Principal Component Analysis Method Acquiring Soil Moisture Information from MODIS Data

Monitoring soil moisture exactly is very important. Soil moisture is one important factor of agricultural meteorology, which can reflect humid condition of soil, and is a main base to forecast drought of farmland. But it is influenced by too many factors, so it is difficult to monitor real-time soil moisture of largescale areas. General method such as soil sampling method, neutron probe method and TDR method takemuch time and efforts, and can only monitor limited spots. However, the development of remote sensingtechnology can provide assistance to monitoring real-time soil moisture of large-scale areas dynamically.Thermal inertial is a matured technological method to monitor bare soil moisture applying MODIS data. But it needs remote sensing data of both daytime and nighttime, which is difficult to obtain in practicaloperation. It inherits the idea of K-L transformation that is applied in the remote sensing system of targetclassification, uses principal component analysis, regression analysis, residual image, and relative reflection of internal average methods to correct remote sensing radiation data, and establishes soil moisturemodel of multiple-dimensional feature space based on mono temporal normalized MODIS data.Then the result from the image of monitoring is obtained and checked up. Monitored and model results of 35 spots are compared, and the accuracy of the model is 80%. It shows that the model has potentialto be applied in operation. More problems are discussed, including representative of the data, the calibration of remote sensing detector and so on.