Retrieval Of Surface Soil Moisture In Winter Wheat Field Based On Sentinel-1/2 Data

Wheat plays an important role in xinjiang’s food security construction,among which winter wheat accounts for a large proportion in the output,so it is particularly important to ensure the normal growth of winter wheat.Soil moisture is closely related to crop health.Monitoring surface soil moisture is one of the important factors to judge the growth and yield of winter wheat.Too low soil water content will inhibit the normal growth and development of winter wheat,which will not only reduce the winter wheat yield,but also cause crop failure.In this study,the surface soil of winter wheat field was used as the research object,and the soil moisture estimation model was established based on the backscattering coefficients obtained from Sentinel-1 microwave remote sensing image data under different polarization modes.At the same time,in order to eliminate the influence of vegetation layer in soil inversion,a variety of vegetation indices based on Sentinel-2 optical image were used to analyze their correlation with the field measurement data of winter wheat canopy water,and representative vegetation indices were selected to establish the empirical equation of winter wheat canopy water estimation.The water cloud model was improved by using this equation,and combined with the measured soil moisture data in the field,a semi-empirical model for retrieving soil moisture in the field of winter wheat was established by partial least square method,so as to realize the cooperative inversion of microwave remote sensing and optical remote sensing.The main conclusions are as follows:（1）Based on the correlation analysis of normalized vegetation index（NDVI）,improved soil adjusted vegetation index（MSAVI）,difference vegetation index（DVI）,normalized water index（NDWI）and winter wheat canopy water content,the results showed that the four vegetation indices were significantly correlated,indicating that these vegetation indices were closely related to winter wheat canopy water content.（2）The water retrieval model of winter wheat was established by linear regression and the results were analyzed.The determination coefficients of NDVI,DVI and MSAVI remained between 0.4 and 0.6.While NDWI is greater than 0.6.MSAVI is the vegetation index obtained by adding soil regulation coefficient and improving NDVI,which is more suitable for monitoring vegetation status under the interference of bare soil background.The inversion accuracy of NDWI is the highest,indicating that short-wave infrared is very suitable for reflecting water stress of winter wheat.Therefore,MSAVI and NDWI were selected to construct the empirical equation of winter wheat canopy water.（3）The correlation between soil moisture and backscattering coefficients of two single polarization modes VV and VH on Sentinel-1 and two double polarization modes VV/VH and（VV-VH）/（VV+VH）obtained by band calculation were analyzed.The results showed extremely significant correlation,and the correlation degree of VV polarization was higher.The absolute value of correlation coefficient exceeds 0.8.The correlation coefficient of VH polarization is about 0.4,while the correlation coefficient of VV/VH and（VV-VH）/（VV+VH）are very close,and the absolute value of the correlation coefficient is 0.624 and 0.632,respectively.（4）Two improved water cloud models were obtained by substituting the empirical equation of winter wheat canopy water represented by MSAVI and NDWI.These two models eliminate the influence of winter wheat canopy water on soil water retrieval to a certain extent.Partial least square method（PLSR）was used to fit the improved water cloud model based on MSAVI,the model set of measured soil moisture data,and the backscattering coefficients of the four polarization modes,respectively,to obtain the semi-empirical model of soil moisture inversion.Similarly,the semi-empirical model of soil moisture inversion is also established by using the above method for the improved water cloud model based on NDWI.The R~2of VV-MSAVI is 0.634,VH-MSAVI is 0.138,VV/VH-MSAVI is 0.505,（VV-VH）/（VV+VH）-MSAVI is 0.731.And R~2of VV-NDWI is 0.669.VH-NDWI is 0.214,VV/VH-NDWI is 0.676,（VV-VH）/（VV+VH）-NDWI is 0.751.（5）By comparing the accuracy analysis of soil water inversion models of 8 different combinations,it is concluded that the improved water cloud model based on NDWI can effectively eliminate the interference of winter wheat water during the estimation of soil water under the condition of winter wheat covering.At the same time,the inversion model based on（VV-VH）/（VV+VH）polarization mode was least affected by the attenuation effect of vegetation layer,and could effectively represent the spatial distribution of soil water in winter wheat field.