Evaluation of the accuracy of seven gridded satellite precipitation products over the Godavari River basin, India

Water is an important part of global circulation. Precipitation is the principal input into these cycles, and its measurement and evaluation in spatial and temporal scales are necessary. This paper assessed the performance of seven satellite precipitation products, i.e., Climate Hazards group Infrared Precipitation with Stations, Princeton Global Forcing’s, Tropical Rainfall Measuring Mission, Climate Prediction Center, Climate prediction center MORPHing technique, Precipitation Estimation from Remotely Sensed Information Using Artificial Neural Networks-Climate Data Record, Multi-Source Weighted-Ensemble Precipitation using India Meteorological Department gridded precipitation as a reference from 1998 to 2016 over the Godavari River basin, India, applying continuous and categorical metrics. The Nash–Sutcliffe efficiency, coefficient of determination, and root mean square error for Multi-Source Weighted-Ensemble Precipitation were 0.806, 0.831, and 56.734 mm/mon and for Tropical Rainfall Measuring Mission were 0.768, 0.846, and 57.413 mm/mon. Similarly, categorical metrics, i.e., highest accuracy, Peirce’s skill score, and lowest false alarm ratio, were recorded by Multi-Source Weighted-Ensemble Precipitation with 0.844, 0.571, and 0.462, respectively. Cumulative distribution function was also assessed for all the datasets, representing all products that overestimated low to medium precipitation events except Multi-Source Weighted-Ensemble Precipitation and followed a similar pattern of India Meteorological Department except for low precipitation events. All satellite precipitation products were estimated accurately for low-lying areas and inaccurately over low precipitation regions, large forests, and hilly regions. Overall, Multi-Source Weighted-Ensemble Precipitation and Tropical Rainfall Measuring Mission datasets performed better in evaluating precipitation, and these datasets can be used in hydrological applications.