IMPROVING EXTREME RAINFALL EVENT PREDICTION USING MICROWAVE SATELLITE DATA ASSIMILATION

Abstract

Microwave Satellite data has become very useful in weather prediction through data assimilation for improving Numerical Weather Prediction (NWP) initial conditions. By assimilating 23GHZ (sensitive to water vapor) and 89GHZ (sensitive to cloud water) brightness temperature, the predictability of an extreme rain weather event over Lake Victoria, East Africa is investigated. The assimilation of AMSR-E brightness temperature (TB) through a Cloud Microphysics Data Assimilation (CMDAS) into a NWP considerably improves the spatial distribution of this event. Assimilating cloud water, water mixing ratio and other ice cloud components inserts clouds, triggers weak surface convergence and consequent convection. The spatial distribution of the simulated event is comparable to observed satellite rainfall. Through assimilation, smaller events missed by the downscaling experiment are introduced into the model state and this leads to a better forecast. The integrated cloud condensate follows a similar pattern to observed cloud top temperature with the probability of detection improving as well. However, even though the system is able to reproduce a reasonable distribution of this event, there still remains overestimation especially over the regions of maximum precipitation.