A Forward Model of Soil Moisture Estimation Based on Millimeter Wave PolSAR Images

Using synthetic aperture radar (SAR) data to predict soil moisture is one of the most important applications of remote sensing. There are currently two mainstream methods: the forward prediction models based on the internal influence mechanism of soil moisture and the data-driven methods. Due to the extreme complexity of the internal influence mechanism of soil moisture, the corresponding forward prediction models established always rely on a number of empirical parameters, resulting in their lack of precise adaptability in some scenarios. Although data-driven methods can achieve ideal prediction accuracy, they lack exploration of internal mechanisms. In response to the above issues, this paper investigates the possibility of using data-driven methods to assist in establishing soil moisture forward prediction models. Firstly, the empirical parameters of the Oh model are used as optimization variables to establish a mathematical optimization model. Then, based on the optimized model, the optimal soil moisture model parameters were obtained by using the sample data of millimeter band polarimetric SAR images in the study area. Finally, a prediction model with stronger adaptability to the study area was obtained. The soil moisture prediction results of the test samples indicate that the model established by this method has higher prediction accuracy compared to the traditional Oh model.