saturated condition indicated similar trends, where all the FoS increased for both left and right banks for month 1 and month 2. Hence, it can be concluded that the presence of vegetation cover increases the stability of the bank, for multiple water table condition (dry condition, partially saturated, and fully saturated). For Case 2, the BSTEM model was simulated against the case with tension crack (H = 0.84 m), with vegetation and without vegetation. The model was simulated under dry, partially saturated, and fully saturated condition. As predicted, the presence of the tension crack on the banks reduces the stability of the bank. The FoS for the right bank without vegetation under dry condition reduced to 2.31 (month 1) and 1.16 (month 2), in comparison to the condition in Case 1 (without tension crack). Similar trends are observed in FoS values for month 2 for both left and right banks. For partially saturated condition, the FoS value for right bank (month 2) decreased to 1.67 from 1.82 for the case without tension crack. The same trends observed in fully saturated condition for the right banks (month 1 and month 2), the value of FoS decreased with the presence of tension crack. FoS for partially saturated condition reduced to 0.65 (unstable) for month 2. The simulated BSTEM model Case 2 incorporating tension crack (H = 0.84 m), with vegetation under the dry condition, the stability decreases from 2.61 (without tension crack) to 1.16 (conditionally stable) for the right bank (month 1). The same trend can be seen for left bank with a decreased in the FoS from 3.92 to 3.15 (month 1). The same trend yielded for month 2 for both right bank and left bank. Under partially saturated condition, it can be observed that the stability of the right bank indicated stable FoS of 1.69 for month 1, however, the bank is not stable for month 2

Short Timescale Riverbank Erosion and Bank Stability of Sg. Bernam …

153

(FoS of 0.65). Although the left bank under partially saturated condition is in stable, the value of FoS decreased. The right bank under fully saturated condition yielded unstable FoS of 0.25 respectively for month 2, with the presence of tension crack.

4.2 Toe Model Output All BSTEM models generated from the stability evaluation were simulated to quantify the amount of the toe erosion based on three conditions, namely, (i) initial condition; (ii) vegetation effect; and (iii) bank protection effect. These conditions were assigned to assess the potential amount of lateral retreat, quantifying eroded materials at the bank and the toe. Finally, the total area of eroded materials can be quantified. As the result, comparison in term of eroded areas and the average boundary shear stress were made between month 1 and month 2. The input duration for the used software was twenty-four hours. The percentage of vegetation used was (no vegetation 67.67% + 33.37% of Perennial Ryegrass), while the age of the plant is 10 years, producing a cohesion of 0.5 KPa due to roots. The proposed bank protection was (rip rap wall with D50 0.256 m) induced on the Bank Toe Model. Table 4 shows the BSTEM analysis obtained after running the toe Model Output at the right bank and left bank of Sg. Bernam for month 1 and month 2. For the right bank, the initial condition of the bank indicated the lowest amount of 17.88 Pa, in comparison to vegetation effect (18.16 Pa) and bank protection effect (18.16 Pa). These results are significantly concurrent with the amount of lateral retreat and eroded area of the bank and toe, as the average applied boundary shear stress on the bank increases, the amount of lateral retreat and eroded area decreases with respect to vegetation effect and bank protection effect. The results to lateral retreat for the right bank yielded 28.46 cm (month 1) and 76.80 cm (month 2). However, when vegetation cover effect applied to the bank, the lateral retreat reduced to 11.76 cm (month 1) and 0.48 cm (month 2). The effect of bank protection improved the stability Table 4 BSTEM analysis obtained after running the Toe Model Output at the right bank and left bank of Sg. Bernam for month 1 and month 2 Bank

Right

Left

Case

Average Applied Boundary Shear Stress (Pa)

Maximum Lateral Retreat (cm)

Eroded Area-Total (m2 )

Month 1

Month 2

Month 1

Month 2

Month 1

Month 2

Initial

17.88

30.25

28.46

76.80

0.66

0.33

Vegetation

18.16

33.26

11.76

0.48

0.06

0.001

Bank Protection

18.16

33.26

0

0

0

0

Initial

31.65

40.84

17.43

40.05

0.18

0.31

Vegetation

41.02

39.65

27.94

72.13

0.11

0.25

Bank Protection

41.02

39.65

0

0

0

0

154

A. Saadon et al.

Fig. 6 Toe model output for the right bank under (a) initial condition, (b) vegetation cover, and (c) bank protection using rip rap.

of the bank significantly, where it can be seen from Table 4 that the effect of bank protection (rip rap wall with D50 0.256 m) stabilized the bank with zero lateral retreat. Similar trend occurred to account for the total eroded area, initial condition yielded 0.66 m2 of eroded materials for month 1, and 0.33 m2 of eroded materials for month 2. Total eroded area reduced with the effect of vegetation (0.06 m2 for month 1 and 0.001 m2 for month 2). Figure 6 shows comparison of toe model output for the right bank under initial condition, vegetation cover and bank protection using rip rap for month 1. The initial case for the left bank indicated average applied boundary shear stress of 31.65 Pa (month 1), in comparison to vegetation effect (40.12 Pa) and bank protection effect (40.12 Pa). For the second month, the initial condition indicated 40.84 Pa of average applied boundary shear stress, 39.65 Pa for vegetation effect and bank protection. These results are significantly concurrent with the amount of lateral retreat and eroded area of the bank and toe, as the average applied boundary shear stress on the bank increases, the amount of lateral retreat and eroded area decreases with respect to vegetation effect and bank protection effect. The results to lateral retreat for the left bank yielded 17.43 cm (month 1) and increased to 40.05 cm (month 2). However, for the left bank when vegetation cover effect applied to the bank, the lateral retreat increased to 27.94 cm (month 1) and 72.13 cm (month 2). Therefore, the option for vegetation cover was not suitable to increase the bank stability since the weight of the vegetation cover is not suitable for the existing bank geometry. To stabilize the bank with additional vegetation protection, the bank angle and bank height should be redesign. The effect of bank protection improved the stability of the bank significantly, where it can be seen from Table 4 that the effect of bank protection (rip rap wall with D50 0.256 m) stabilized the bank with zero lateral retreat. Similar trend occurred to account for the total eroded area, initial condition yielded 0.18 m2 of eroded materials for month 1, and 0.31 m2 of eroded materials for month 2. Total eroded area reduced with the effect of vegetation (0.11 m2 for month 1 and 0.25 m2 for month 2). Figure 7 (a), (b) and (c) show the toe model output for right bank (initial condition, vegetation effect and bank protection effect) for month 1.

