Anomaly Analysis Of The Sensing Data Before And After The Earthquakes

Abnormal pre-seismic phenomena of Satellite infrared remote sensing data is an effective supplementary means for earthquake forecasting warning. And the related research becomes one of the hot research topics in recent years. Because the existing research methods of pre-seismic anomaly lack of a large number of earthquake cases and comprehensive analysis of many kinds of infrared remote sensing data, here we propose three abnormal analysis algorithms step by step. The main research work includes:1、 In view of the CUSUM algorithm can detect the mean change in the statistical process, We propose the algorithm for anomalous analysis based on CUSUM algorithm of Outgoing Longwave Radiation (OLR) data at a resolution of 1°×1°. Firstly, calculating cumulative sum of sequence’s mean with sample observation data; Secondly, using the local smoothing method to smooth the cumulative sum; Thirdly, using the local maximum and minimum structure to obtain characteristic curve; Finally, adopting double threshold value control method for anomaly detection. The experiments of the Wenchuan earthquake and Argentina earthquake show that the abnormal degree gradually change from small to large before the earthquake, and up to the maximum value during the earthquake, then fall down gradually after the earthquake. Experimental results show that the proposed method is feasible and can effectively find the abnormal changes before, during and after the earthquake.2、We propose the average power spectrum (APS) seismic anomalous analyzing algorithm based on wavelet transform and short time Fourier transform. Algorithm assumes that OLR data at a resolution of 2.5°×2.5°is the additive model, and using "db8" wavelet to mining the hidden composition which is closely related to the earthquake in the OLR data. And using short time Fourier transform to analyze its related spectrum features, put forward the average spectrum with specific frequency range as the basis of seismic anomalous analysis. We use the algorithm to analyze the eleven strong earthquakes with magnitude greater than 7.0 at home and abroad. These experimental results show this characteristic of seismic anomaly before and after is ascending-stable-descending.3、 For the question of infrared anomaly study before the earthquake lack of a variety of comprehensive analysis method for infrared remote sensing data, we propose the anomalous analyzing algorithm based on Random Walk. The Outgoing Longwave Radiation data of NOAA meteorological satellite and NCEP/NCAR reanalysis data covers the surface temperature data, potential temperature and pressure data at a resolution of 2.5°×2.5°, comprehensive to analyze eight strong earthquakes with magnitude greater than 8.0 at home and abroad in 2008-2013. We analyze the intensity, location, and the number of days with greater than 2 times the variance based on the Random Walk results, and compare results of four kinds of datas. The experimental results show that at least two kinds of data have synchronized anomalous variance with greater than 2 times in every earthquake, and the anomalies occur within 2 weeks before the earthquake.