Evaluating CORDEX-CORE Regional Climate Models for Precipitation Simulation: A Multi-Criteria Ranking Approach for Agro-Hydrological Applications in the Cauvery Delta, Tamil Nadu

Climate change has significantly influenced regional precipitation patterns, affecting agricultural productivity and water-resource sustainability in monsoon-dependent regions. Reliable climate projections are therefore essential for climate impact assessment and adaptation planning. However, the performance of Regional Climate Models (RCMs) varies considerably across regions and climatic conditions, necessitating rigorous evaluation before their application in local-scale studies. In this study, eight CORDEX-CORE South Asia (0.22°) RCMs were evaluated for their ability to reproduce historical precipitation over Thanjavur district, Tamil Nadu, India, during 1976–2005 using observed rainfall data from 13 rain gauge stations. Model performance was assessed using seven statistical metrics integrated through the Compromise Programming Index (CPI), together with evaluations of precipitation occurrence, rainfall intensity occurrence, and cumulative precipitation distribution. The results revealed considerable variability among the RCMs. RegCM-based models generally exhibited wet biases, whereas REMO and CCLIM tended to underestimate seasonal rainfall. Based on the selected metrics and CPI framework, NorESM–RegCM, NorESM-CCLIM, and NorESM-REMO demonstrated comparatively better overall performance. However, analyses of precipitation occurrence and distribution showed that model performance depended on the precipitation characteristic considered, with no single RCM consistently outperforming the others across all evaluation criteria. These findings demonstrate that a multi-criteria evaluation framework provides a more comprehensive assessment of RCM performance than conventional statistical metrics alone and offers a robust basis for selecting suitable RCMs for future climate projections and agro-hydrological applications in the Cauvery Delta region.