Research on soil moisture inversion method based on GA-BP neural network model

Global Navigation Satellite System Interferometric Reflectometry (GNSS-IR) is a new remote-sensing technique, and it can be used to estimate near-surface soil moisture from Signal-to-Noise Ratio (SNR) data. Considering the effects of vegetation changes on GNSS-IR in some environments, a non-linear inversion method for soil moisture is proposed. Firstly, the SNR data and satellite elevation angles are solved using Translation, Editing, and Quality Checking. The direct and reflected signals are separated using a low-order polynomial; then, a sinusoidal fitting model of the reflection signal is established; it is used to obtain the amplitude and phase of the SNR interferogram. Finally, an estimation model of vegetation water content and prediction model of the vegetation phase changes are established to modify the original phase and weaken the influence on the vegetation changes. Based on the corrected phase, a Genetic Algorithm Back Propagation Neural Network (BPNN) model is established for soil moisture inversion. According to the GPS monitoring data from the Plate Boundary Observatory H2O network, the experiment indicates that (1) The BPNN is introduced to inverse the soil moisture content, and the non-linear fitting ability of the model is well developed, and the fitting process is stable; (2) the modified phase effectively reduced the effects of vegetation changes on the soil moisture inversion. The correlation coefficient (r) between the inversion results and soil moisture value greatly improved, and the root mean square error and mean absolute error are less than 0.060 and 0.050, respectively. Therefore, the soil moisture problem can be treated as a non-linear event, and the algorithm is feasible and effective.