Evaluation of SMAP Soil Moisture Relative to Five Other Satellite Products Using Ground-Based Observations over the United States

Satellite platforms provide a unique opportunity to retrieve soil moisture over large areas. The Level 3 (L3SMP) and Level 4 (L4SMP) products of the latest soil moisture satellite mission, SMAP, are evaluated relative to five other satellite products. Results are dependent on the representativeness of the observed data, and in this presentation, we discuss how subpixel precipitation variability and in situ data source affect the perceived performance of satellite products.

The satellite platforms are evaluated against the Climate Reference Network (CRN) and the Cosmic Ray Soil Moisture Observing System (COSMOS) ground-based networks, each with around 100 in situ stations over CONUS. The SMAP products as well as Aquarius clearly outperform the other products. Over all land cover types, L3SMP and L4SMP are better in JJA than DJF and they perform best over short vegetation. L4SMP is better than L3SMP product over tall and short vegetation whereas L3SMP has higher correlation compared with CRN soil moisture over crops.

Just 14% (21%) of L4SMP stations meet the SMAP mission requirement of the unbiased error (ubRMSE) less than 0.04 m³m^-3 in DJF (JJA) compared with CRN, primarily due to the representativeness of the station mainly caused by the subpixel precipitation variability. To partially account for this mismatch, a relaxed criterion for ubRMSE less than 0.06 m³m^-3 is chosen, and L4SMP meets this criterion at much higher percentages of stations (~ 45% in DJF and JJA). When stations with vegetation water content greater than 5 kg m^-2 are removed, L4SMP meets the criterion at more than half of the stations.