The SMOPS Blended Soil Moisture Climate Data Record

Soil Moisture (SM) is a vital state variable of land surface in hydrological, meteorological and climatological studies as it impacts energy, carbon and water interactions in the boundary layer. To make effective use of all available microwave-based datasets, our research team has developed and been maintaining a SM Operational Product System (SMOPS) at National Oceanic and Atmospheric Administration (NOAA)-National Environmental Satellite, Data, and Information Service (NESDIS) since 2012. What makes the SMOPS product unique is that it provides the 6-hourly and daily global blended SM data products with 3-hour and 24-hour latency, respectively. However, the lack of long-term consistent SMOPS climate data record (SMOPScdr) means data quality varies significantly among the different versions, resulting in uncertainties of using multi-year SMOPS blended data for climatological studies and long-term data assimilations. To bridge this gap and meet the data requirements of our users, we developed the SMOPScdr by reprocessing and in turn combining all available microwave satellite SM observations since 2002.

The SMOPScdr data product takes advantage of three advances: 1) refinement of the currently operational Advanced Microwave Scanning Radiometer-2 (AMSR-2) SM retrievals using the optimal machine learning models. The refined AMSR-2 SM retrievals (AMSRr) show an overwhelming advantage in data accuracy over the currently operational AMSR2 product, while the AMSRr datasets are comparable with the latest version SM Active and Passive (SMAP) observations. As an important component of the SMOPS blended product, higher quality AMSR-2 SM retrievals enhance the data accuracy of the developed SMOPScdr. 2) Calibration of the multifrequency AMSR for EOS (AMSR-E) brightness (Tb) observations toward the AMSR2 measurements through building the ascending and descending calibration models at pixel level using the Simultaneous Conical Overpass (SCO) method. Compared to other commonly-used calibration strategies, results for the SCO method indicate that AMSR-E Tb observations are more successfully scaled to AMSR-2 datasets, which allows to generate long-term consistent AMSR-E/-2 SM data by implementing the trained AMSR-2 machine learning model, and eventually benefits the SMOPScdr. 3). Bias-correction of the individual satellite SM data products toward the SMAP observations. The climatological patterns and dynamical characteristics of the individual satellite SM retrievals vary significantly from each other. The SMAP data product was used as a benchmark for the blending procedure, which makes the SMOPScdr depend solely on readily available satellite observations.

In this study, the comprehensive validations were conducted to evaluate the developed SMOPScdr. With respect to the SM measurements from 295 North American Soil Moisture Database sites, SMOPScdr shows better accuracy in comparison with the individual satellite SM datasets and much higher spatial and temporal coverage. Results also indicate that the developed SMOPScdr is comparable with the reference SMAP observations.