基于多源遥感数据的宁夏河东沙地土壤水分反演研究OA

In order to accurately understand the distribution characteristics of soil moisture in the sandy land east of the Yellow River in Ningxia,this study explores inversion models and methods for soil moisture content in sandy land,which provides theoretical support and decision-making reference for research on the spatial distribution of soil moisture.Using microwave remote sensing data from Sentinel-1 SAR combined with multispectral data from Sentinel-2,the soil moisture content at different depths(5,10 and 20 cm)in the sandy land east of the Yellow River in Ningxia was inverted by integrating the Water Cloud Model and the Random Forest algorithm.The results showed:①Correlation analysis between different environmental variables and soil moisture content across layers revealed significant differences.Band reflectance(B11,B12)showed a significantly negative correlation with soil moisture at all depths.The normalized difference vegetation index was positively correlated with soil moisture at 5 cm and 10 cm,while the backscattering coefficient(VV+VH)exhibited a significantly negative correlation with soil moisture at 5,10 and 20 cm.②Analysis of model evaluation metrics showed that the coefficient of determination between measured and validated values for soil moisture across the three layers ranged from 0.75 to 0.85,with root mean square errors between 0.02%and 0.04%,indicating that the Random Forest model achieves high accuracy in soil moisture inversion.③ The soil water content in the study area was generally low,and the overall soil water content was less than 3.00%.The inversion results of soil water content show strong consistency with the actual measured values.By integrating Sentinel-1/2 remote sensing data with vegetation indices and backscattering coefficients,the method demonstrates high accuracy in soil moisture inversion for this area and can effectively capture the spatial distribution characteristics of regional moisture.