Short Timescale Riverbank Erosion and Bank Stability of Sg. Bernam …

155

Fig. 7 Toe model output for the right bank under (a) initial condition, (b) vegetation cover, and (c) bank protection using rip rap

5 Conclusion In this study, BSTEM was utilized to predict bank stability, bank retreat and eroded area for left bank and right bank of Sg. Bernam. Data extraction from Sg. Bernam has been completed based on fieldwork investigation for two consecutive months. Two major analyses have been conducted from BSTEM, (i) assessing the bank stability in quantifying the Factor of Safety (FoS) based on the case with tension crack and without tension crack; and (ii) quantifying the amount of lateral retreat in cm and total eroded area in m2 using toe model function in BSTEM. For bank stability analysis, the BSTEM model was simulated for two cases. For case 1, the banks were assigned with no tension crack (H = 0 m), with and without vegetation, incorporating additional three sub-conditions to account for the effect of groundwater level, namely, dry condition, partially saturated condition, and fuller saturated condition. Similar effects were applied to model case 2, were the tension crack effect (H = 0.84 m) was added to the banks. The models were simulated with and without vegetation, incorporating additional three sub-conditions to account for the effect of groundwater level, namely, dry condition, partially saturated condition, and fuller saturated condition. The model output represents the FoS obtained from the three cases simulations. For month 1, the right bank of the river when the water was at the dry condition, the FoS obtained was 2.61, while for second month, the FoS obtained was 1.24. Partially saturated condition for month 1 yielded FoS of 1.85 while FoS for the second month was 0.72. Fully saturated condition yielded FoS of 1.04 (month 1) while 0.29, unstable bank for month 2. It can be concluded that the FoS obtained for the first month on the right bank was stable in comparison to the second month. From the results, it is observed that the FoS for the second month was not stable in both partially and fully saturated. The FoS obtained on the left bank when the water table was at dry condition was 3.92 (month 1) while 5.37 (month 2) of which it was considered the highest FoS obtained from both months. When the water table was partially saturated, the FoS reduced to 3.19 (month 1) while 3.23 (month 2). The FoS obtained when the water table was considered to be fully saturated was 1.75 (month 1) and 2.23 (month 2). Twelve BSTEM model were simulated using toe model output in quantifying the amount of lateral retreat in cm and total eroded area in m2 . The toe models were simulated under initial condition, vegetation effect and bank protection. Analysis

156

A. Saadon et al.

of both right bank and left bank indicated that erosion and lateral retreat occurred more severe in the second month, in comparison to the first month. Right bank recorded 76.80 cm lateral retreat with eroded area of 0.33m2 in the initial condition. The amount of eroded area and lateral retreat reduced significantly when vegetation cover assigned to the models. The simulated model using bank protection (rip rap with average diameter of 0.256 m) stabilized both right bank and left bank, subsequently reduced the amount of bank lateral retreat. In conclusion, BSTEM model successfully simulated the short timescale riverbank erosion and bank stability of Sg. Bernam.

References 1. AlMadhhachi AST, AlMussawy HA, Basheer MI, AbdulSahib AA (2020) Quantifying Tigris riverbanks stability of southeast Baghdad city using BSTEM. Int. J. Hydrol. Sci. Technol. 10(3):230–247 2. AlMadhhachi AT, Hanson GJ, Fox GA, Tyagi AK, Bulut R (2011) Measuring erodibility of cohesive soils using laboratory jet erosion tests. In: World Environmental and Water Resources Congress 2011: Bearing Knowledge for Sustainability, pp. 2350–2359 3. AlMadhhachi AST, Hanson GJ, Fox GA, Tyagi AK, Bulut R (2013) Measuring soil erodibility using a laboratory “mini” JET. Trans ASABE 56(3):901–910 4. AlMadhhachi AST, Hanson GJ, Fox GA, Tyagi AK, Bulut R (2013) Deriving parameters of a fundamental detachment model for cohesive soils from flume and jet erosion tests. Trans ASABE 56(2):489–504 5. Ashraf M, Shakir AS (2018) Prediction of river bank erosion and protection works in a reach of Chenab River. Pakistan. Arab J Geosci 11(7):1–11 6. Cancienne RM, Fox GA, Simon A (2008) Influence of seepage undercutting on the stability of root-reinforced streambanks. Earth Surf Process Landf Br Geomorphol Res Gr 33(11):1769– 1786 7. Constantine CR, Dunne T, Hanson GJ (2009) Examining the physical meaning of the bank erosion coefficient used in meander migration modeling. Geomorphology 106(3–4):242–252 8. Daly ER, Miller RB, Fox GA (2015) Modeling streambank erosion and failure along protected and unprotected composite streambanks. Adv Water Resour 81:114–127. https://doi.org/10. 1016/j.advwatres.2015.01.004 9. Duan JG, Julien PY (2010) Numerical simulation of meandering evolution. J Hydrol 391(1– 2):34–46 10. El KadiAbderrezzak K, Moran AD, Mosselman E, Bouchard J-P, Haberscak H, Aelbrecht D (2014) A physical, movable-bed model for non-uniform sediment transport, fluvial erosion, and bank failure in rivers. J Hydro- Environ Res 8(2):95–114 11. Enlow H, Fox G, Guertault L (2017) Watershed variability in streambank erodibility and implications for erosion prediction. Water 9(8):605. https://doi.org/10.3390/w9080605 12. Enlow HK et al (2018) A modeling framework for evaluating streambank stabilization practices for reach-scale sediment reduction. Environ Model Softw 100:201–212 13. Hamshaw SD, Engel T, Rizzo DM, O’NeilDunne J, Dewoolkar MM (2019) Application of unmanned aircraft system (UAS) for monitoring bank erosion along river corridors. Geomat Nat Haz Risk 10(1):1285–1305 14. Hanson GJ, Simon A (2001) Erodibility of cohesive streambeds in the loess area of the midwestern USA. Hydrol Process 15(1):23–38 15. Midgley TL, Fox GA, Heeren DM (2012) Evaluation of the bank stability and toe erosion model (BSTEM) for predicting lateral retreat on composite streambanks. Geomorphology 145:107–114. https://doi.org/10.1016/j.geomorph.2011.12.044

