Comparison of different rainfall products with gauge-based measurements over Narmada River Basin, India

The present study evaluates and compares the performance of different rainfall products, namely, Climate Hazards Group InfraRed Precipitation with Station (CHIRPS), India Meteorological Department (IMD) gridded, Prediction of Worldwide Energy Resource (POWER), and Precipitation Estimation from Remotely Sensed Information using Artificial Neural Network-Climate Data Record (PERSIANN-CDR) with gauge-based measurements over Narmada River basin, India. The ground-based daily rainfall data (1981-2020) of 11 gauging stations have been collected from the Water Resources Department, Madhya Pradesh and the evaluation of rainfall product has been accomplished on a point-to-grid basis (nearest neighbor method) at annual and seasonal scales with the help of continuous and categorical statistical metrics. The results reveal a strong positive correlation (> 0.75) between rainfall estimates of different products and gauge-based measurements at annual scale demonstrating higher similarity in rainfall estimates and observed data, whereas seasonal estimates have exhibited comparatively weaker relationship. Likewise, percent bias (PBIAS) demonstrates least bias in annual and monsoon rainfall estimates and high in other seasons. These findings reveal that rainfall estimates tend to improve with increasing time scale (season to annual). However, majority of the rainfall products have overestimated the low rainfall (western region) and underestimated the high rainfall (eastern and southeastern regions). Further, the values of critical success index (CSI) indicate IMD gridded product outperforms in detecting rainfall events accurately followed by POWER, PERSIANN-CDR, and CHIRPS. These results suggest that IMD gridded estimates provide the best alternate to ground-based rain measurements. However, rainfall estimates from POWER, PERSIANN-CDR and CHIRPS can also be used in various hydrometeorological investigations over Narmada River basin.