With the large volume of satellite remote sensing data of the earth terrestrial surface becoming available, precisely monitoring the dynamics of the land surface state variables for agricultural and land use management becomes possible. Currently, the moderate resolution imaging spectroradiometers on board NASA’s Earth Observing Satellites (EOS) Terra and Aqua make it possible to derive a global coverage of land surface vegetation indices, leaf area index, and surface temperature data products at 1 km spatial resolution every day. The advanced microwave scanning radiometers (AMSR) on board Aqua and Japan's ADEOS satellites start sending back a global coverage of rainfall and land surface soil moisture data products at up to 25km spatial resolution every two to three days. It is also well known that these land surface remote sensing products contain uncertainties due to imperfect instrumentation calibration and inversion algorithms, geophysical noise, representativeness error, communication breakdowns, and other sources while land surface model can continuously simulate these land surface state or storage variables for all time steps and all covered areas. Therefore a combination of satellite remote sensing products and land surface model simulations may provide more continuous, precise and comprehensive depiction of the dynamics of the land surface states. This paper introduces the state-of-the-arts technologies in the development of NASA's Land Data Assimilation System, and then proposes a procedure to combine the simulations of a simple land surface model and the remote sensing products from MODIS and AMSR. After the results of testing the procedure for an arid area in Southwest USA are presented, the application of the procedure for the oases in Fukang Count of Xinjiang Autonomous Region is proposed.