Study On Land Surface Soil Moisture Retrieval Method With ERS Wind Scatterometer Over Tibetan Plateau

Although soil moisture constitutes only about 0.005 % of the global water resources, it's important as a boundary condition for hydrologic and climate models. Tibetan Plateau is the largest and highest area with complex landform in the world, whose water and energy cycle play a crucial role on the Asia monsoon, East Asia atmosphere general circulation and global climate change. Retrieving nice continuant soil moisture data on Tibetan Plateau is one of the most important work for scientists who focus their studies on the interactions between land and atmosphere of Tibetan Plateau.This study focuses on the development of a consistant methodology for soil moisture inversion in Tibetan Plateau from ERS-1/2 Wind SCatterometer data(WSC). To gain a higher precision of soil moisture data, we firstly introduced a simplified scat-tering(SSM) model calibrated by the latest developed Advanced Integral Equation Model(AIEM) which predicts the backscattering coefficient of random rough surface. In light of the characteristics of WSC data in incident angle, we built the relation between the roughness parameters of SSM from two observations with different incident angles to reduce unknown parameters in retrieving process, thus we could derive the surface soil moisture over bare soil from two simultaneous observation from fore and mid beam of WSC respectively. Secondly, we tried to revise (?)e vegetation effect on the soil moisture estimation. We used the Water-Cloud vegetation model for estimating vegetation effects quantitatively with the reasonablly estimated canopy parameters for each vegetation canopy on Tibetan Plateau. Lastly, we derived the spatial and time series soil moisture data of Tibetan Plateau and the analysis of the results was carried out.To develop our soil moisture retrieval model over bare soil with WSC data, the validity of different theoretical scattering models over random rough surface and empirical and semi-empirical retrieval models over bare soil were investigated with WSC data over Tibetan Plateau. The results indicated AIEM could simulate the surface backscattering with higher precision than other models and exponential autocorrela-tion function would be more suitable than other autocorrelation function when AIEM simulation was conducted over large-scale area and Tibetan Plateau. Because AIEM model is so complex that we cann't use it in soil moisture retrieval without any simplification, we then introduced a simplification scattering model, after calibrated by AIEM model, the correlation function between the roughness parameter "A" of SSM with two different incident angle of WSC were analyzed. The parameter "b", on the other hand, can be regressed by radar incident angle. Thus we can retrieve the bare soil moisture only with two observation data from fore and mid beam of WSC respectively. The above retrieval method performed well in Tibetan Plateau with WSC data, and the results were cross-validated with the measured soil moisture data collected from GAME-Tibet IOP98. The correlation coefficient between the estimated and measured soil moisture reached about 0.6 to 0.7, the stand deviation of estimated soil moisture was of the order of 0.04 cm3/cm3.If we want retrieve land surface soil moisture under vegetation cover, more vegetation .parameter should be known or the unknown parameters should be reduced. Because of the obvious seasonal characteristics of precipitation and frozen-thawed soil, WSC data between Apr. 15th and May 20th of each year in each observation cell was chosen to retrieve parameter "A" of SSM, and these "A" were regressed with incident angle, so we could get the value of parameter "A" of SSM for each observation cell of WSC in each year. Retrieval results with above method in GAME-Tibet experiment sites demonstrated the retrieved soil moisture by two different method, one was the derived A from two observation, the other was from regressed "A", were equivalent, but the derived soil moisture data with regressed A was more smooth with changing time series than those with derived "A" from two observation.When revising the vegetation scattering effects on soil moisture retrieval, we firstly corrected the value of vegetation parameter "b" by the relation of "b" between two different wave length, namely C-band and L-band electromagnetic wave. Secondly, we estimated the vegetation water content(Wc) using AVHRR-NDVI using the empirical relation between Wc and NDVI. At last, we determined the value of vegetation backscattering, and the vegetation effects were explicitly incorporated in our retrieval model.