Short Timescale Riverbank Erosion and Bank Stability of Sg. Bernam …

157

16. Mittelstet AR, Storm DE, Fox GA, Allen PM (2017) Modeling streambank erosion on composite streambanks on a watershed scale. Trans ASABE 60(3):753–767 17. Motta D, Abad JD, Langendoen EJ, Garcia MH (2012) A simplified 2D model for meander migration with physically-based bank evolution. Geomorphology 163:10–25 18. Piégay H, Darby SE, Mosselman E, Surian N (2005) A review of techniques available for delimiting the erodible river corridor: a sustainable approach to managing bank erosion. River Res Appl 21(7):773–789 19. Randle TJ (2006) Channel migration model for meandering rivers. In: Proceedings of the 8th Federal Interagency Sedimentation Conference, pp. 241–248 20. Avila JJR, McAnally WH, Langendoen EJ, Achury SLO, Martin JL (2011) The role of streambank erosion contributions to sediment loads in the Town Creek Watershed in Mississippi. In: International Symposium on Erosion and Landscape Evolution (ISELE), 18–21 September 2011, Anchorage, Alaska, vol. 124. American Society of Agricultural and Biological Engineers 21. Rinaldi M, Casagli N, Dapporto S, Gargini A (2004) Monitoring and modelling of pore water pressure changes and riverbank stability during flow events. Earth Surf Proc Land 29(2):237– 254 22. Rinaldi M, Darby SE (2007) 9 Modelling river-bank-erosion processes and mass failure mechanisms: progress towards fully coupled simulations. Dev Earth Surf Process 11:213–239 23. Rinaldi M, Nardi L (2013) Modeling interactions between riverbank hydrology and mass failures. J Hydrol Eng 18(10):1231–1240 24. Saadon A, Abdullah J, Muhammad NS, Ariffin J, Julien PY (2021) Predictive models for the estimation of riverbank erosion rates. CATENA 196:104917. https://doi.org/10.1016/j.catena. 2020.104917 25. Saadon A, Abdullah J, Muhammad NS, Ariffin J (2020) Development of riverbank erosion rate predictor for natural channels using NARX-QR Factorization model: a case study of Sg. Bernam, Selangor, Malaysia. Neural Comput Appl 32(18):14839–14849. https://doi.org/10. 1007/s00521-020-04835-5 26. Saadon A, Ariffin J, Abdullah J, Daud NM (2016) Dimensional analysis relationships of streambank erosion rates. Jurnal Teknologi 78(5–5):1–7. https://doi.org/10.11113/jt.v78.8580 27. Simon A, Curini A, Darby SE, Langendoen EJ (2000) Bank and near-bank processes in an incised channel. Geomorphology 35(3–4):193–217 28. Simon A, PollenBankhead N, Mahacek V, Langendoen E (2009) Quantifying reductions of mass-failure frequency and sediment loadings from streambanks using toe protection and other means: Lake Tahoe, United States. JAWRA J Am Water Resour Assoc 45(1):170–186. https:/ /doi.org/10.1111/j.1752-1688.2008.00268.x 29. US Department of Agriculture (2016) Bank Stability and Toe Erosion Model, v5.4, Model documentation. USDA-ARS, Oxford, MS 30. Klavon K, Fox G, Guertault L, Langendoen E, Enlow H, Miller R, Khanal A (2017) Evaluating a process-based model for use in streambank stabilization: insights on the bank stability and toe erosion model (BSTEM). Earth Surf Proc Land 42(1):191–213

3D Simulation on 90 Degree Off-Take Branching Channel with Separation Zones Siti Aimi Asyarah Zakaria, Mohd Ridza Mohd Haniffah, Amyrhul Abu Bakar, M Faizal Ahmad, and Iskandar Shah Mohd Zawawi

Abstract Flow separation occurs often in river diversion, particularly in branch channels. The flow separation is generally seen through its separation area, which is influenced by a number of factors such as off-take angles, discharge ratios, and channel widths. The study was conducted through computational modelling of FLOW-3D to create a relationship between discharge ratio with the size of the separation area. The model domain was setup with a channel bifurcating into two, the main channel downstream and a channel representing the river diversion at a 90°. The model was first validated with previous laboratory work. Then, three discharge ratios were simulated, which are 0.21, 0.49 and 0.70. The layers within the depth of the water were also investigated to understand whether the separation area is homogenous within the water depth via the vertical confinement product. The length of separation area was observed for a few layers within the water depth. The results shows that the separation area is homogenous within the water depth and a single depth can be applied to represent the whole channel. As for the separation area, the result shows a good agreement with previous work. The length of separation zone decreases as discharge ratio increases. Keywords River bifurcation · River diversion · Branching channel · Flow separation · FLOW-3D · Computational modeling S. A. A. Zakaria (B) · M. R. M. Haniffah · A. Abu Bakar · M. F. Ahmad Department of Water and Environment Engineering, School of Civil Engineering, Faculty of Engineering, Universiti Teknologi Malaysia, 81310 Johor Bahru, Johor, Malaysia e-mail: [email protected] M. R. M. Haniffah e-mail: [email protected] A. Abu Bakar e-mail: [email protected] I. S. M. Zawawi Centre for Mathematics Studies, Faculty of Computer & Mathematical Sciences, Universiti Technology MARA, 40450 Shah Alam, Selangor, Malaysia e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023 I. K. Othman et al. (eds.), Proceedings of the 5th International Conference on Water Resources (ICWR) – Volume 2, Lecture Notes in Civil Engineering 365, https://doi.org/10.1007/978-981-99-3577-2_11

159

160

S. A. A. Zakaria et al.

1 Introduction In Malaysia, there are several methods of dealing with flood that was established by the Department of Irrigation and Drainage under the Ministry of Environment and Water. Among them are flood control dam, flood diversion channel, flood detention pond, etc. Batu and Jinjang diversions and SMART Tunnel are flood diversion projects that have been completed in Kuala Lumpur. The purpose of these projects is to reduce the flowrate by diverting excess water from the main river into a channel during heavy rainfall. Flow regulation is required at the diversion and the floodplain topography and its characteristics need to be considered to eliminate flood risk. With regards to river bifurcation, there are studies of discharge ratio and separation zone in main and branch channels. Physical models were an early approach where the research focused on 90 degree off-take angle [1, 3, 5]. [3] construct an experiment with subcritical and supercritical flow in the main and branch channel, respectively. The experiment found that, at a specific width ratio, the discharge ratio between branch and upstream decreased as the Froude number increased up to Froude number is 0.8 which cause flow depth increase to critical depth that cause fluctuation flow. [5] discussed the return flow in branch channels and division streamlines that correlated to Froude numbers in main and branch channels. From the study, the size and shape of recirculation zone increased as the Froude number in branch channel decreased. [1] found the relationship between the depth ratio and discharge ratio and energy-loss coefficient for subcritical flow in branch. The depth ratio increased as discharge ratio increased and Froude number at the downstream decreased. Besides physical model, [9] apply a computational model of FLUENT-2D to study relationship between discharge ratio and separation zone dimensions in 90degree junction. There is good agreement between the model and the experiment by [2] where the length and width of separation zone decreased with increased of discharge ratio. [7] use analytical approach to study the mechanism of combining and dividing flow in rectangular and trapezoidal channel. They found that, the shape of channel also effects the depth ratio even the same bottom and flow area with constant discharge were applied. The rectangular channel can obtain higher depth ratio compared to trapezoidal for equal flow area. [4] use local frame axis method to study the separation flow at a large angle between the main and branch channel. The separating streamline and recirculation zone began to appear from the upstream corner to downstream of the branch channel. [11] solve multiple off-takes bifurcation channel with the aid of momentum principle and mass continuity. The study found that, the right-angled river bifurcation is preferable to mitigate floods. The aim of this paper is to compare the simulation results of FLOW-3D to the numerical modelling of FLUENT-2D by [9] that incorporates the experiment by [2] on the separation zone at 90-degree off-take angle. The numerical modelling was validated first through previous experimental work by [8].

