Ahmad Keshavarz

The subject of the research is to compare the backscattering coefficients in the empirical models for SAR with their in-situ measurements and to present a model based on machine learning algorithms for soil moisture retrieval. The main goal is to present a model based on machine learning algorithms to fix the error in the empirical models and to estimate soil moisture with high accuracy. The empirical model used is Baghdadi model and the machine learning algorithm is support vector regression. The sampling method is random selection of training and test data based on the date of the data and also based on the number of soil moisture ground data measurement stations. Data was also collected by pre-processing SAR images with the help of open source algorithms of the European Space Agency and using machine learning algorithms. Finally, the information was analyzed with common criteria for evaluating system accuracy, including Pearson’s correlation coefficient, coefficient of determination, and root mean square error. The results showed that with the support vector regression, the error between the soil moisture in Baghdadi model and in-situ measurements was reduced and the soil moisture was estimated with high accuracy.