Inferring the impact of radar incidence angle on soil moisture retrieval skill using data assimilation

The impact of measurement incidence angle (θ) on the accuracy of radar-based surface soil moisture (Θ_s) retrievals is largely unknown due to discrepancies in theoretical backscatter models as well as limitations in the availability of sufficientlyextensive ground-based Θ_s observations for validation. Here, we apply a data assimilation-based evaluation technique for remotely-sensed Θ_s retrievals that does not require groundbased soil moisture observations to examine the sensitivity of skill in surface Θ_s retrievals to variations in θ. Application of the evaluation approach to the TU-Wien European Remote Sensing (ERS) scatterometer Θ_s data set over regional-scale (~10002 km²) domains in the Southern Great Plains (SGP) and Southeastern (SE) regions of the United States indicate a relative reduction in correlation-based skill of 23% to 30% for Θ_s retrievals obtained from far-field (θ > 50°) ERS observations relative to Θ_s estimates obtained at θ <; 26°. Such relatively modest sensitivity to θ is consistent with Θ_s retrieval noise predictions made using the TU-Wien ERS Water Retrieval Package 5 (WARP5) backscatter model. However, over moderate vegetation cover in the SE domain, the coupling of a bare soil backscatter model with a "vegetation water cloud" canopy model is shown to overestimate the impact of θ on Θ_s retrieval skill.