Study On Inversion Of Grassland Soil Moisture Based On Dual Polarization Data

Grassland soil moisture is one of the important indicators for evaluating soil quality.It maintains the stability of the ?soil-vegetation-atmosphere? grassland ecosystem.The loss of grassland soil moisture directly threatens the survival and development of human beings.The development of synthetic aperture radar technology has made it possible to continuously monitor soil moisture over a large area.However,the backscattering coefficient of synthetic aperture radar is not only affected by soil moisture,but also affected by surface roughness and vegetation coverage.This is the difficulty in retrieving soil moisture from synthetic aperture radar.According to these problem,this paper does the following research:(1)Taking the Xilinhot desert grassland as the research area,considering the influence of vegetation cover,the applicability of the empirical scattering model Dubois and Oh model is discussed.The soil moisture of Xilinhot is measured based on Dubois model and Oh model respectively,and the inversion accuracy is evaluated by using the measured data.The results show that the Oh model has a narrow application range and poor applicability to the dry surface or rough surface.The Oh model inversion accuracy in the study area is better than the Dubois model.(2)Based on the AIEM model,the relationship between radar backscattering characteristics and surface parameters is simulated,and the variation of backscattering characteristics with surface parameters under X-band is revealed.(3)Based on the AIEM model,the relationship between HH polarization,VV polarization,HH-VV polarization difference with surface roughness parameters is discussed respectively.HH-VV polarization difference information is introduced to establish soil semi-empirical moisture inversion model based on AIEM model.The model inversion value is in good agreement with the theoretical input value,and the model accuracy is high.(4)Taking the Baori Hiller meadow grassland as the research area,using Sentinel-2A data to extract vegetation canopy water content as the input of water cloud model,combined with water cloud model and established semi-empirical soil moisture inversion model,mapping Baori Hi Hiller soil moisture.The results show that the water content of vegetation can be effectively extracted by Sentinel-2A;the established soil water inversion model can be used to invert soil moisture under the cover of meadow grassland in this study area.(5)Selecting the feature band of soil moisture Based on Sentinel-2A,combining with measured soil moisture and SAR image values to establish BP neural network to predict soil moisture.The results show that the Sentinel-2A red edge band has a good correlation with soil moisture.Compared with the Landsat data,this is an advantage for estimating soil moisture;BP neural network can express the the nonlinear relationship between backscatter coefficient and soil moisture well,it has great potential in multi-source soil moisture inversion.