Estimation of surface soil moisture from onboard FY-2Dsatellite multi-temporal data

Surface soil moisture (SSM) is a critical element of the hydrologic processes that influences exchange of water and energy fluxes at the land/atmospheric interface. Current remote sensing applications in SSM studies are largely limited to polar orbiting satellites. With the development of new generation geostationary satellites such as MSG and GOES-O&P, land surface visible and thermal infrared data can be acquired with high spatial and temporal resolutions. Consequently, great opportunities exist to analyze land surface soil moisture with retrieval methods of satellite-observed data. Fengyun-2D is a Chinese operated geostationary satellite with one visible and four infrared channels of optical imaging radiometer with temporal image acquisition frequency of 30 min. This al-lows mapping diurnal variations in land surface shortwave radiation (SSR) and land surface temperature (LST). The objective of this study was to estimate SSM from diurnal evolutions of SSR and LST. FY-2D data (with a spatial resolution of 5 km) were download along with advanced mechanically scanned radiometer (AMSR) soil moisture products (with spatial resolution of 25 km). The sets of data were geographically corrected via geo-referencing using geo-locational tools. The thermal and visible infrared soil moisture data products from FY-2D and AMSR were matched using a linear resampling method. Next, an algorithm for estimating SSM via two thermal infrared channels (IR1: 10.3~11.3 m and IR2: 11.5~12.5 m) and one visible channel (0.55~0.9 m) of the geostationary satellite data was proposed based on linear relationship between SSR and LST diurnal evolutions. Finally, the method was validated using FY-2D and AMSR SSM data products for September 30, 2010 and SSM estimated using FY-2D data for October 20, 2010. The results showed that the method was applicable in calculating SSM. SSM correlation based on analysis of FY-2D and AMSR was 0.52, root mean square error (RMSE) of 0.025 g·cm^-3 and maximum errors < 0.07 g·cm^-3. The proposed method was easy use and output mid-scale SSM that improved the spatial resolution of AMSR SSM products from 25 km to 5 km. However, it was noted that the method must be used with care as it was prone to severe error under vegetation or cloud cover.