Abstract: In this investigation,we proposed a local correlation analysis method by extending the traditional correlation method to the local space,as the basis of the quantitative analysis of time series. Furthermore,we proposed an interference subtraction method, by integrating several methods including FFT smoothing filtering, the above proposed local correlation (Local-CC) analysis and llnear regression (LR), based on llnear response of deformation to temperature,to subtract the effect of the interferences on the premonition (e. g. cave deformation). As an example,we use the measured data of the cave deforma tion and the temperature recorded by Kunming standard seismic station, and apply the method of FFT smoothing filtering to extract the higher frequency parts from the original signals,and extract the high correlation time interval between the two data by using the Local-CC analysis,and then solve the responsibility of the cave deformation data to the temperature using LR. We get the final deformation data after using the interference subtraction method. It seemed that our method can properly subtract the interference of temperature to deformation.

Key words: Local correlation analysis, Interference subtraction method, FFT smoothing tiltering, Cave deformation, Linear regression