Challenge in the Application of Machine Learning in Hydrological Modelling
When dealing with hydrology systems , we are actually dealing with the stochasticity of nature . Therefore , it is not easy to perfectly replicate the streamflow data due to its complexity and dynamic nature . The climate conditions in Malaysia are very inconsistent , especially for the two planting seasons . The difference in climate nature between these two seasons somewhat disrupts the streamflow modelling process under ML ( Adib & Harun , 2022 ; Nasir et al ., 2024 ), and this also eventually impacts the future projections under GCMs . Although the model performance for streamflow modelling under ML easily achieves acceptable results , to maximise the model performance and especially to capture irregular extreme low flow or extreme high flow , it is suggested to either run the model development separately for the off-season and the main-season or use different ML algorithms for these two different seasons ( Nasir et al ., 2024 ). In addition , Adib and Harun ( 2022 ) put together two ML algorithms , SVR and RF , to minimise errors in the prediction of low flow and peak flow . These solutions could potentially improve the model performance but require additional work or more than one ML algorithm .
Conclusion
The intricate balance between water availability and the demands of rice cultivation underscores the critical importance of effective water resource management in sustaining food security , economic stability , and environmental sustainability , particularly in regions like Malaysia which has a significant rice production sector . The reliance on river basins and rivers for irrigation supply necessitates proactive planning and management strategies to address challenges posed by climate change and population growth . Integrating ML with GCMs presents a promising approach to projecting future irrigation water availability for rice granaries . By harnessing the power of ML algorithms to analyse climate data and simulate streamflow patterns , researchers and policymakers can gain valuable insights into potential water resource scenarios and develop adaptive strategies to ensure the resilience of rice production systems . In essence , the synergy between ML and GCMs offers a pathway towards a more resilient and water-secure future for rice cultivation , not only in Malaysia but also in other rice-producing regions globally . Through collaborative efforts among researchers , practitioners , and policymakers , we can leverage advanced technological tools to address the complex challenges facing agricultural water management and pave the way for sustainable rice production in the face of evolving environmental dynamics .
REFERENCE
Abbasian , M ., Moghim , S ., & Abrishamchi , A . ( 2019 ). Performance of the general circulation models in simulating temperature and precipitation over Iran . Theoretical and Applied Climatology , 135 ( 3 – 4 ), 1465 – 1483 .
Adib , M . N . M ., & Harun , S . ( 2022 ). Metalearning approach coupled with CMIP6 multi-GCM for future monthly streamflow forecasting . Journal of Hydrologic Engineering , 27 ( 6 ), 05022004 .
Adib , M . N . M ., & Harun , S . ( 2023 ). Machine learning algorithms with hydro-meteorological data for monthly streamflow forecasting of Kurau River , Malaysia . In Othman , I . K ., Haniffah , M . R . M ., & Jamal , M . H . ( Eds .), Lecture Notes in Civil Engineering ( Vol . 2 , pp . 29 – 41 ). Singapore : Springer .
Ahmadi , H ., Rostami , N ., & Dadashi-roudbari , A . ( 2020 ). Projected climate change in the Karkheh Basin , Iran , based on CORDEX models . Theoretical and Applied Climatology , 142 ( 1 – 2 ), 661 – 673 .
Ardabili , S ., Mosavi , A ., Dehghani , M ., & Várkonyi- Kóczy , A . R . ( 2020 ). Deep learning and machine learning in hydrological processes climate change and earth systems a systematic review . Lecture Notes in Networks and Systems , 101 , 52 – 62 .
Dorairaj , D ., & Govender , N . T . ( 2023 ). Rice and paddy industry in Malaysia : governance and policies , research trends , technology adoption and resilience . Frontiers in Sustainable Food Systems , 7 , 1093605 .
31