An estimation model of soil moisture from the AMSR\|E brightness temperature data was coupled into the WRF\|Noah mesoscale numerical model for improving the observation and model simulation ability of soil moisture over Northeast Plateau. The main condusions are as follows: Firstly the land surface soil moisture(SM) in the northeastern part of the Qinghai-Tibet Plateauwas estimated by the AMSR-E brightness temperature data. Its value was somethinglower than the observedand the NCEP reanalysed data, but regional distribution was reasonable and the temporalvariation showed a quick response to the daily precipitation. Sencondly,the estimated SM was assimilated into the WRF-Noahmodel byusingthe Newtonianrelaxation assimilation method. By fine tuning of the quality factor inthe Newtonianrelaxationmethod, the simulated SM valueswere obviouslyimproved in desert areas, somewhat improvements also occurred grassland, shruband grass mixed zones. Ontemporal scale, the difference between the simulated and observed SM were alsodecreased.