Examining the performance of precipitation products in characterizing the Indian summer monsoon rainfall (ISMR) using triple collocation

Abstract

Precipitation datasets are crucial for understanding and studying the hydroclimatic processes. The advent of satellite observations and reanalysis has enabled to produce the precipitation observation at considerably fine spatial and temporal scales. However, the applicability and representativeness of these data products are still not definite, specifically over a vast topographic, ecologic, and climatic gradient and data-scarce region like India, where reliable (quality-controlled) and dense ground observations spread homogenously over space are inaccessible. Additionally, Indian Summer Monsoonal Rainfall (ISMR) produces ∼ 75 % of annual rainfall and determines the success of the agrarian Indian society and economy and subsequently, the sustenance of millions of people. In this context, reanalysis and satellite datasets could play a vital role by providing historical and near-real-time accurate precipitation estimates in understanding the climatic variability and forecasting the hydrometeorological extremes such as drought, urban floods. Hence, this study aims to evaluate four state-of-the-art precipitation data products over India by deploying the robust triple collocation (TC) technique during the principal rainy season (June to September). TC is a powerful formulation that allows the quantification of the error and agreement of the independent datasets without having the knowledge of reference observation. We further determine the order of suitability (rank) of examined data products based on the minimum error and maximum agreement estimated from TC. Furthermore, we derive the regional suitability of the datasets based on the Köppen–Gieger climate zones. Results suggest that: (1) reanalysis data products (IMDAA and ERA5-Land) outperform (rank one) other datasets in the wet tropical monsoon regions (west coastal plain and northeastern India), the highest rainfall receiving regions; (2) CHIRPS is the most suitable dataset in the transitional climatic conditions and facilitates the characterization of moderate to low-intensity ISMR events; (3) IMD is found to be suitable in peninsular India, where high gauge density is present.