3D Simulation on 90 Degree Off-Take Branching Channel …

161

2 Simulation Setting up This study consists of simulations of 90 degree off-take angle using 3D modelling through FLOW-3D software. The discharge ratio and flow separation were observed and compared to the previous work of [9] and [2]. Figure 1 shows the simulation set up for 90° off-take angle branching channel. The set up consists of main channel and intake (branch). The length of main channel and branch channel are 8.15 m and 5.0 m respectively. Both of the channels are 0.3 m wide and 0.25 m deep. The surface is assumed to be smooth with horizontal bed slope. The water is coming from the left (Inlet) to the right and then diverted into branch channel (Outlet 2). Outlet 1 is the downstream of the main channel. The discharge from the inlet is constant at 0.0024 m3 /s, while, the discharge in the branch channel is varied by three discharge ratios, (Qr) which are 0.21, 0.49 and 0.7. The discharge ratio is the ratio of the flowrate in the branch to the flowrate in the upstream main channel. Then, the vertical confinement and flow separation were observed.

Fig. 1 Simulation layout with boundary condition (Q for Volume Flowrate, O for Outflow, S for Symmetry and W for Wall)

162

S. A. A. Zakaria et al.

3 Physics Option in FLOW-3D In FLOW-3D, there are many different physics available. For this study, two physics were activated, gravity and turbulence model. The gravitational acceleration in the vertical (z) direction is set to -9.81 m/s2. As for the viscosity turbulence model for 90 degree branching channel the k-ε turbulence model was applied. The option of wall shear boundary conditions is automatically set to no-slip condition when viscous flow is selected. For the fluid properties, water at 20°C was applied with density and viscosity of 1000 kg/m3 and 0.001 kg/m/s respectively.

4 Initial and Boundary Conditions Table 1 shows the initial and boundary conditions of the simulation. The settings of boundary conditions will affect the simulation results. The boundary condition at the main channel upstream was set to volumetric flowrate (Q) and the outlet of downstream of the main channel (Outlet 1) was set to outflow (O) boundary condition. The outlet of downstream branch channel (Outlet 2) was set to wall (W) boundary condition and a pump is used to control flowrate into branch channel to represent the same flow condition with the previous work. The boundary at the top of the channel is set to pressure (P) boundary condition. The initial condition follows the actual flow data of the water elevation and hydrostatic pressure. The flowrate at the upstream was gradually increased from zero to the required flowrate to reduce water splashing and the immediate change of velocity, hence increasing the stability of the simulation and shorten the time to reach its steady state. Table 1 Initial and boundary condition in FLOW-3D Boundary Condition Upstream Main Channel

Downstream Main Channel (Outlet 1)

Branch Channel (Outlet 2)

Top of the Channel

Volumetric flowrate

Outflow

Wall + pump

Pressure

Turbulence Model

Initial Fluid Elevation (m)

k-ε

0.219 (From origin)

3D Simulation on 90 Degree Off-Take Branching Channel …

163

5 Separation Zones and Vertical Confinement Separation zone will occur near the entrance of the branch channel due to the incapability of the flow in the main channel to adapt to the sudden change in geometry at the junction as shown in Figure 2. Thus, flow will be divided into two regions within the branching channel; the flow trapped in circulation of the separation zone and the flow that moves straight forward towards the end of the branch. In order to identify the separation zone, the variation of the flow field in the depth direction must be known. The velocity field can be characterized as homogeneous in the z direction by calculating the correlation product between the velocity fields at different horizontal slices along the depth. The correlation product is as shown in Equation (1) by [10]. C=

 i, j

1 u 1 u 2 + v1 v2   u 1 2 + v1 2 u 2 2 + v2 2 nb points

(1)

where (u 1 v1 ), (u 2 v2 ) are the velocity components given at each point of the mesh grid (nb points ) on two different slices, and i, j represent the mesh grid constituted by the nb points . A correlation product of 1 means that the velocity fields are totally identical between the two slices.

Fig. 2 Separation zone in branch channel

164

S. A. A. Zakaria et al.

6 Results and Discussion 6.1 Model Validation for Upstream Flowrate with Discharge Ratio The numerical modelling of branching channel in FLOW-3D software was first validated with another laboratory experiment by [8] by comparing the effect of varying upstream flowrates (Qu) to the discharge ratios. The configuration of this 90° branching channel consists of 8.0 m length of main channel and 3.0 m length of branch channel as shown in Figure 3. The width and depth of both channels are 0.2 m. The downstream part of both the main and branch channels are free from any structure. Three flux surfaces and probes were implemented in the channels. The purpose of the flux surfaces are to measure discharge in the channel while the probe is to measure flow depth and Froude number at a point during the simulation. The discharges in the main channel were measured at 0.15 m and 8.0 m from upstream, while discharge in branch channel was measured at 2.9 m from the junction of the bifurcated channel. Then, discharge ratios were calculated.

Fig. 3 Experimental layout with flux surface and probes

The comparison of FLOW-3D results to the experimental study is as shown in Fig. 4. Both the results show that as the upstream flowrate increases, the amount of flowrate into the diverted channel or the branched channel reduces. The results from FLOW-3D simulations and experiment data differ at a minimal value and this validates that the numerical model is able to recreate the 90 degree off-take branching channel flow.

3D Simulation on 90 Degree Off-Take Branching Channel …

165

Fig. 4 Validation result between experiment [8] and FLOW-3D

6.2 Vertical Confinement Once the model has been validated, the model is set up according to previous works by [2, 8] as shown in Figure 1. For the vertical confinement, correlation products between the velocity fields on four horizontal slices were made; Slice 1 = 0.06 m from the bottom of the channel and followed by Slice 2 = 0.09 m, Slice 3 = 0.12 m, Slice 4 = 0.15 m, and Slice 5 = 0.18 m. The correlation product was computed based on the values provided between Slice 1 and Slice 2, Slice 2 and Slice 3, Slice 3 and Slice 4, and Slice 4 and Slice 5. Figure 5, 6 and 7 show the streamlines at the branching channel for each slice, for the three different discharge ratios. For each discharge ratio, the streamlines are almost similar to each other for all slices. This is further proven through the correlation products, as shown in Figure 8. The correlation products for all cases lay between 0.997 and 0.999, indicating that the variation in velocity across the water depth is very small and the separation zones are homogenous across the water depth.

166

S. A. A. Zakaria et al.

Fig. 5 Streamlines at the branching channel by slices for Qr = 0.21

6.3 Separation Zones As the separation zones are homogenous across the water depth, only one slice is applied to represent the separation zone for each discharge ratio. Separation zones in the branching channel can be observed through the streamlines for the three discharge ratios as shown in Figure 9. As the discharge ratio increases, the length and width of separation zone decreases. The separation zone is a zone of circulated flow as can be seen by the elliptic shape of the streamlines in the separation zone shown in Figure 9. Some part of the flow will be able to follow through until the end of the branch channel, as shown by the narrow streamlines which diverges out at the end of the separation zone. Typically, the velocities in the separation zone will be lower compared to velocity in the main channel [6]. The low velocity area occurs at the left of the entrance in the branch channel where the velocity is close to 0 m/s at the centre of the separation zone. In addition, the effect of low velocity will increase the deposition rate of sediment. However, erosion might appear when the velocity is too high [4].

