Soil Moisture Retrieval Of Field Parcel In Karst Mountainous Area Based On Sentinel-1 SAR Data

Soil moisture is an important parameter and component of global water cycle and land water cycle,and plays an important role in crop yield estimation,drought monitoring and refinement of cultivated land.In karst mountainous areas,due to the fragmentation of cultivated land,data loss of multi-temporal optical image caused by cloud and fog interference,soil water inversion is susceptible to interference of complex environment,etc.To solve the above problems,active microwave remote sensing has been widely used in soil water inversion due to its advantages of high resolution,immunity from cloud and fog,and strong surface penetration.Based on sentinel-1 SAR images and field data in April,June and July 2021,soil moisture of cultivated land in Guanling County,Anshun City,Guizhou Province was retrieved.In order to avoid the influence of mixed pixels on the accuracy of soil inversion in karst mountainous area,the block scale was taken as the constraint condition.According to the actual situation of the cultivated land in the study area,the soil water inversion is divided into two parts by using vegetation water content to distinguish whether the cultivated land is bare soil or not.On bare land or sparse vegetation covered land,the functional relationship between soil backscattering coefficient and soil moisture and soil roughness is established by using AIEM model database.Vegetation in vegetation covered surface,selecting different radar parameters,using water clouds model to remove the influence of vegetation at the same time compared to the radar vegetation parameters inversion precision of soil moisture,based on the optimum parameters of the vegetation in the study area in the study area is calculated after the final soil to the scattering coefficient,input parameters into the SVR model in the study area of cultivated land plot soil moisture inversion.The spatial and temporal distribution characteristics of soil water in cultivated land were analyzed in combination with planting structure,grade of rocky desertification,topography and soil physical properties.The main research contents and results are as follows:(1)In view of the exposed surface in order to improve the inversion accuracy calibration of correlation length l study choose composite roughness Z_s=s~3/l~2 as soil roughness parameters,using the AIEM model to simulate the backward scattering field of soil,soil roughness coefficient and a function of soil moisture after the soil moisture and soil is deduced to the scattering coefficients of the quadratic equation.R~2 and RMSE were 0.6684 and 0.0857 cm~3/cm~3 respectively.(2)Four radar vegetation indices(CR,PRVI,RVI and Dp SVI)were selected to retrieve soil moisture of cultivated land based on water cloud model and SVR model.The results show that the accuracy of radar vegetation index inversion is PRVI>Dp SVI>RVI>CR.The inversion accuracy R~2 of April,June and July are 0.8335,0.8425 and 0.8967,and RMSE are 0.0423 cm~3/cm~3,0.0428 cm~3/cm~3 and 0.0365 cm~3/cm~3,respectively.The inversion results of the three phases are reliable.By comparing the inversion accuracy of pixel scale and block scale,the inversion accuracy of block scale is higher than that of pixel scale,and its R~2 is 0.7249,indicating that the inversion accuracy of block scale soil moisture is suitable for karst mountainous areas and high.(3)Soil moisture in the study area was obtained by multi-model combination inversion,and the results were analyzed by considering planting structure,topographic factors,rocky desertification grade and soil properties.The results showed that soil moisture in June and July in the study area was higher than that in April.The soil moisture in the cultivated land near the urban area was slightly lower than that near the village,which was mainly related to the local microclimate and rainfall in the study area.Based on the analysis of planting structure,the soil water content of maize was lower than that of rice and vegetables,which was mainly related to the planting characteristics of crops.According to the topographic factor analysis,the soil water distribution index of cultivated land at elevations of 1100?1300m was high and the dominant distribution was significant.The distribution of soil moisture in non-karst areas was uniform and stratification was not obvious.According to the analysis of soil physical properties,the mean value of soil water in loam soil is higher,that in clay soil is lower,and that in coarse bone soil is lower,which is related to the soil particle size.The planting structure and soil physical properties of environmental factors had great influence on the distribution characteristics of soil water.The spatial distribution of soil water was affected by topographic factors and rocky desertification grade in terms of microtopography and artificial cultivation.