Assessment of Multiple Satellite-Based Precipitation Estimates Over Muneru Watershed of India

Abstract

Satellite-based precipitation estimates (SPEs) having the advantage of near-global coverage and high spatial-temporal resolution, can improve the present hydrological prediction potential for a river basin or watershed. In the present study, three popular SPEs i.e., Climate Prediction Centre Morphing algorithm (CMORPH), Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks (PERSIANN), and TRMM Multi-satellite Precipitation Analysis (TMPA) precipitation datasets were evaluated with gauge based India Meteorological Department (IMD) gridded dataset using contingency and statistical table methods over an agricultural Watershed of South India. The evaluation was carried out at a spatial resolution of 0.25° × 0.25° (~26.5 km × 26.5 km) on a daily, monthly, seasonal, and yearly time scales for the years 1998–2013. Similar characteristics were observed with different SPEs in detecting rainfall events on various time scales. Overall, statistical and contingency indices analysis revealed that the performance of the TRMM dataset in predicting rainfall was better than the PERSIANN and CMORPH datasets. It was observed that the correlation between watershed-wide monthly average precipitations of TRMM dataset against the IMD dataset was good (R² = 0.79) as compared to PERSIANN (R² = 0.68) and CMORPH (R² = 0.57) datasets. The occurrence frequencies of very light and very heavy rainfalls were observed maximum and minimum, respectively with the IMD dataset, and the same pattern was observed with all the SPEs. Although, the TRMM dataset performed better than PERSIANN and CMORPH datasets, considerable interannual and seasonal variations were observed due to bias in SPEs. The present analysis reveals that the application of SPEs could be a potential alternative approach for hydrologic simulation over an un-gauged or data-sparse regions after suitable bias corrections.