3D Simulation on 90 Degree Off-Take Branching Channel …

167

Fig. 6 Streamlines at the branching channel by slices for Qr = 0.49

For this simulation, the width and length of the separation zones are highly correlated to the previous works. The highest length and width of separation is when Qr = 0.21, followed by Qr = 0.49 and the smallest is when Qr = 0.70. Figure 10 shows the dimensionless length (SL /B) of separation zone versus discharge ratio. The length of the separation zone is measured parallel to the water flow. Results from laboratory experiment by [2] and numerical modelling using FLUENT-2D investigated by [9] are also included. The length of separation zone decreases as the discharge ratio increases. The trends of all the results are the same but with slight differences in values for FLOW-3D as compared to the two previous works. The agreement in the trend of the sizes of the separation zones when the discharge ratios change shows that the FLOW-3D simulation is able to reproduce the work done experimentally.

168

Fig. 7 Streamlines at the branching channel by slices for Qr = 0.7

Fig. 8 Correlation product for each discharge ratio

S. A. A. Zakaria et al.

3D Simulation on 90 Degree Off-Take Branching Channel …

Fig. 9 Streamlines at the branch channel for three discharge ratios

Fig. 10 Dimensionless length of separation zone (S L /B) versus discharge ratio (Qr)

169

170

S. A. A. Zakaria et al.

7 Conclusion The aim of this present study is to observe and understand the separation zone that occurs in the branching channel of a river bifurcation with three different discharge ratios. The numerical modelling of branching channel in FLOW-3D was first validated with laboratory experiment by [8]. The results from FLOW-3D simulations and experiment data differ with a minimal value, thus, the numerical was validated. The separation length in 90° branching channel with three discharge ratios were observed. The results show that as the discharge ratio increases, the length of the separation zone decreases. This is consistent with the previous works but with slight differences in values when compared in terms of the dimensionless length of the separation zone. The agreement in the trend of the sizes of the separation zones when the discharge ratios change shows that the 3D simulation is able to reproduce the work done experimentally. The information available in this 3D simulation results will help to realize the relationship between inlet flowrate and discharge ratio through the resulting streamlines and separation zones. Acknowledgements The research work is funded under the Collaborative Research Grant, UTM of references Q.J130000.2451.08G89. This particular research is in collaboration with Universiti Teknologi MARA, under a general collaboration involving University of Nottingham Malaysia and Universiti Teknologi Petronas.

References 1. Hsu CC, Tang CJ, Lee WJ, Shieh MY (2002) Subcritical 90 equal-width open-channel dividing flow. J Hydraul Eng 128(7):716–720. https://doi.org/10.1061/(asce)0733-9429(200 2)128:7(716) 2. Kasthuri B, Pundarikanthan NV (1987) Discussion of “Separation Zone at Open-Channel Junctions” by James L. Best and Ian Reid (November, 1984). J Hydraul Eng 113:543–544 3. Krishnappa G, Seetharamiah K (1963) A new method of predicting the flow in a 90-branch channel. La Houille Blanche 7:775–778. https://doi.org/10.1051/lhb/1963055 4. Mignot E, Doppler D, Riviere N, Vinkovic I, Gence JN, Simoens S (2014) Analysis of flow separation using a local frame-axis: application to the open-channel bifurcation. J Hydraul Eng 140:280–290. https://doi.org/10.1061/(asce)hy.1943-7900.0000828 5. N. S. Lakshamana, K. Sridharan, M. Y. A. B. (1968). Division Flow in Open Channel.Pdf (pp. 139–142). 6. Alomari NK, Yusuf B, Ali TAM (2016) Flow in a branching open channel: a review. Pertanika J Scholar Res Rev 2(2):40–57 7. Pandey AK, Mishra R (2012) Comparison of flow characteristics at rectangular and trapezoidal channel junctions. J Phys Conf Ser 364:012141. https://doi.org/10.1088/1742-6596/ 364/1/012141 8. Sayed T (2019) An experimental study of branching flow in open channels. Limnol Rev 19(2):93–101. https://doi.org/10.2478/limre-2019-0008 9. Shamloo H, Pirzadeh B (2008) Investigation of characteristics of separation zones in T-junctions. WSEAS Trans Math 7(5):303–312

3D Simulation on 90 Degree Off-Take Branching Channel …

171

10. Sui B, Huang SH (2017) Numerical analysis of flow separation zone in a confluent meander bend channel. J Hydrodyn 29(4):716–723. https://doi.org/10.1016/S1001-6058(16)60783-7 11. Zawawi ISM, Abdullah NL, Aris H, Jaafar BA, Norwaza NAH, Yunos MHFM (2019) Mathematical modeling for flood mitigation: effect of bifurcation angles in river flowrates. Civil Eng Archit 7(6A):50–57. https://doi.org/10.13189/cea.2019.071406

Saline Water and Freshwater Interactions in a Narrow Meandering Channel Mazlin Jumain, Zulkiflee Ibrahim, Wan Nor Afiqa Wan Mustafah Kamal, Sharifah Nurfarain Syed Abdul Jabar, Md.Ridzuan Makhtar, Noorarbania Abd Rani, Nurfarhain Mohamed Rusli, and Mohd Zulkhairi Mat Salleh

Abstract Saltwater intrusion has been a global issue in water resources management and ecological engineering. This phenomenon leads to problems such as encroachment into water intake zone, loss of freshwater vegetation and disturbance to aquatic life habitat. Undeniably climate change increases the saline water flow into the river system. The meandering rivers are common, and the hydraulics is more complex than straight rivers. Experimental hydraulic research was carried out in the Universiti Teknologi Malaysia to elucidate the hydrodynamic interactions between saline water and freshwater in a narrow meandering channel. The spatio-temporal salinity profiles M. Jumain (B) · Z. Ibrahim Centre for River and Coastal Engineering, UniversitiTeknologiMalaysia, UTM, 81310 Johor Bahru, Johor, Malaysia e-mail: [email protected] Z. Ibrahim e-mail: [email protected] M. Jumain · Z. Ibrahim · W. N. A. Wan Mustafah Kamal · S. N. Syed Abdul Jabar · M. Makhtar · N. Abd Rani · N. Mohamed Rusli School of Civil Engineering, Faculty of Engineering, Universiti Teknologi Malaysia, UTM, 81310 Johor Bahru, Johor, Malaysia e-mail: [email protected] S. N. Syed Abdul Jabar e-mail: [email protected] M. Makhtar e-mail: [email protected] N. Abd Rani e-mail: [email protected] N. Mohamed Rusli e-mail: [email protected] M. Z. Mat Salleh Faculty of Civil Engineering, Universiti Teknologi MARA Pasir Gudang Campus, 81750 Masai, Johor, Malaysia © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023 I. K. Othman et al. (eds.), Proceedings of the 5th International Conference on Water Resources (ICWR) – Volume 2, Lecture Notes in Civil Engineering 365, https://doi.org/10.1007/978-981-99-3577-2_12

