Research On Method For Radar Retrieval Of Soil Moisture On Vegetation Covered Surface

Soil moisture is an important parameter affecting agricultural production,hydrological process and climate system,and its temporal and spatial distribution information plays a key role in monitoring soil moisture,guiding irrigation operations,and predicting crop yields.However,due to the high spatial variabiliaty of soil moisture,traditional ground measurement methods cannot meet the needs of large-scale monitoring.With the development of remote sensing technology,it is possible to monitor soil moisture spatiotemporally at various scales.Passive microwave remote sensing technology has been successfully used to monitor soil moisture,but the spatial resolution of passive microwave remote sensing soil moisture products is relatively low,usually at the level of tens of kilometers.Because of its high spatial resolution,synthetic aperture radar has attracted much attention in obtaining high spatial resolution soil moisture information.Compared with passive microwave,the backscattering obtained by radar is more easily affected by surface geometric information such as vegetation structure characteristics and soil roughness,which has always been a difficult and hot issue in the research of soil moisture retrieval by radar.In this paper,ground-based,airborne,and spaceborne radar data and ground measured data are the main data sources,and the research is carried out in the Soil Moisture Experiment in the Luan River in China,the Soil Moisture Active Passive Validation Experiment 2012 in Canada and Cloud and Land Surface Interaction Campaign in the United States.Based on the existing rough surface scattering model and vegetation scattering model,the vegetation and roughness correction problems in soil moisture retrieval are studied,and soil moisture retrieval methods suitable for different radar data characteristics are constructed.The main research work of this paper is as follows:(1)In order to study the correction of surface roughness in the process of soil moisture retrieval,and to discover the ground feature information contained in multi-angle observations,this paper develops a soil moisture retrieval method based on multi-angle radar data according to the characteristics of ground-based radar data.Firstly,this paper analyzes the correlation between the backscattering coefficient and soil moisture at different incident angles and retrieves soil moisture by combining radar data with different polarization and different incident angles.The results of soil moisture retrieval with HV polarization in the C band and VV polarization in the L band are the best.The correlation coefficients of the retrieval results are 0.586 and 0.687 respectively,and the ub RMSE are 0.073 and 0.067 cm~3/cm~3 respectively.Secondly,the retrieval results of low,medium and high incident angles are analyzed.The influence of different incident angles on the retrieval results is analyzed,and it is found that the HV polarization retrieval accuracy in the C band is the highest at low incidence.At a low incidence angle,the VV polarization retrieval accuracy in the L band is the highest,and at medium and high incidence angle,the HV polarization retrieval accuracy in the L band is the highest.In addition,the SMAP 9 km soil moisture product is used to limit the retrieval process,and the ub RMSE of the final soil moisture retrieval result is 0.03 cm~3/cm~3,which proves that when using multi-angle data to retrieve soil moisture,the retrieval accuracy can be improved by passive microwave data(2)In order to explore the problem of vegetation correction in the process of soil moisture retrieval,realize the quantitative description and parameterization of vegetation based on multi-polarization data,and overcome the problem that vegetation information obtained by optical observation is easily affected by clouds and rain in the traditional retrieval method.Based on airborne radar data,this paper establishes a soil moisture retrieval method relying on radar vegetation index.Firstly,the sensitivity of three radar vegetation indices to the measured vegetation water content is analyzed,and the feasibility of using radar vegetation index is confirmed.Secondly,by comparing the backscattering coefficient simulated by the model with the backscattering coefficient observed by radar,it is found that VH polarization observation has the best fitting accuracy in the parameterization process of water cloud model.The final results show that the ub RMSE values of soil moisture retrieval results of rape,corn,soybean and wheat are 0.056,0.053,0.098 and 0.079 cm~3/cm~3,respectively,using VV polarization backscattering coefficient and VH polarization observation.The research proves that multi-polarization radar observation can be used to reflect vegetation change and parameterize vegetation influence,and it is also found that the parameters of water cloud model change with vegetation growth in different vegetation growth stages.This study provides a new idea for retrieving soil moisture from L-band radar data.(3)In order to solve the“ill-conditioned retrieval”and“multi-solutin problems”affected by vegetation and roughness in the process of soil moisture retrieval,this paper makes full use of the high time revisit ability of the current space-borne SAR(Sentinel-A/B),and constructs a multi-temporal dual-channel(MTDC)soil moisture retrieval method based on spaceborne radar data.Firstly,aiming at the problem that the incident angles of multi-track radar data are inconsistent,this paper innovatively proposes an angle normalization method based on time series,which corrects different ground tracks data to the same incident angle level while retaining the original backscattering(soil moisture)change information.Secondly,aiming at the problem that multiple unknown parameters(ill-conditioned retrieval)cannot be retrieved simultaneously,according to the time-varying characteristics of vegetation and roughness,it is assumed that they do not change in the retrieval window,thus effectively reducing the number of parameters to be retrieved.Besides,considering the strong sensitivity of radar backscattering to surface roughness and the interference to the iterative retrieval process,taking advantage of the relatively stable liquid water content during the freezing period,the freezing period is used to obtain parameters such as surface roughness and initial liquid water content preferentially.Finally,by setting a certain time window,the sliding retrieval is carried out by using multi-temporal dual-channel data,The comparison between the ground measured soil moisture data and the SMAP soil moisture product data shows that the ub RMSE of the spaceborne soil moisture retrieval results are 0.066 and 0.048 cm~3/cm~3,respectively,which confirms the role of multi-temporal information in spaceborne radar soil moisture retrieval and the feasibility of developing radar regular soil moisture products.In summary,based on the characteristics of radar data carried on different platforms(ground/multi-angle,airborne/multi-polarization,spaceborne/multi-temporal),this paper studies the correction methods of surface roughness and vegetation parameters.And a series of radar soil moisture retrieval methods for vegetation-covered surface is developed,which characterizes that improving the dimension of observation data is an effective means to improve the retrieval ability,and provides a reference for the subsequent operation of radar soil moisture products.