Evaluation of SMAP Level 2-4 soil moisture products over Michigan, USA

Abstract:

The satellite sensor systems for soil moisture measurements have been continuously evolving. The Soil Moisture Active Passive (SMAP) mission represents one of the latest advances in this regard. So far, much of our knowledge on the accuracy of SMAP soil moisture over the Great Lakes region originated from the evaluation results over the contiguous United States using in situ data from the USDA Natural Resources Conservation Service Soil Climate Analysis Network and/or the U.S. Climate Reference Network, which provided only several in situ sensor stations for this region. As such, these results typically under-represented the accuracy of SMAP soil moisture in this domain. In this work, the SMAP Level 2-4 soil moisture products (SPL2SMAP_S, SPL3SMP_E, and SPL4SMAU) are evaluated over the southern portion of the Great Lakes region using in situ measurements from the Michigan Automated Weather Network. The unbiased root-mean-square error (ubRMSE) values for both surface and root zone SPL4SMAU soil moisture estimates are below 0.04 m³ m^-3 at the 36-km scale. The ubRMSE for SPL4SMAU soil moisture typically ranged from 0.02 to 0.06 m³ m^-3 at the point-scale, with an average error of 0.045 m³m^-3 (0.037 m³ m^-3) for the surface (root-zone) soil moisture. The ubRMSE values for SPL3SMPE a.m. (i.e., from descending orbits) soil moisture retrievals are close to 0.04 m³m^-3 or better at the 36-km scale, with an average ubRMSE of ~0.06 m³m^-3 against the sparse network. The SPL3SMPE p.m. (i.e., from ascending orbits) soil moisture retrievals were slightly less accurate than their a.m. equivalents. The average ubRMSE values were ~0.05-0.06 m³m^-3 for high resolution (3 km and 1 km) SPL2SMAP_S soil moisture retrievals at the point-scale, with the skill of the baseline algorithm-based soil moisture retrievals exceeding that of the optional algorithm-based counterparts. Clearly, the skill of SPL4SMAU surface soil moisture exceeds that of SPL3SMP_E and SPL2SMAP_S soil moisture retrievals. The higher resolution SPL2SMAP_S soil moisture estimates did not present an obvious improvement over the 9-km SPL3SMP_E product. The evaluation results would provide important insights on the application of these products in hydrological and meteorological studies, especially in humid regions.