Evaluation of CMIP6 models for future rainfall projections over India: Insights on bias correction and scenario-based variability

Rainfall in India is essential for agriculture, water resources, and socio-economic stability. Understanding its future behaviour under changing climate conditions is therefore critical. This study evaluates 41 CMIP6 global climate models using the compromise programming method to identify the most suitable models for projecting future rainfall over India. Five performance indicators viz. Correlation Coefficient, Normalized Root Mean Square Deviation, Skill Score, Nash–Sutcliffe Efficiency, and Kling–Gupta Efficiency were used to compute the Lₚ-metric. The top-performing models were EC-Earth3-AerChem, MPI-ESM1-2-LR, AWI-CM1-1-MR, AWI-ESM1-1-LR, and EC-Earth3-CC. However, since future data were unavailable for EC-Earth3-AerChem and AWI-CM1-1-LR, EC-Earth3-Veg-LR and FIO-ESM-2-0 were selected for bias-corrected multi-model ensemble (MME) projections. Bias correction using quantile mapping improved the representation of seasonal rainfall but showed some overestimation during highly variable months. Future projections under SSP2-4.5 and SSP5-8.5 scenarios show intensified monsoon rainfall, especially under SSP5-8.5, due to higher atmospheric moisture in warmer conditions. Non-monsoon seasons exhibit mixed rainfall trends. Overall, the study provides a systematic framework for model intercomparison and highlights the importance of incorporating these projections into regional planning to strengthen climate resilience in India’s rainfall distribution under changing climate conditions.