A Satellite Remote Sensing Monitoring Model for Soil Moisture Based on Artificial Neural Network

Based on the satellite data such as precipitation estimation, incident radiation, brightness temperature etc. and the assimilation data of CLSMDAS, combined with B-P neural network to develop a new model of soil moisture monitoring. The model mining the relationship between the soil moisture and the satellite products, then calculate the weight and build models using artificial neural network which has ability of nonlinear processing, and finally output the soil moisture data which is high precision, continuous time and space. Experiments show that the monitor product of the soil moisture by the new model is more accurate than inversion by the AMSR-E, so that it can be used in large-scale to monitor the soil moisture by remote sensing.