Optimizing category-based statistical metrics for selecting global climate models in rainfall projections for Peninsular Malaysia

Abstract Rainfall projections are crucial for climate impact assessments and water resource management, especially in regions like Peninsular Malaysia (PM), where climate variability has a profound effect on socioeconomic and environmental systems. This study aims to optimize category-based statistical metrics, namely, error, efficiency, correlation, composite, and others, to reduce redundancy and identify the most reliable global climate models (GCMs) for improving rainfall projection accuracy in PM. Using CHIRPS daily rainfall data and GCM outputs from the Coupled Model Intercomparison Project Phase 6 (CMIP6), a multi-filtration approach was employed to rank GCM performance based on optimized metrics. The selected metrics span five categories, namely, 7 error metrics, 5 efficiency metrics, 3 correlation metrics, 2 composite metrics, and 5 other metrics, to address a broad range of evaluation needs, ensuring a rigorous and non-redundant assessment of GCMs. A systematic selection of GCM evaluation metrics reduced redundancy by up to 64%, retaining eight key metrics. Principal component analysis (PCA) confirmed these metrics as the most representative, with the first two principal components (PCs) capturing 95.5–100% of the total variance. Using compromise programming index (CPI) and Fisher-Jenks classification, five top-performing GCMs, namely, TaiESM1 (0.5507), CMCC-ESM2 (0.6753), MRI-ESM2-0 (0.7473), CMCC-CM2-SR5 (0.8402), and EC-Earth3-Veg (1.0233), were identified, with TaiESM1 (overlap efficiency, OVL = 0.96) and CMCC-ESM2 (OVL = 0.91) most closely replicating CHIRPS rainfall, making them highly recommended for reliable climate projections. The study concludes that this optimization framework significantly enhances rainfall projection reliability, thereby supporting better-informed decision-making in climate adaptation and water resource planning. Furthermore, the methodology provides a replicable approach for improving GCM selection, particularly in regions with complex climate dynamics, advancing climate modeling for sustainable development planning.