Abstract
Droughts pose a critical threat to water-dependent sectors, particularly hydropower, a vital low-carbon energy source in Sarawak, necessitating an assessment of future drought impacts. Utilising Climate Hazards Group InfraRed Precipitation with Station data (CHIRPS) dataset, Standard Precipitation Index at 3 months (SPI-3) was identified as the most representative drought time scale, contributing most to pincipal component 1 (PC1), which explained 73.86% of the total variance. MPI-ESM1-2-HR (CPI = 0.057) and NorESM2-LM (CPI = 0.005), were identified as top-performing based on compromise programming index (CPI) and Fisher-Jenks classification, effectively simulating low rainfall percentiles with Mean Absolute Error (MAE) and Root Mean Square Error (RMSE) below 0.06 across 1st to 15th percentiles. Random Forest (RF) achieved the highest distributional agreement Overlap Coefficient (OVL = 0.83), with reduced Standard Deviation (SD) (55 mm) and near-zero skewness (-0.25), indicating superior bias correction performance. Temporally, drought frequency was projected to decrease by mid-century, then rose again by 2075–2100 to 5,037 moderate, 2,315 severe, and 1,058 extreme events, indicating a mid-century dip followed by a late-century increase. Hydropower vulnerability to drought in Sarawak varies temporally and spatially, with Batang Ai peaking mid-century (vulnerability index 1.00), Bakun and Murum most vulnerable by century’s end (indices 1.00 and 0.93), and Baleh consistently least affected (=0.17).
