The assimilation of satellite data plays an important role in the numerical weather forecast,especially for the region where there is less conventional observations.This study focused on the use of satellite microwave radiances over central Asia,especially on the use of lower sounding channels of microwave sounder over land including Tibetan Plateau and Taklimakan Desert where there is only a few conventional observations covered.To improve the assimilation of microwave sounder radiance over Asia in Global/Regional Assimilation PrEdiction System(GRAPES)at CMA,a land surface emissivity scheme is developed.The land surface emissivity of those window channels is firstly retrieved from the satellite observation.The retrieval is then used in the assimilation of satellite data.It is not only the window channel takes the emissivity retrieval,but also the sounder channel near the window channel gets the retrieved emissivity.The emissivity of database is used in case of the retrieval is failed.It was found that the use of lower sounding channels of microwave sounder over land improves the upstream mid-latitude weather systems and have big impact on the track forecasts of Typhoon.Future plan on the assimilation of satellite data in numerical weather prediction over central Asia will also be presented.