Validation and improvements of brain precipitation satellite estimator using CHUVA field campaign dataset and passive microwave radiometer data in the Megha-Tropiques satellite context

Abstract

The present proposal deals with the development and improvement of rain retrieval over Brazil using a combination of passive microwave radiometer data (completed with infra-red data from geostationary satellite. The base tool will be the BRAIN (Bayesian Rain retrieval Algorithm Including Neural network). This algorithm was developed initially for tropical rain retrieval using any of the passive microwave radiometer mentioned above. BRAIN is a Bayes/Monte-Carlo method that relies on an a-priori retrieval database to reduce the world of possible solutions to those physically meaningful. From the existing algorithm, a number of studies will be done as follow: Retrieval database building using a combination of TRMM PR and TMI a retrieval database specifically designed for Brazil and its various regions will be set up. Implementation of a first version of the retrieval and performance assessment using CHUVA data. An error assessment over the various region of Brazil will be made. Address the warm rain problem. The specific case of warm rains will be looked upon. This problem is particularly critical for some of the Brazilian regions and is also unfortunately extremely difficult to solve using PMR data over land. The proposed study will take place over a period of 6 month, extending from August 2012 to December 2012. (AU)