Research Of Soil Moisture Inversion Using Time Series SAR Data Based On Improved Alpha Model

As an important surface parameter,soil moisture is a key input parameter in many models,such as agriculture,meteorology and disasters.As a promising means to monitor the spatial and temporal changes of soil moisture in a wide range,radar has the advantages of all-weather and all-weather relative to optical remote sensing,and has been widely studied by scholars in the past.The difficulty in using radar to monitor soil moisture is that the backscattering coefficient is related to multiple surface parameters.How to eliminate or reduce the influence of roughness and vegetation factors on the backscattering signal,and determine the exact relationship between it and soil moisture is still the current research focus.Previous studies have shown that multi-temporal methods based on change detection techniques have great potential in soil moisture inversion.The method assumes that surface roughness and vegetation change slowly compared to soil moisture(regardless of tillage),so it is approximately constant,and the change of backscatter coefficient observed by radar during the observation period is attributed to soil moisture.Establishing a connection between the two to minimize the interference factor,this method provides a better idea for solving the above problem,and does not require additional auxiliary data.Among the methods based on change detection techniques,the Alpha approximation model has received extensive attention due to its model simplicity.However,when the incident angle is changed in the acquired time-series SAR data,the change of the backscatter coefficient will also include the influence of the incident angle.The premise of the method is no longer valid,and the applicability of the Alpha model is also affected.Aiming at the above problems,this paper took the change of the incident angle of time-series SAR data into consideration and proposed an improved Alpha model.Based on the empirical linear model and cosine model,the incident angle correction of time-series SAR data and the effect on soil moisture inversion results are discussed.The power exponent in the cosine model and slope in the empirical linear model are determined by similar pixel data with the support of optical remote sensing image classification.Two bare farmlands were selected as experimental areas,which were located in the Heihe River Basin of China and the Manitoba Province of Canada.The improved model was validated using the Sentinel-1 data from the ESA and the soil moisture observation network data of the corresponding area,and the results were analyzed and discussed.The main conclusions of the study are as follows:(1)Using multi-temporal SAR data for time-series soil moisture inversion based on the Alpha model,the change of incident angle will affect the inversion accuracy of the soil moisture.The improved Alpha model proposed in this paper takes the influence of angle change into consideration by coupling the incident angle correction method,which improved the inversion accuracy of soil moisture.Therefore,the Alpha model could be extended to time-series S AR data with varying incident angles,which improves the utilization of radar data.(2)In this paper,two kinds of incident angle correction methods are used,which are empirical linear correction and cosine correction.According to the comparative analysis of soil moisture inversion results at two experimental sites,we found that the cosine method perform better.The backscattered signal received by the radar is more consistent with the Lambertian cosine law with the change of the incident angle.(3)The exponent of the cosine of incident angle for different surface types is different.The commonly used cosine correction forms(COS2θ and COSθ)couldn’t correct the effect caused by the change of incident angle in bare farmland.For this type of surface,this paper recommends using the cosine correction form COS4θ～COS5θ.