Regional soil moisture retrievals and simulations from assimilation of satellite microwave brightness temperature observations

Low-frequency microwave satellite observations are sensitive to land surface soil moisture (SM). Using satellite microwave brightness temperature observations to improve SM simulations of numerical weather, climate and hydrological predictions is one of the most active research areas of the geoscience community. In this paper, Yan and Jins' (J Radio Sci 19(4):386-392, 2004) theory on the relationship between satellite microwave remote sensing polarization index and SM is used to estimate land surface SM values from the advanced microwave scanning radiometer-E (AMSR-E) brightness temperature data. With consideration of soil texture, surface roughness, optical thickness, and the monthly means of NASA AMSR-E SM data products, the regional daily land surface SM values are estimated over the eastern part of the Qinghai-Tibet Plateau. The resulting SM retrievals are better than the NASA daily AMSR-E SM product. The retrieved SM values are generally lower than the ground measurements from the Maqu Station (33.85A degrees N, 102.57A degrees E) and the Tanglha Station (33.07A degrees N, 91.94A degrees E) and the US NCEP reanalysis data, but the temporal variations of the retrieved SM demonstrate more realistic response to the observed precipitation events. In order to improve the land surface SM simulating ability of the weather research and forecasting model, the retrieved SM was assimilated into the Noah land surface model by the Newtonian relaxation (NR) method. A direct insertion method was also applied for comparison. The results indicate that fine-tuning the quality factor in the NR method improves the simulated SM values most for desert areas, followed by grasslands, and shrub and grass mixed zones at the regional scale. At the temporal scale, the NR method decreased the root mean square error between the simulated SM and actual observed SM by 0.03 and 0.07 m(3)/m(3) at the Maqu and Tanglha Stations, respectively, and the temporal variation of simulated SM values was much closer to the ground-measured SM values.