INGENIEUR
INGENIEUR
to contribute to sustainable water management practices , mitigate risks , and ensure the efficient utilisation of this precious resource . Collaborative efforts between hydrologists , data scientists , and policymakers are essential to harness the power of ML for a more resilient and water-secure future .
Employing ML techniques is crucial to deal with the increasing amounts of data to cope with the challenges of climate change and the ever-increasing human impact on the environment . The rapid progress made in ML is arguably the most relevant development currently in the field of hydrology and climate change . Only in the past few years have these ML techniques become more prominent among researchers in hydrological processes and climate change ( Ardabili et al ., 2020 ).
Several ML algorithms have proven to be very efficient in hydrological modelling , as listed in Table 2 , including support vector regression ( SVR ), random forest ( RF ), extreme learning machine ( ELM ), back-propagation neural network ( BPNN ), artificial neural network-extreme learning machines ( ANN-ELM ), gaussian process regression ( GPR ), multilayer perceptron ( MLP ), long short-term memory ( LSTM ), process-based ( PB ), elastic net regression ( ENR ), extreme gradient boosting ( XGB ), modified stacking ensemble strategy ( MSES ), multiple linear regression ( MLR ), M5-pruned model ( M5P ), and linear regression ( LR ).
Climate Change Impact and Assessment
Today , climate change caused by increased greenhouse gas ( GHG ) concentrations and CO 2 levels in the atmosphere has worsened the water availability status for rice irrigation in Malaysia . Rice cultivation has shown a sign of critical irrigation , particularly in the northern region , where the water stress index for rice production surpasses that of other states ( Hanafiah et al ., 2019 ), and the western part is predicted to experience future depletions in water sources from river basins ( Ismail et al ., 2020 ). The uncertain trajectory of climate change in the future is poised to have an impact on water resources and the need for crop irrigation , which is highly sensitive to changes in precipitation and temperature , making the effective management and allocation of water resources a challenging task ( Srivastav et al ., 2021 ). The expected rise in food demand under climate change would also significantly impact irrigation demand , adversely impacting rice production . Therefore , the predicted shortage in future available water resources makes it imperative to evaluate the fluctuations and uncertainty of future water supply under the influence of climate changes to ensure the sustainability of rice production and water resources ( Mirzabaev et al ., 2023 ; Molotoks et al ., 2021 ).
Within Southeast Asia , climate change primarily stems from alterations in precipitation , surface temperature , sea temperature , and sea level ( IPCC , 2022 ). Scientists utilise advanced climate models to understand the complex dynamics of the evolving climate to make wellinformed decisions . Global Climate Models ( GCMs ) have been developed by many international groups participating in the Coupled Model Intercomparison Project ( CMIP ). GCMs are the climate models covering Earth at a relatively lower resolution with their grid systems , including ocean circulation , which are used to simulate decades to centuries . The period covered is typically up until 2100 , but in some scenarios , experimental data can reach up to 2300 . As shown in Figure 3 , GCMs divide the globe ( atmosphere and oceans ) into horizontal and vertical grids , creating gridboxes . The climate models are the most dependable datasets for assessing forthcoming challenges , signifying a collective and inter disciplinary endeavour that significantly advances the comprehension of climate dynamics . These models provide valuable insights into possible strategies for mitigation and adaptation ( Han et al ., 2020 ). The latest of its kind , CMIP and CMIP6 , fall under the Fifth Assessment Report ( AR5 ), presenting advancements in land use , energy considerations , and uncertainties linked to greenhouse gas emissions compared to the previous versions ( Moss et al ., 2010 ). Compared to its predecessors , CMIP6 is more expansive regarding the number of models and scenarios it encompasses ( Zarrin & Dadashi-Roudbari , 2021 ), with an increased horizontal resolution , increased complexity in the physical modelling schemes , and novel shared socioeconomic pathways ( SSPs ) ( Abbasian et al ., 2019 ; Ahmadi et al ., 2020 ; Riahi et al ., 2017 ; Taylor et al ., 2012 ). Specifically , CMIP6 stands as a foundational element in modern climate science , delivering a comprehensive perspective of Earth ’ s climate system and its
28 VOL 99 JULY - SEPTEMBER 2024