Application of Artificial Neural Network to Distributed Precipitation Estimation Based on EOS/MODIS Remotely Sensed Imagery

Abstract

With the meteorological factors extracted from EOS/MODIS satellite remotely sensed imagery and the corresponding observed precipitation being the input layer and output layer respectively, a back propagation(BP) artificial neural network(ANN) is learned and trained. As the test and application, the distributed precipitations in Qingjiang river basin located in central China are estimated. It is concluded that the precipitations estimated by the BP ANN based on EOS/MODIS are nearly equal to the observed ones at the rainfall stations distributed in the river basin. It is revealed that the integration of EOS/MODIS and ANN provides a new effective way to estimate the distributed precipitation in river basin.