Downscaling Of SMOS Soil Moisture Using Sentinel-SAR And Modis Data

Soil moisture is an important part of the"four water".Several microwave remote sensing satellites continue to provide wide spatial coverage and high temporal resolution of surface soil moisture,but the coarse resolution of surface soil moisture of microwave products is difficult to meet the hydrologic watershed hydrological simulation and farmland scale drought monitoring accuracy requirement and lack of a wide range of deep soil moisture data.The radar and optical surface information data can provide high spatial inversion of soil moisture.According to the characteristics of radar data and optical data,with Daqing River Basin as the study area,collected of the Sentinel-1 data,the optical data,SMOS L2surface soil moisture products（0-5 cm）,the surface measured data and the terrain data,extract the surface environment of the factors from 2016-2019.Based Random Forest algorithm to build a downscaling model,the research got high spatial resolution.Then,based on the SWI model of the iterative index algorithm,the surface soil moisture was used to calculate the root zone soil moisture（0-50cm）,and the accuracy was evaluated.The main conclusions of the study are as follows:（1）The R language programming was used to calculate the contribution of environmental factors in each table.The contribution of the backscattering coefficient representing the radar factor was as high as 0.25,indicating that the backscattering coefficient⁰_vv had a high correlation with SMOS soil moisture.The contribution of drought factors such as temperature vegetation drought index TVDI and apparent thermal inertia ATI was also higher than 0.1.The results confirmed the effectiveness of radar factors and drought factors in soil moisture monitoring,and the fitting relationship between optical indexes such as LST and NDVI and SMOS soil moisture was good,which was conducive to the construction of a framework for soil moisture retrieval by multi-source factors.（2）Based on random forest algorithm synergy microwave and optical remote sensing data of SMOS downscaling surface soil moisture,and build a good training model and the forecast of validation set,and the SMOS raw data fitting precision,R range of 0.900-0.998,and ub RMSE range of 0.002-0.052 cm³/cm³.The soil moisture data after downscaling with site measured data correlation is good（R²>0.804,the RMSE<0.028 cm³/cm³）.The soil moisture in the Daqing River Basin with a spatial resolution of 1km can reflect the regional differences well,and the results show that the proposed algorithm is suitable for obtaining the surface soil moisture with a medium and high spatial resolution.（3）Based on the downscaling surface soil moisture,SWI model suitable for Daqing River Basin was constructed to estimate the deep soil moisture,which was generally consistent with the spatial distribution of CLDAS soil moisture data based on the measured data assimilation.The spatial resolution of the inversion of deep soil moisture was higher,showing more spatial differences of soil moisture.By comparing the inversion results with the measured data,the median correlation coefficient is 0.81,which indicates that the deep soil moisture products are in good agreement with the ground observation data,and the change trend of the time series of surface and root zone soil moisture is consistent,and there is a strong coupling relationship.