Intercomparison of Alternate Soil Moisture Downscaling Algorithms Using Active–Passive Microwave Observations

Abstract

Three active-passive soil moisture downscaling algorithms are tested to demonstrate the feasibility of each for application to NASA's Soil Moisture Active Passive (SMAP) mission launched in January 2015. These algorithms include the official baseline and optional downscaling algorithms, and a change detection method. These synergistically use 1-km synthetic aperture radar backscatter to downscale 36-km brightness temperature to 9 km, which is then converted into soil moisture at 9 km, or downscale soil moisture directly to 9-km resolution. While these algorithms have been tested previously, this was mostly using synthetic data sets. Moreover, there has never before been a direct comparison of the alternate methods using the same data sets. Thus, it is imperative that they be tested against each other for a comprehensive range of land surface conditions prior to global application. Consequently, this letter evaluates these three algorithms using data collected from the soil moisture active passive experiments (SMAPExs) in Australia, designed to closely simulate the SMAP data stream for a single SMAP radiometer pixel over a three-week interval. Results suggested that the average root-mean-square error (RMSE) in downscaled soil moisture at 9-km resolution was 0.019, 0.021, and 0.026 cm 3 /cm 3 for the baseline, optional, and change detection method, respectively. While there was a little difference in the RMSE, the optional method showed the best correlation between the downscaled soil moisture and the reference soil moisture map. Therefore, the optional algorithm is recommended for global implantation by SMAP.