Key points are not available for this paper at this time.
In this study, we inter-compare soil moisture from in situ measurement, reanalysis data (ERA-interim), land data assimilation system simulations (the Global Land Data Assimilation System, GLDAS) and two satellite remote sensing retrievals: L-band products from Soil Moisture Ocean Salinity (SMOS) and C-band products from the Japan Aerospace Exploration Agency Advanced Microwave Scanning Radiometer-Earth Observing System (AMSR-E). The stationaveraged surface soil moisture data, measured during May to September 2010, from the CEOP Mongolia network are used as “ground truth”. Major findings are: (1) from the point view of root mean square error (RMSE), the accuracy of the remote sensing products is clearly higher than the ERA-interim and GLDAS. AMSR-E has the smallest RMSE (0.032), while the highly-expected SMOS has an RMSE of 0.065, larger than the mission requirement (RMSE<0.04). Both GLDAS (RMSE=0.132) and ERA-interim (RMSE=0.115) evidently overestimate soil moisture. (2) According to the correlation coefficient (R), ERA-interim has the highest one (0.77), and next came AMSR-E (0.47), GLDAS (0.06) and SMOS (0.04), indicating that both GLDAS and SMOS fails to capture the soil moisture temporal dynamics. Our results reveal that the remote sensing product still need further develop, for both C-Band algorithm (AMSR-E) and Lband one (SMOS). The coincident of high R of ERA-interim and low RMSE of AMSR-E implies a potential for integration within a land data assimilation system.
Wen et al. (Sat,) studied this question.