173

174

M. Jumain et al.

along the river are discussed in this paper. The findings prevailed a typical characteristic of a salt-wedge estuary and indicated the processes of estuarine mixing. As the saltwater flows upstream, the salinity level drops due to dilution process. Freshwater discharge also significantly influenced the mixing of freshwater and saline water. Velocity of saline water, Us decreased up to 25% when freshwater discharge increased for Case A and B. A rapid dilution rate was observed in Case C due to strong velocity forces produced by a high freshwater discharge. Furthermore, the interaction between freshwater and saline water in a narrow meandering river might be influenced by the flow resistance induced by the channel boundaries and meander planform itself. Keywords Saline water · Freshwater · Meandering channel · Experimental hydraulics · Mixing

1 Introduction Saltwater intrusion has been a main issue in recent decades. In coastal areas, saltwater intrusion is a common hydrologic process in which saline water invades a river and mixes with freshwater. It is one of the main mechanisms that degrades the consistency of both surface water and groundwater making it unsuitable for drinking. Apart from that, saltwater intrusion will alter the species abundance and composition of terrestrial and marine ecosystems because of loss of adequate habitat. Global climate change is predicted to cause an increase in sea level rise, and the frequency and size of storms and storm usage. This will contribute further to shoreline erosion; flood damage, inundation of land, saltwater intrusion into the freshwater lens aquifer, among others [9, 12]. Rapid urbanisation has increased daily water supply demands, exacerbating the saltwater intrusion crisis, especially during the drought season [1]. Malaysia was hit by extreme El-Nino phenomenon in March 2014 and 2016. Due to extremely dry weather, the water supply system in Muar and Kuantan River basins had been affected. Complaints were received by the water authority from Muar and Kuantan citizens claiming that the tap water has a taste of saltiness. The authority also confirmed that this is due to the pumping of saline water into the treatment plant [5]. Salt intrusion in an estuary is subjected to many external forcing factors such as river discharge, tides, and topography [3, 14, 15]. Saltwater and freshwater tend to mix in estuaries where mixing characteristics can be distinguished as vertically mixed, slightly stratified, highly stratified or salinewedge [4, 13]. In partially mixed and highly stratified mixed estuaries, the structure of stratified flow is complicated by density gradient and tidal reciprocating flow. The combination of strong density stratification, turbulent diffusion and advection mean velocity led to complex circulation and mixing processes in estuaries [8, 11]. Apart from that, human activities such as dredging, and shoreline development influence the estuarine dynamics. These resulted on salt intrusion problems which directly

Saline Water and Freshwater Interactions in a Narrow Meandering Channel

175

affect the quality of water resources and give a threat to marine species. Hence, the study of salt intrusion is necessary to ensure water quality is safe and suitable for consumption. Meandering rivers are common feature in Malaysia and meander dynamics has been the focus of river engineering for decades. However, it is a challenge for researchers to precisely predict the flow behaviour and its geomorphological processes. The flow structure in a meandering river is highly three dimensional and extremely complicated as compared to a straight river. Furthermore, the understanding on the mixing between saltwater and freshwater in a meandering river is not fully understood. Field studies on saltwater-freshwater dispersion study is difficult and costly. Alternatively, a laboratory investigation can be implemented to elucidate the saltwater-freshwater mixing in a meandering river. The aim of the study was to establish fundamental knowledge on the saltwater and freshwater interactions in a narrow meandering river. This study focussed on the spatio-temporal salinity intrusion profiles along the meandering channel. Thus, the objectives of research were to investigate the freshwater salinisation profiles in a narrow meandering channel due to variation of freshwater discharges. This research output is expected to contribute to a better understanding of saltwaterfreshwater interactions in a meandering river, which will lead to better water resource management and river management practices in the future.

2 Experimental Investigation A laboratory experiment was carried out in a 10 cm width narrow meandering channel with 690 cm length in Hydraulic Laboratory, Universiti Teknologi Malaysia. The meandering channel bed was set at a gradient of 0.1%. The channel was supported by steel structures and the wall of the channel was built using transparent acrylic sheets. Meanwhile, freshwater was supplied from a sump via a centrifugal pump with a maximum capacity of 15 litres per second (L/s). Strainers were provided in inflow tank to reduce the turbulence effect in the channel. A tailgate placed at the downstream was used to control water level in the channel practically to establish uniform flow condition during the experimental work. Freshwater ran from upstream of the flume moving to the downstream and saltwater from a constant head tank with a capacity of 50 L was released in opposite direction. The experiment was carried out with salinity level of 15 ppt. A bluecoloured dye in the saline water was used as a tracer to visualise the mixing in the channel. The experiment was conducted using a constant saline water discharge of 0.32 L/s. Meanwhile, the freshwater discharges were varied at 0.15 L/s and 0.45 L/ s. The plan view of meandering channel is illustrated in Fig. 1. During the study, the water surface levels along the channel were checked regularly until an equilibrium flow depth was developed. Freshwater and saltwater discharge were measured using a volumetric approach. The water samples were collected using siphoning system as shown in Fig. 2. There were 5 sampling stations established

176

M. Jumain et al.

Fig. 1 Experimental setup in the laboratory study

along the channel starting from downstream of the channel as shown in Table 1. Station 1 was the point where saltwater released (estuary) and Sect. 5 was the farthest location approaching the upstream (river). Station 1 to 5 were selected as boundary of measurement station for data collection in this study. The salinity levels were measured using YSI 30 SCT salinometer for a total duration of 180 s with a 60 s interval. x is longitudinal distance from saltwater released station and L is total length of experimental studied area. The total length of experimental studied area, L was 2620 cm.

Fig. 2 Water sampling system in a meandering channel

Table 1 Sampling stations

Station

x (cm)

x/L

1

0

0.00

2

278

0.11

3

1310

0.50

4

2342

0.89

5

2620

1.00

Saline Water and Freshwater Interactions in a Narrow Meandering Channel

177

Table 2 Details of experimental cases Case

Q f (L/s)

Q s (L/s)

Q f / Qs

D (cm)

U f (cm/s)

U s (cm/s)

A

0.15

0.32

0.47

4.5

3.33

7.11

B

0.28

0.32

0.88

6.0

4.67

5.33

C

0.40

0.32

1.25

7.5

5.33

4.27

3 Results and Discussion The experimental investigations were conducted under uniform flow condition. Meanwhile, the Reynolds and Froude numbers for freshwater and saltwater were greater than 4000 and less than one, respectively. Hence, the regimes of flow were classified as turbulent with sub-critical condition. Q f , Q s , D, U f and Us represent freshwater and saline water discharges, water depth, freshwater velocity, and saline water velocity, respectively. S is the salinity level at each measurement point, So is initial of salinity level (15 ppt), x is longitudinal distance, L is total length of studied area and z is the sampling elevation measured from the bed channel. Table 2 summarises the details of experimental study. The velocities of freshwater (U f ) and saline water (Us ) were in range of 3.33 cm/s to 5.33 cm/s and 4.27 cm/s to 7.11 cm/ s, respectively. It was noticed that U f increased up to 40%, while Us decreased up to 25% as the freshwater discharge rose.

