Catchment specific evaluation of Aphrodite’s and TRMM derived gridded precipitation data products for predicting runoff in a semi gauged watershed of Tropical India

Abstract

Data scarcity poses difficulties to rainfall-runoff modelling in a semi/un-gauged river basin. There are certain alternative precipitation data sources, including satellite derived and gauge data derived interpolated products. However, their efficiency in estimating runoff is questioned and proved to be condition specific. In this article, an attempt has been made to compare two mostly used gridded precipitation data products, i.e. TRMM and Aphrodites in predicting runoff for semigauged Kangshabati river basin of tropical part of India during three consecutive years using a simplistic NRCS-CN approach. Considering the purpose of comparing the results in an uncalibrated model setting was used and performance of the outputs of runoff simulations based on both the precipitation datasets were tested using coefficient of determination, RMSE and Nash–Sutcliffe methods. Catchment wise model performance was compared to understand the relationship pattern of the model and input data. The accuracy level of the model outputs for upper, middle and lower catchments using two different types of input precipitation data is distinguishable. Comparing the discharge magnitude as predicted by two simulations using statistical parameters, it is apparent that the accuracy level of TRMM based discharge prediction is more in upper catchment and Aphrodite based discharge prediction is more in lower catchment. In other words, TRMM is more effective than Aphrodite’s in recording higher precipitation or high intensity discharge than that of lower precipitation or low intensity discharge and vice versa.