Multiscale Evaluation of Global Gridded Precipitation Datasets: A Comparison with Observed Precipitation Dataset Over Indo-Gangetic Delta Region

Abstract

Accurate precipitation characterisation is vital for agricultural productivity and water resource management in the densely populated Indo-Gangetic Delta. This study provides a comprehensive multiscale evaluation of three prominent global gridded precipitation datasets, i.e., CHIRPS, ERA5 Land, and MERRA2, against 0.25° gridded observational data from the India Meteorological Department (IMD) at 90 grid points across Gangetic West Bengal (1981-2025). Performance was evaluated at daily, monthly, seasonal (Winter, Pre-Monsoon, Monsoon, and Post-Monsoon), and annual timescales using a statistical framework comprising Pearson Correlation [r], Mean Bias [MB], Nash-Sutcliffe Efficiency [NSE], Root Mean Square Error [RMSE], Kling-Gupta Efficiency (KGE), Mean Absolute Error (MAE), and Percent Bias (PBIAS). Results indicate pronounced scale-dependent performance across all products. Monthly accumulations exhibited the highest overall reliability, led by CHIRPS (r = 0.899, NSE = 0.784, KGE = 0.85, MAE = 38 mm, PBIAS = +8.0%). Conversely, at the daily scale, MERRA2 demonstrated superior overall fidelity (KGE = 0.45, r = 0.542, NSE = 0.260, RMSE = 9.62 mm, PBIAS = -6.0%), whereas CHIRPS (KGE = 0.35) and ERA5-Land (KGE = 0.36) showed greater high-frequency errors. Seasonal assessments revealed that, while CHIRPS captured core monsoonal dynamics effectively (KGE = 0.48, MAE = 208 mm), MERRA2 exhibited superior performance during dry winter conditions (KGE = 0.67, MAE = 12 mm, PBIAS ≈ 0.0%). These findings offer a critical quantitative benchmark for selecting reliable forcing datasets for hydrological modelling and climate adaptation strategies in the Indo-Gangetic Delta region.