Rainfall estimation by inverting SMOS soil moisture estimates: A comparison of different methods over Australia

Remote sensing of soil moisture has reached a level of maturity and accuracy for which the retrieved products can be used to improve hydrological and meteorological applications. In this study, the soil moisture product from the Soil Moisture and Ocean Salinity (SMOS) satellite is used for improving satellite rainfall estimates obtained from the Tropical Rainfall Measuring Mission multi-satellite precipitation analysis product (TMPA) using three different “bottom up” techniques: SM2RAIN, SMART and API-mod. The implementation of these techniques aims at improving the well-known “top down” rainfall estimate derived from TMPA products (version 7) available in near real time. Ground observations provided by the Australian Water Availability Project (AWAP) are considered as a separate validation dataset. The three algorithms are calibrated against the gauge-corrected TMPA re-analysis product, 3B42, and used for adjusting the TMPA real-time product, 3B42RT, using SMOS soil moisture data. The study area covers the entire Australian continent and the analysis period ranges from January 2010 to November 2013. Results show that all the SMOS-based rainfall products improve the performance of 3B42RT, even at daily time scale (differently from previous investigations). The major improvements are obtained in terms of estimation of accumulated rainfall with a reduction of the root mean square error of more than 25%. Also in terms of temporal dynamic (correlation) and rainfall detection (categorical scores) the SMOS-based products provide slightly better results with respect to 3B42RT, even though the relative performance between the methods is not always the same. The strengths and weaknesses of each algorithm and the spatial variability of their performances are identified in order to indicate the ways forward for this promising research activity. Results show that the integration of “bottom up” and “top down” approaches has the potential to improve the quality of near real-time rainfall estimates from remote sensing in the near future.