The Earthquake Probability Prediction Based on Weighted Factor Coefficients of Principal Components

Considering there are many seismicity factors with information overlap, we apply a gray generating-A generating GM model which has higher precision to mine the potential law of these factors, and apply principal component analysis (PCA) approach to reduce the dimension of these factors, choose several seismicity factors which represent most of the information that the original seismicity factors carry, by utilizing Bayes posterior probability discriminant analysis, to predict the largest magnitude for certain region and period numerically and synthetically. As an example, we predict the annual largest magnitude for North China (30°~42°E, 108°~125°N). The application results show the practical value of this model in the medium-short-term earthquake prediction.