The Megha-Tropiques accumulated rainfall algorithm TAPEER (Tropical Amount of Precipitation with Estimation of ERrors):investigating the error budget of satellite quantitative precipitation estimations

摘要

@@ 1. Introduction Precipitation estimation algorithms based on satellite observations have shown an increase of their overall quality, especially at the meteorologically relevant scales, over the recent decade. The need for a comprehensive error estimate along with the basic rain amount has become a key issue for quantitative precipitation estimation. The TAPEER (Tropical Amount of Precipitation with Estimation of ERrors) algorithm is based on a merging approach of Passive Microwave (PM) instantaneous rain retrievals together with observations from geostationary cloud top sensitive Infrared (IR) channels. This technique relies on the Universally Adjusted GOES Precipitation Index (Xu et al., 1999) and on the PMIR algorithm (Kidd et al., 2003). It is developed for the Megha-Tropiques mission as a tool for a better understanding of the error budget on accumulated rain estimations. TAPEER will provide daily rain accumulations at onedegree resolution over the whole Tropical belt and their associated errors. Roca et al. (2010) developed a sampling error model for satellite accumulated rainfall estimation. It was then adapted to provide global sampling error maps for the TAPEER algorithm. While sampling errors are often assumed to have the most significant contribution to the error budget (Bell et al., 1990), algorithmic issues can also lead to significant bias and random errors (Xu et al., 1999). Therefore they need to be quantified. This study shows a methodology developed as a first step to derive algorithmic errors for satellite precipitation estimation algorithms. A novel approach to quantitative precipitation verification, taking into account the errors on both satellite and surface rain datasets when comparing the estimates, is then used as a tool to infer the importance of each component of the budget to the total error.