Triple Collocation Analysis of Soil Moisture From Metop-A ASCAT and SMOS Against JRA-55 and ERA-Interim

Abstract

This study investigates the quality of Advanced Scatterometer (ASCAT) surface soil moisture (SSM) retrievals with respect to other SSM products derived from the passive Soil Moisture and Ocean Salinity (SMOS) mission and two reanalysis datasets, i.e., the JRA-55 and the ERA-Interim. In particular, the purposes of this study are to 1) characterize the global error structure of the satellite products, 2) understand the spatiotemporal variability of SSM at global scale, and 3) investigate in which areas the assimilation of satellite data may add value to reanalysis. For these purposes, we applied standard statistical methods as well as triple collocation analysis (TCA) for estimating signal-to-noise ratios (SNR). In line with previous studies, we find large and spatially variable biases between all four datasets, but overall spatiotemporal dynamics as reflected in Hovmoller diagrams agree well. With the exception of arid and semiarid environments, ASCAT performs better than SMOS in terms of both its correlation with the models and the SNR. As a result of TCA, we recognize the potential areas for assimilation of ASCAT data, characterized by a high SNR of the satellite data compared to the models, to be the savanna regions in Africa and Central Asia, southwestern North America, and eastern Australia.