Abstract
Soil moisture often governs the exchange of water between the land surface and the atmosphere and, as a result, has a profound effect on the global water and energy cycles. With the launch of the two new L-band satellite missions tasked with retrieving surface soil moisture (i.e., Soil Moisture and Ocean Salinity (SMOS) and Soil Moisture Active/Passive (SMAP) missions), new possibilities exist for quasi-global soil moisture monitoring. Here, new soil moisture estimates (CLDAS-BPNN), generated by fusing a land surface model soil moisture product (CLDAS) using in-situ observations via a Back Propagation Neural Network (BPNN) method, are used to evaluate four satellite-derived soil moisture products (SMOS-L3, SMOS-IC, SMAP_L3_SM_P, and SMAP_L3_SM_P_E) over the highly irrigated Huai River Basin in China. Results indicate that the post-processing of CLDAS (to generate CLDAS-BPNN) reduces soil moisture errors (particularly bias) and preserves temporal correlations with in-situ observations. Due to extensive irrigation present in the basin, CLDAS-BPNN soil moisture time series have weak seasonal differences and the spatial distribution of their means does not reflect known precipitation patterns. Based on the validation of SMOS and SMAP soil moisture retrievals against CLDAS-BPNN, several key findings can be obtained. First, SMAP retrievals are recommended for locations possessing both valid SMOS and SMAP retrievals. However, SMOS provides generally better spatial support than SMAP. Second, the four satellite retrievals present no significant seasonal variation in time-varying errors, and larger soil moisture underestimation is observed in winter than in summer. Finally, all four satellite products exhibit degraded accuracy in mountainous and forested areas of the Huai River Basin relative to flatter terrain in the Huaibei plain.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
