Multifeature classification based rainfall estimation using visible infrared TRMM data

Abstract

In this study an attempt has been made to estimate the rain potential of clouds using the signatures in visible and infrared spectral channels by multi spectral classification approach. The data for this study was obtained from Visible Infrared Scanner (VIRS) and TRMM Microwave Instrument (TMI) onboard Tropical Rainfall Measuring Mission (TRMM) satellite. Fifteen VIRS derived parameters have been used for classification and the clouds were separated into 24 classes using K-Mean classification algorithm. Six out of these 24 classes were found to have high raining probability as well as high cumulative contribution to the total rainfall (~80%).  A regression analysis has been performed to explain the TMI rainfall rate in terms of features derived from VIRS observations for these six classes.  The results of classification and verification have been discussed.