Streamflow projection under CMIP6 climate scenarios using a support vector regression: a case study of the Kurau River Basin of Northern Malaysia

• Published in: Environmental earth sciences 2024-02, Vol.83 (4), p.125, Article 125
• Main Authors: Nasir, Muhammad Adib Mohd, Zainuddin, Zaitul Marlizawati, Harun, Sobri, Kamal, Md Rowshon, Ismail, Habibu
• Format: Article
• Language: English
• Subjects: Agricultural management; Biogeosciences; Climate change; Climate effects; Climate models; Climate variations; Climatic data; Data analysis; Data processing; Dry season; Earth and Environmental Science; Earth Sciences; Environmental Science and Engineering; Geochemistry; Geology; Global climate; Global climate models; Global warming; Hydroclimate; Hydrologic models; Hydrology; Hydrology/Water Resources; Intercomparison; Irrigation; Minimum temperatures; Original Article; Precipitation; Rainfall; Rainy season; Resource availability; River basins; Root-mean-square errors; Seasons; Statistical analysis; Stream discharge; Stream flow; Streamflow changes; Support vector machines; Terrestrial Pollution; Water management; Water resources; Wet season
• ISSN: 1866-6280; 1866-6299
• DOI: 10.1007/s12665-024-11435-2

The forecasting of future streamflow aids researchers and policymakers to understand how changes in climate affect hydrological systems. However, traditional computational approaches demand intensive data specifically for the basin, and it is costly. The shift towards more contemporary and data-driven approaches known as support vector regression (SVR) in hydrological modeling utilizing only the hydro-climate data from Coupled Model Intercomparison Project Phase 6 (CMIP6) provides rapid input–output data processing with accurate future projection. CMIP6 is an updated and improved Global Climate Models (GCMs) for the exploration of the specific impacts of changing streamflow patterns for improved water management in agricultural areas. The delta change factor method was used to generate climate sequences, fed into the SVR model to project streamflow from 2021 to 2080. The SVR model fitted reasonably well, demonstrated by several statistical indicators, including Kling-Gupta Efficiency (KGE), Nash–Sutcliffe Efficiency (NSE), Percent Bias (PBias), and Root Mean Squared Error (RMSE), with the training phase performance surpassing the testing phase. Future projections indicated increased rainfall during the dry season for most months, excluding April to June. The rise in precipitation was particularly pronounced during the wet season. Maximum and minimum temperature projections increased for all SSPs, with SSP5-8.5 predicted a substantial increase. The projection revealed that seasonal streamflow changes would range between  – 19.1% to – 1.2% and – 7.5% to – 3.1% in the dry and wet seasons, respectively. A considerable streamflow reduction is anticipated for all SSPs in April and May due to increased temperatures, with the most pronounced impact in the SSP5-8.5. Assessing the effects of climate variations on water resource availability is crucial for identifying effective adaptation strategies to address the anticipated rise in irrigation demands due to global warming. The projected streamflow changes due to potential climate impacts are significant for Bukit Merah Reservoir, aiding the formulation of appropriate operational strategies for irrigation releases.