Inversion of the soil moisture based upon neural network

An inversion method based upon a neural network is applied to retrieve the soil moisture and roughness parameters from the radar backscattering coefficients. The structure and the training algorithm of the neural network (NN) are presented in the paper where the training patterns are generated by an integral equation model (IEM). By choosing the proper type of input data, the necessary input data number is minimized. Analyze of the calculation results shows that the NN inversion method has high accuracy.