Comparative assessment of IMD and CHIRPS rainfall products over Southern Peninsular India: A multi-scale validation study (2001-2020)

Abstract

Accurate precipitation estimation over topographically complex, monsoon-dominated regions remains a persistent challenge for satellite-based products. This study presents a comprehensive multi-scale evaluation of the Climate Hazards Group InfraRed Precipitation with Station data (CHIRPS) against India Meteorological Department (IMD) gridded rainfall observations (0.25° × 0.25°) over Southern Peninsular India for the period 2001–2020. By integrating spatial bias diagnostics, pixel-level correlation analysis, seasonal stratification, anomaly-based stress testing, interannual variability assessment, and time–frequency analysis using continuous wavelet power spectra, the study moves beyond conventional validation to interrogate the physical consistency and scale-dependent reliability of CHIRPS under distinct monsoon regimes. CHIRPS exhibits strong temporal coherence with IMD at the regional scale (r = 0.888; RMSE = 180 mm), accurately reproducing the seasonal progression of the southwest (June, July, August and September) and northeast (October and November) monsoons, the characteristic west–east rainfall gradient, and interannual variability across the 20-year record. Time–frequency diagnostics reveal statistically significant co-dominant periodicities in the 4– 8-year band in both datasets, consistent with ENSO-scale modulation of monsoon rainfall, indicating that CHIRPS captures externally forced climate signals rather than dataset-specific retrieval artifacts. A key contribution of this study is the demonstration that CHIRPS performance is not spatially or seasonally uniform. Pixel-level bias analysis reveals a pronounced and physically organized positive bias along the windward slopes of the Western Ghats, dynamically linked to orographically enhanced deep convection during the southwest monsoon, while interior semi-arid and eastern coastal regions exhibit near-zero bias and high spatial agreement (r > 0.8). Despite discrepancies in absolute magnitude, CHIRPS consistently captures the sign, timing, and relative ranking of extreme drought (2002, 2004, 2009, 2015) and surplus (2005, 2010, 2020) years, confirming its reliability for anomaly-based monitoring applications. By synthesizing these diagnostics, the study establishes a physiographically grounded reliability zonation for CHIRPS over Southern Peninsular India, clearly delineating regions where the product can be applied without correction and zones where spatially adaptive bias correction is scientifically mandatory. These findings provide process-oriented guidance for the informed use of CHIRPS in climate variability assessment, drought monitoring, and hydroclimatic applications in monsoon-dominated tropical regions.