Rainfall screening methodology using TRMM data over a river basin

Abstract

A regionalized rain/no-rain classification (RNC) based on scattering index methodology is developed for detecting rainfall signatures over the land regions of the Mahanadi basin (India), using data products from the passive and active sensors onboard the Tropical Rainfall Measuring Mission (TRMM), namely the TRMM Microwave Imager (TMI) and Precipitation Radar (PR). The proposed model, developed using data for two years from the orbital database, was validated using PR and in-situ data for selected case study events in 2011 and 2012. Performance evaluation of the model is discussed using 10 metrics derived from the contingency table. Overall, the results show superior performance, with an average probability of detection of 0.83, bias of 1.10 and odds ratio skill score greater than 0.93. Accurate rainfall detection is obtained for 95% of case study events. The relative performance of the proposed model is dependent on rainfall type, but it should be useful in rainfall retrieval algorithms for current missions such as the Global Precipitation Measurement Mission.