3.1 Vertical Distribution of Fresh-Saline Waters Mixing Mixing in estuaries is generally being driven by several factors such as river discharge, tides, topography, and wind. Figure 3 shows the vertical distribution of fresh-saline waters mixing due to discharge ratio (Q f /Q s ) in a narrow meandering channel. The variation of salinity levels at selected measurement stations were plotted at a duration of 180 s. The finding prevailed that the salinity levels at bottom elevation were higher compared to the intermediate and water surface elevations in all cases. The freshwater lies on top flow layer due to its lower density while the saltwater remained in the mid-depth and bottom of the channel. As the discharge ratio (Q f /Q s ) increased, the salinity level (S) lessened. The salinity levels were observed to be 13.56 ppt, 10.29 ppt and 4.39 ppt at Q f /Q s of 0.47, 0.88 and 1.25, respectively. The salinity levels in Case C were lower than in Case A and B. It indicates that saline water having difficulty to intruded into upstream. The salinity profiles for Q f /Q s of 1.25 became nearly vertical line and seen as an almost zero salinity level at x/L of 0.89. These phenomena took place when the river discharge was very much larger than the saltwater discharge. The dilution process in Case C implies the effects of mixing processes in the estuarine system by strong velocity forces of

178

M. Jumain et al.

Fig. 3 Vertical distribution of salinity levels for distance x/L of (a) 0.11, (b) 0.50, and (c) 0.89

Saline Water and Freshwater Interactions in a Narrow Meandering Channel

179

ambient freshwater flow. Similar results on the salinity distribution were reported by [7, 10].

3.2 Longitudinal Salinity Profiles Naturally, mixing in open channel occurs in two directions: vertical and transverse. Density stratification affects transverse mixing more than vertical mixing [12]. Figure 4 depicts the normalised transverse/longitudinal saltwater intrusion patterns (S/So ) along the relative distance (x/L) in meandering channels at a duration of 180 s. It was noticed that the freshwater lies on top flow layer, while the saltwater accumulated in the mid-depth and bottom of the channel. The salinity level decreased from 12.1% to 90.2% as the measurement depth close to water surface in the channel. This was due to the lower density of freshwater compared to saline density. This represents the formation of salt wedge in the estuary as found and reported by [6]. It can also be noticed that the S/So in Case A was highest compared to Case B and Case C. As the discrepancy ratio of freshwater discharges to saltwater discharge (Q f /Q s ) drops, the saltwater can intrude more further into upstream. This indicates that the lower freshwater discharge influences the mixing and intrusion patterns. For Case A and Case B, it can be seen the sudden drop of salinity x/L at 0.89 and 1.0, especially for the z = 2 cm due to limitation of selection studied zone. A drastic drop of salinity also might be occurred due to velocity forces by ambient freshwater flow [2, 15].

3.3 Temporal Patterns of Saline Water Intrusion Saline water intrusion in the estuary is primarily regulated by river discharge and tide. The river discharge is the most important element influencing salinity intrusion in the estuarine system, with higher river discharge resulting in less salt intrusion and vice versa. Figure 5 exhibits the variation temporal patterns of saline water intrusion in a narrow meandering channel. Lower value of z/D indicates the position of deep channel and vice versa. Salinity levels measured at 60 (t1), 120 (t2) and 180 (t3) seconds for Case A were plotted. The finding prevailed that the salinity level along the channel decreases from t1 to t3. This provides evidence to the occurrence of dilution processes occurs when the saltwater is washed away by ambient freshwater. The salinity level S/So at bottom elevation z/D of 0.44 was higher compared to the intermediate z/D of 0.67 and water surface elevations z/D of 0.89. This shows that the presence of deep channel has led to substantial movement of saline wedge toe towards the upstream direction. Moreover, the highest percentage difference of salinity level S/So can be observed at relative depth (z/D) of 0.67 compared to the bottom and water surface elevations. The percentage difference was up to 78.9%. This suggests that the effects of the channel

180

M. Jumain et al.

  Fig. 4 Longitudinal salinity profiles for Q f /Q s of (a) 0.47, (b) 0.88, and (c) 1.25

Saline Water and Freshwater Interactions in a Narrow Meandering Channel

Fig. 5 Temporal saline water profiles at different z/D of (a) 0.44, (b) 0.67, and (c) 0.89

181

182

M. Jumain et al.

planform itself on the mixing processes and freshwater-saline water interactions in a narrow meandering river.

4 Conclusion The saline-fresh water interactions in a narrow meandering channel have been experimentally investigated. The conclusion can be drawn from the findings are: 1. Freshwater discharge influences the saline water intrusion profiles in a narrow meandering channel. As the saline water moves upstream, the salinity level gradually decreases towards upstream (river flow) due to dilution processes. The dilution rate and process occurred in short time when the discharge ratio (Q f /Q s ) increased. 2. Salinity level is greater at the bottom surface compared to the water surface layer, due to the high density of saline water as compared to freshwater. The salinity level decreases from 12.1% to 90.2% as the measurement depth close to water surface in the channel. In addition, the fresh-saline water mixing profiles in this study were classified as typical characteristic of a salt-wedge estuary. 3. Salinity level along the channel decreases with time. This provides evidence to the occurrence of dilution processes occurs when the saltwater is washed away by ambient freshwater. The variation of salinity levels in a narrow meandering channel were up to 78.9% and might be influenced by the flow resistance induced by the channel boundaries and meander planform itself. Acknowledgements The implementation of this experimental hydraulic research was funded under the Fundamental Research Grant Scheme (FRGS) and UTM grant of references FRGS No. 5F420, and UTMER No. 17J75, respectively. We would like to acknowledge all who involved directly and indirectly in this project, especially to Hydraulic Laboratory staff for providing us research facilities and manpower.

References 1. Adams JB, Pretorius L, Snow GC (2019) Deterioration in the water quality of an urbanised estuary with recommendations for improvement. Water SA 45(1):86–96 2. AlFuady MF, Azzubaidi RZ (2021) An experimental study on investigating and controlling salt wedge propagation. IOP Conf Ser Earth Environ Sci 779(1):012079. https://doi.org/10.1088/ 1755-1315/779/1/012079 3. Chen Q, Zhu J, Lyu H, Pan S, Chen S (2019) Impacts of topography change on saltwater intrusion over the past decade in the Changjiang Estuary. Estuar Coast Shelf Sci 231:106469 4. Fischer HB, List JE, Koh CR, Imberger J, Brooks NH (1979) Mixing in Inland and Coastal Waters. Academic Press 5. Gisen JI, Adnan SS, AhmadTajudin AA, Chan TX (2018) Investigation of salt intrusion condition in the Belat estuary. MATEC Web Conf 150:1–6

