干旱区生态环境知识资源中心

Monitoring surface soil moisture (SSM) variability is essential for understanding hydrological processes, vegetation growth, and interactions between land and atmosphere. Due to sparse distribution of in-situ soil moisture networks, over the last two decades, several active and passive radar satellite missions have been launched to provide information that can be used to estimate surface conditions and subsequently soil moisture content of the upper few cm soil layers. Some recent studies reported the potential of satellite altimeter backscatter to estimate SSM, especially in arid and semi-arid regions. They also pointed out some difficulties of such technique including: (i) the noisy behavior of the backscatter estimations mainly caused by surface water in the radar foot-print, (ii) the assumptions for converting altimetry backscatter to SSM, and (iii) the need for interpolating between the tracks.

In this study, we introduce a new inversion framework to retrieve soil moisture information from along-track altimetry measurements. First, 20Hz along-track nadir radar backscatter is estimated by post-processing waveforms from Jason-2 (Ku-and C-Band during 2008-2014) and Envisat (Ku-and S-Band during 2002-2008). This provides backscatter measurements every similar to 300m along-track within every similar to 10 days from Jason, and every similar to 35days from Envisat observations. Empirical orthogonal base-functions (EOFs) are then derived from soil moisture simulations of a hydrological model, and used as constraints within the inversion. Finally, along-track altimetry reconstructed surface soil moisture (ARSSM) storage is inverted by fitting these EOFs to the altimeter backscatter. The framework is tested in arid and semiarid Western Australia, for which a high resolution hydrological model (the Australian Water Resource Assessment, AWRA model) is available. Our ARSSM products are also validated against Soil Moisture and Ocean Salinity (SMOS) L3 products, for which maximum correlation coefficients of bigger than 0.8 are found. Our results also indicate that ARSSM can validate the simulation of hydrological models at least at seasonal time scales. (C) 2017 Elsevier Inc. All rights reserved.