| Surface soil moisture plays an important role in controlling water and energy cycles at the land-atmosphere boundary.In addition,accurate surface soil moisture monitoring and prediction play a vital role in crop growth studies,flood and drought event prediction,hydrological research and global change research.Radar remote sensing is considered as one of the most promising methods for soil moisture retrieval.However,radar remote sensing has obvious incident angle effect,that is,radar echo is strong at short distance,and will gradually weaken with longer distance.This may affect the accuracy of soil moisture retrieval.How to accurately correct the radar incidence angle effect has great research value.This research is supported by the national key research and development program of China"Development of Resource and Environmental Carrying Capacity Measurement System for Rural Construction"(No.2018YFD1100100)and the National Natural Science Foundation of China"Research on collaborative Inversion algorithm of Optical and Radar Remote sensing for soil moisture and surface roughness"(No.41971323).The paper introduces the development status of radar sensors in detail,summarizes the current normalization method of radar incident angle,and condenses the electromagnetic wave radiation transfer theory and the method of radar remote sensing inversion of soil moisture.On this basis,according to the electromagnetic wave propagation theory,the incident angle effect of the radar signal under the state of bare soil and crop-covered surface is simulated,and the correction of the incident angle effect of the crop radar signal is carried out in depth.The difference of the incident angle effect of radar signals of different crop types is analyzed,and the newly proposed dynamic cosine model is finally applied to the field of soil moisture inversion,which improves the accuracy of soil moisture inversion.The main research results are as follows:(1)Based on the microwave scattering model of bare surface(IEM and Oh models)and the microwave scattering model of vegetation covered surface(WCM),the radar signals under different surface conditions are simulated,and the incident angle effect of radar signals under different surface conditions is analyzed.The radar signals of bare land and vegetation cover have incident angle effect,and the radar signals of VV polarization have more significant incident angle effect.With the increase of vegetation index(NDVI),the effect of radar incidence angle gradually weakens.(2)The dynamic cosine model and performance evaluation based on vegetation index are realized.In this study,we found that vegetation indice has great potential in representing the dynamic change of N value.In this study,three typical vegetation indices(NDVI,EVI and SAVI)and four fitting methods(linear,logarithmic,exponential and quadratic polynomial function)were selected to determine the optimal vegetation index representing the dynamic N value and the optimal fitting relationship between the optimal vegetation index and N value.The results show that SAVI can better simulate the dynamic change of N value,and the exponential fitting relation is optimal under both VV and VH polarization.Compared with first-order cosine model(1.37 d B)and cosine squared model(1.08 d B),the average RMSE of radar signal corrected by the newly developed dynamic cosine model is 0.47 d B,which is 66%and56%higher than that of first-order cosine model(1.37 d B)and cosine squared model(1.08 d B),respectively.(3)There are obvious differences in incidence angle effect of radar signal among maize,soybean and rice.The radar incidence angle effect of maize and rice is strong,but that of soybean is relatively weak.The high accuracy correction of radar incidence angle effect needs to consider the influence of radar polarization mode,crop growth stage and crop type.(4)The sensitivity of radar signal to soil moisture was significantly improved(R~2=0.57)by using the new dynamic cosine model to correct incident angle effect.The sensitivity of original radar data,radar data corrected by first-order cosine model and radar data corrected by cosine square law model to soil moisture were 0.41,0.47 and0.52 respectively.The soil moisture inversion accuracy of radar signal after dynamic cosine model correction is improved,and the soil moisture inversion result(R~2=0.604,RMSE=0.019 cm~3/cm~3).The dynamic cosine model based on vegetation index proposed in this study can not only serve the field of soil moisture retrieval,but also has great potential in the field of visual interpretation of SAR images,crop classification and quantitative inversion of land surface parameters. |