Saline Water and Freshwater Interactions in a Narrow Meandering Channel

183

6. Haron NF, Tahir W (2014) Physical model of estuarine salinity intrusion into rivers: a review. Adv Mater Res 905(2014):348–352 7. Haron NF (2018) Modelling of Salinity Intrusion for Transverse Flow during Extreme Flood Event in Kuala Selangor. Ph.D. thesis, Universiti Teknologi Mara, Malaysia 8. Haddout S, Maslouhi A, Igouzal M (2017) Predicting of saltwater intrusion in the Sebou river estuary (Morocco). J Appl Water Eng Res 5(1):40–50 9. Hoover DJ, Odigie KO, Swarzenski PW, Barnard P (2017) Sea-level rise and coastal groundwater inundation and shoaling at select sites in California, USA. J Hydrol Reg Stud 11:234–249 10. Ibrahim Z, Ab Halim AH, Bakar NA, Subramaniam S (2008) Experimental studies on mixing in a salt wedge estuary. Malays J Civil Eng. 20(2):188–189 11. Liu B, Peng S, Liao Y, Wang H (2019) The characteristics and causes of increasingly severe saltwater intrusion in Pearl River estuary. Estuar Coast Shelf Sci 220:54–63 12. Mohamad MF, AbdHamid MR, Awang NA, MohdShah A, Hamzah AF (2018) Impact of sea level rise due to climate change: case study of Klang and Kuala Langat districts. Int J Eng Technol 10(1):59–64 13. Scarlatos PD (1996) Estuarine Hydraulics, Environmental Hydraulics. In: Singh VP, Hager WH (eds.) Kluwer Academic Publishers, pp. 289–348 14. Tinh NX, Wang J, Tanaka H, Ito K (2020) Response of salinity intrusion to the hydrodynamic conditions and river mouth morphological changes induced by the 2011 tsunami. J Sci Technol Civil Eng (STCE) - NUCE 14(2):1–16. https://doi.org/10.31814/stce.nuce2020-14(2)-01 15. Zhu J, Cheng X, Li L, Wu H, Gu J, Lyu H (2020) Dynamic mechanism of an extremely severe saltwater intrusion in the Changjiang estuary in February 2014. Hydrol Earth Syst Sci 24(10):5043–5056

Understanding Variability of Groundwater Potentials in Western Sokoto Basin: Implications for Sustainable Groundwater Development Saadu Umar Wali, Noraliani Binti Alias, and Sobri Bin Harun

Abstract This study evaluates potential groundwater variability using hydrogeological data from 113 boreholes in the western Sokoto basin. The data comprised of static water level (Swl), pumping test (Pt), pumping water level (Pwl), estimated yield (Ey), and hand pump setting (Hps). Data were obtained from the Department for Rural Water Supply and Sanitation (RWASSA) Birnin kebbi. Multivariate analysis - Factor analysis (FA) was applied to analyse data. The FA indicated that most of the variability in groundwater is explained by variation in estimated yields. Boreholes in the Basement Complex section are characterised by lower yields than those in the Cretaceous and Tertiary sediments. Although the Gwandu formation is the most prolific aquifer, boreholes tapping the Illo formation are equally characterised by good water yield. Thus, the two aquifers present an excellent groundwater development potential which can be harnessed for large-scale irrigation, municipal and industrial uses. However, the lack of groundwater development restrictions driven by increased groundwater withdrawals has presented a tremendous challenge for sustainable groundwater resources management in the Sokoto basin. Thus, a policy guiding groundwater management is recommended.

S. U. Wali · N. B. Alias (B) · S. B. Harun Department of Water and Environmental Engineering, School of Civil Engineering, Faculty of Engineering, Universiti Teknologi Malaysia, UTM Skudai, 81310 Johor Bahru, Johor, Malaysia e-mail: [email protected] S. U. Wali e-mail: [email protected] S. B. Harun e-mail: [email protected] S. U. Wali Department of Geography, Federal University Birnin Kebbi. P.M.B., 1057, Birnin Kebbi, Kebbi State, Nigeria © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2023 I. K. Othman et al. (eds.), Proceedings of the 5th International Conference on Water Resources (ICWR) – Volume 2, Lecture Notes in Civil Engineering 365, https://doi.org/10.1007/978-981-99-3577-2_13

185

186

S. U. Wali et al.

Keywords Pre-Cretaceous Basement Complex · Cretaceous and Tertiary Sediments · Static Water Level · Estimated Yield · Factor Analysis · Groundwater Management

1 Introduction Groundwater is a significant water supply source for industrial, irrigation, and domestic uses, especially in developing countries [35, 47]. In remote areas, it is the only dependable source of water supply. While groundwater availability is primarily a geology function, it is increasingly impacted by climatic and anthropological factors [21, 53, 54]. Estimates show that about 2.5 billion people depend on groundwater for drinking [32]. In addition, groundwater aquifers offer 43% of the water used for agriculture globally. In past decades, groundwater withdrawals mostly linked to agriculture have resulted in substantial declines in the water table [32]. The Sokoto basin is underlain by interbedded partially-fused sand, clay, limestone, and gravel [52]. The formations ranged from Cretaceous to Quaternary in age and attained a depth of about 1067 m. The Illo and Gundumi group’s Tertiary deposits are the oldest deposits of the Cretaceous age [37, 38]. The Cretaceous and Tertiary deposits strike northward and plunge about 7 m/2 km in a north-westward direction. The deposits also thicken downdip, though southward along the outcrop was thicker, and the Sokoto and Rima Groups pinched out. Groundwater in the study area was confined as artesian water and unconfined just below the water table [5, 50, 52]. This condition is found in most permeable Cretaceous and Tertiary alluvial series members [1, 6]. Hydrological data, including hydraulic assessments for wells tapping the three aquifers, are discussed in detail by [5]. The sedimentary formations of the Sokoto basin are underlain by the Crystalline Basement Complex rock [4]. Groundwater mainly occurs in cracks and tabular partings under the basement complex formation. Analysis of boreholes data in the preCretaceous Basement Complex areas of Northern Nigeria revealed an average borehole yield of 880 gallons per hour (gph) from a mean depth of 37 m [5]. Generally, groundwater conditions of the basement complex section of the Sokoto basin are comparable to Nigeria’s Basement Complex [49, 51]. Therefore, evaluating the potential groundwater variability of the Sokoto basin’s basement and sedimentary aquifers is essential for improved groundwater management. Thus, the hydrogeological condition’s variation is expected to impact static water level, pumping test, pumping water level, estimated yield, and hand pump setting in the Southwestern Sokoto basin. This study seeks to evaluate the variability of the aforementioned hydrogeological variables in the western Sokoto basin and its implications for sustainable groundwater development.

Understanding Variability of Groundwater Potentials in Western Sokoto …

187

2 The Study Area The Sokoto basin is located in the sub-Saharan Sudan Belt of West Africa, commonly categorised as semiarid. The basin lies between Latitudes 10o and 14o N and Longitudes 3o and 7o E (Figure 1). The climate is hot, semiarid tropical (AW) [49–52]. Owing to the basin’s position in extreme Northwestern Nigeria and over 1000 km away from the sea, the basin remains generally dry for most year periods [1–15, 17]. Temperature is generally high and varies significantly with seasons. Annual rainfall ranged from 500 mm to over 1200 mm in the south. The relative humidity is highest in August, ~90%, and is lowest in December, 0.2

0.07 –0.2

0.04–0.07

0.025 – 0.04

7

5.5–7

4–5.5

2.5 - 4

15

10–15

L

Water bodies, spars, Coastal sands vegetation, swamp or bare rock

5–10

2-5