Microwave retrievals of soil moisture and vegetation optical depth with improved resolution using a combined constrained inversion algorithm: Application for SMAP satellite

Abstract

A new algorithm called combined constrained multi-channel algorithm (C-CMCA) is presented for simultaneous retrieval of soil moisture (SM) and vegetation optical depth (VOD) in L-band with improved resolution. Unlike widely used algorithms, the new approach optimally fuses multiple sources of surface temperatures into the inversion process and confines the retrievals to their feasible climatological range rather than the mean and paves the way to account for the slow changes of VOD through a Sobolev-norm regularization. Through controlled numerical experiments that assume a random error in the surface temperatures, it is shown that the algorithm can decrease the root mean squared error (RMSE) by 78% and 81% when compared with the unconstrained version and 54% and 7% when a single source of surface temperature is used in retrievals of SM and VOD, respectively. The use of the Sobolev-norm regularization decreases the RMSE by more than 25% at the expense of a negligible bias. Implementation of the algorithm, using data from the NASA's Soil Moisture Active Passive (SMAP) satellite in 2016, demonstrates that the monthly RMSE of SM retrievals improves by more than 6% when compared with the SMAP enhanced products considering the ground measurements from the International SM Network (ISMN) as a reference while the monthly RMSE of VOD decreases by more than 62% when compared with the VOD climatology used in the SMAP SCA products. Analysis of the results demonstrates that, without increasing the native resolution of radiometric observations, the information content of the a priori constraints cannot only improve resolutions of the retrievals but also make them robust to the background water contamination in the vicinity of coastal zones and over lowland floodplains. For example, over Florida peninsula, the annual SM RMSE and bias are reduced by more than 65%.