APPLICATIONS OF FUZZY MATHEMATICS IN EARTHQUAKE PREDICTION

Abstract

This paper presents a direct method for fuzzy recognition of earthquake precursors and a method of fuzzy clustering analysis of temporal seismicity patterns for earthquake prediction. The direct method for fuzzy recognition of earthquake precursors is based on their fuzzy description by use of membership functions. If we use a suitable membership function, the abnormal features of a precursor can be more clearly recognized. The membership function proposed by Feng et al.. in 1981 is applied to reported changes in V_P/V_S, resistivity, and radon content. The basic idea of application of fuzzy clustering analysis to earthquake prediction is that time intervals with different sizes of the largest earthquake can be classified on the basis of fuzzy similar relations according to the various kinds of statistical indices of seismicity. The method based on the fuzzy equivalent relation includes the following steps: obtaining a fuzzy resemblance relation between samples, transforming it into a fuzzy equivalent relation, and classifying the original samples. The statistical indices of seismicity used by Feng et al. in 1984 is employed for clustering analysis of seismicity in the North-South Seismic Belt of China.