A New Approach to Quantifying Both Random Errors and Systematic Climate Regime Biases in TRMM Rainfall Estimates

ABSTRACT
Recent research efforts using Tropical Rainfall Measuring Mission (TRMM) precipitation radar (PR) observations to investigate the effect of climate regime on rainfall estimates from the TRMM microwave imager (TMI) have identified a number of regime dependent biases [Berg et al., 2002; Poyner, 2002]. These biases stem from regional and time-dependent variations in unobserved cloud properties assumed by the retrieval algorithms to be globally unchanging. While random errors in the instantaneous rainfall estimates quickly diminish when averaged over large time and space scales, the systematic component of the retrieval error can have significant consequences for many climate, regional, and data assimilation applications. As an example, it has been determined that time-dependent regional biases are responsible for large differences between TMI and PR estimates of tropical mean rainfall variability associated with the 1997/98 El Nino. To resolve these differences we propose to quantify errors in the TMI and PR rainfall estimates resulting from random and systematic variations in the algorithm assumptions. Because of differences in the nature of the assumptions made in the TMI versus PR retrievals, we propose a two-pronged approach. The first part of this proposal seeks to integrate previous work by the PI and others to assess regional biases in the passive microwave retrievals based upon information available from the PR along with other satellite and in-situ sources. This effort will focus on quantifying the impact of PR observed changes in the structure of precipitation systems on the TMI retrievals. The second part of the proposed work will center on quantifying the impact of regime changes in the assumed drop size distribution (DSD) on PR rainfall estimates via an end-to-end error model. Because global DSD observations are not available, however, we propose to use the vertical and horizontal structure of reflectivity, observable from both ground and space-borne radars, to relate the structure of rainfall systems with changes in DSD. This approach is based on the principle that distinct microphysical processes give rise to distinguishable raindrop size distributions and, at the same time, manifest themselves as differences in the vertical and horizontal structure of observed radar reflectivity profiles. While initially crude, the development of the proposed error model framework will not only provide uncertainty estimates for the TRMM PR, but also a much needed baseline against which the future Global Precipitation Mission (GPM) dual frequency radar combination can be compared.