Variational retrieval of temperature and humidity profiles using rain rates versus microwave brightness temperatures

Abstract

AbstractThis paper assesses the performance of two different approaches for the retrieval of temperature and humidity profiles from the satellite passive microwave measurements of the Tropical Rainfall Measuring Mission (TRMM) and of the Special Sensor Microwave Imager in rainy areas. Both methods are based on a one‐dimensional variational retrieval (1D‐Var) approach. The input to the first one is rainfall rates estimated by a standard retrieval algorithm from the raw brightness temperatures. In the second technique, 1D‐Var directly uses the brightness temperatures. The retrieval experiments utilize new simplified physical parametrizations of convection and large‐scale condensation specially designed for variational assimilation as well as a microwave radiative‐transfer model.Several tropical and midlatitude meteorological situations are studied. In all cases, it is found that both 1D‐Var approaches are able to converge satisfactorily and produce consistent temperature and specific‐humidity increments. However, the convergence of 1D‐Var retrievals on surface rainfall rates, and to a lesser extent of 1D‐Var retrievals on brightness temperatures, is reduced when precipitation in the background is produced through convection and not by large‐scale processes. The results of the two 1D‐Var methods are then compared in brightness temperature and rainfall rate space and are validated against independent observations from the precipitation radar on‐board the TRMM. Finally, the sensitivity of the 1D‐Var retrievals to the specification of observation‐error statistics is studied. Copyright © 2004 Royal Meteorological Society.