Comparisons of remote sensing and reanalysis soil moisture products over the Tibetan Plateau, China

The availability of reliable long-term and large scale soil moisture observation datasets over the Tibetan Plateau has been of more concern in recent years. It is therefore necessary to verify any such data to have a better guide to their usage. In this study, we have compared satellite soil moisture retrievals of the Special Sensor Microwave Imager (SSM/I), with ERA-Interim/Land and Global Land Data Assimilation System (GLDAS-1) soil moisture products over the Tibetan Plateau. They are found to be largely different from each other regarding the soil moisture magnitudes in spring, summer and autumn, with those of the ERA-Interim/Land particularly higher than the other datasets in the eastern plateau. The SSM/I and the ERA-Interim/Land data are in good agreement considering their seasonal dynamics, which are however considerably different from the GLDAS-1 models (CLM, Noah, Mosaic and VIC). For interannual variations, it is striking to find strong negative correlations (R < -0.6) between the SSM/I and the ERA-Interim/Land data in spring and autumn, especially over the central plateau. The SSM/I data is found to have also strong negative correlations (R < -0.4) with data of the four GLDAS-1 models in the eastern plateau in spring; and such negative correlations appear in central and western plateau in autumn. In summer, the SSM/I data appears to be positively correlated with the ERA-Interim/Land and all the GLDAS-1 data over the plateau. The negative correlations indicate these data have quite different temporal dynamics for their interannual variations, especially in spring and autumn. A look at the spatial variations revealed that the SSM/I data shows a north-south seesaw pattern in all seasons. The GLDAS-1 Mosaic, Noah and CLM data are able to represent this seesaw pattern principally, which however, can hardly be seen in the ERA-Interim/Land and the GLDAS-1 VIC data. Our results have indicated significant uncertainties in the soil moisture data from remote sensing and reanalysis sources; and we have to use such data with great care.