【摘要】 在单参数地震对应概率谱和累计滑动平均概率方法的基础上,研究了多参数滑动极值平均概率方法,用以定量识别地震前兆综合异常特征。首先,该方法基于地震对应概率谱分析,将原始数据时间序列转换成概率时间序列;其次,采用多点累计滑动平均方法得到不同参数滑动平均概率P-ij;再次,求解不同参数的多点滑动极值概率Mij;最后,求解多参数滑动极值平均概率Pc。文中以新疆天山地区不同地震学参数的时间序列为原始数据进行算例分析,结果表明,当考察时段为18个月时,利用多参数滑动极值平均概率可以较好地识别地震前兆综合异常,其异常对应比例和地震对应比例分别为13/13和22/27,异常信度较高,但存在一定程度的漏报现象。更多还原

【Abstract】 Based on single-parameter earthquake corresponding probability spectrum and sliding mean probability research,the article presents multi-parameter sliding maximum-value mean probability method,and quantitatively identifies precursor comprehensive anomaly character.The method first converts original time sequence into probability sequence on the basis of earthquake corresponding probability spectrum analysis.Then by using multi-point cumulative sliding mean method,a sliding mean probability of different parameters [AKP-]_ ij is obtained.After that the multi-point sliding extremum-value probability M_ ij of different parameter is calculated.Finally the multi-parameter sliding extremum-value mean probability P_c is obtained as the result.The example of the original time sequence of different seismic parameters of the Northern Tianshan region in Xinjiang shows that when the studied time-interval is 18 months,multi-parameter sliding extremum-value mean probability can identify precursor comprehensive anomaly better,and the anomaly and earthquake corresponding rate is respectively 13/13 and 22/27.The anomaly confidence is good,but fail alarm is obvious.更多还原