Remote Sensing Retrieval Of Soil Moisture Based On LSMEM Model In Heihe River Basin

Soil moisture works as a basis of understanding the interchanges of land surface and atmosphere.The measurements of soil moisture in large scale can provide valuable references in agricultural activities,numeric weather forecast,probability of disaster like flood,drought and etc.The high accuracy and large scale distribution of soil moisture can't be obtained readily by the in-situ data separately as the measurements only represents instanteneous point values,by the satellite data from the remote sensing platform and appropriate inversion model,the large scale soil water moisture map can be obtained conveniently.In this thesis,the radiative transfer(RT)model is analyzed and the Land Surface Microwave Emission Model(LSMEM)is introduced and modified.The Advanced Microwave Scanning Radiometer-EOS(AMSR-E)daily grid brightness temperature products of the Heihe river basin are processed.The parameter sensitivities of LSMEM model have been tested,according to the test results,the modification of the vegetation optical depth and single scattering albedo has been implemented,the soil moisture distribution has been obtained by inversion of modified LSMEM model using iterative method.The result shows a continuouslyincrement of soil moisture from northern to southern in the basin,the average volumetric soil moisture ranged from 0.02 to 0.05 in northern desert/semi-desert,the upstream and midstream of southern area have the higher values,and the average is about 0.2,the maximum may reach 0.3.The spatiotemporal consistency has been showed after the comparison between the retrieval results and the in-situ data in Linze and Huazhaizi/Yingke area.In order to do validation of the LSMEM retrieval result in large scale,the Variable Infiltration Capacity(VIC)hydrologic model is introduced,the Pearsoncorrelation coefficient map between these two datasets shows a good positive correlation in most areas of the basin,the maximum coefficient even reaches 0.9.The AMSRE Level 3 soil moisture product,provided by the National Aeronautics and Space Administration,and VUA soil moisture product,provided by Vrije Universiteit,Amsterdam are compared as both of them adapt from the RT model and use the AMSR-E platform.The comparisons indicate these two have the similar soil moisture distribution with the LSMEM retrievals,and the maximum correlation coefficient between LSMEM retrievaland AMSRE Level 3 can reach 1,but the coefficient of LSMEM retrieval and VUA dataset indicate a poor correlation as the VUA dataset has the higher soil moisture and standard deviation.Main innovation points are proposed in this thesis,the first is developing a methodology of calculation the vegetation optical depth and single scattering albedo parameters,the improvement of vegetated area soil moisture is virified,and the second is verification among the VIC simulated soil moisture dataset,the AMSR-E level 3 dataset and the VUA dataset using the correlation analyzing methodology in large temporospatial scale.Reliability of the retrieval has been vefified by comparing the in-situ data and other soil moisture productsin catchment scale,and improving the precision of remote sensing retrieval and getting the moisture of deep soil are still the focus in the further research.