利用概率密度分布提取地下流体数字化观测资料中的高频异常信息——以2014年鲁甸6.5级地震为例

摘要

中国地震地下流体观测经过数字化改造以后,观测资料的采样频率显著提高.流体观测资料中的高频数据含有丰富构造信息,为我们捕捉地震孕育及发生过程中的前兆异常信息提供了有利条件.但是,高频信息的出现激发了对数据分析方法改变的需求,如何研发与数字化高频观测资料相匹配的数据处理和异常识别方法,从高频观测数据中挖掘潜在的前兆异常信息,成为目前地震地下研究者首要需解决的关键问题.应用概率密度分布法对2014年鲁甸6.5级地震前南北地震带174组水位、水温分钟值高频观测数据进行分析,结果显示:鲁甸6.5级地震前共有10个水位测点和7个水温测点出现高频信息异常,异常多集中在滇西南构造带的滇中次级块体两侧,且随着时间推移,有向震中区逼近的变化特征.通过对震源区及附近地区地壳结构、构造应力作用及更大范围的区域动力演化特征进行分析,发现异常信息的空间分布特征与川滇地区地壳运动场具有很好的一致性,说明概率密度分布可有效反映出区域构造应力作用,同时也验证了利用概率密度分布在流体观测数据的高频信息异常提取方面具有一定可靠性.%The sampling frequency of data has been obviously improved after the digital transformation for the seismic subsurface fluids observation in China.And there is a great deal of tectonic information in high frequency observational data,creating favorable condi tions for us to catch the precursory information in earthquake nucleation and occurrence process.How to develop data processing and anomaly recognition method matching with the digital high frequency observation data for extracting useful abnormal information become a key issue.We introduced the PDF (probability density distribution) method to analyzed fluid observation data of water level and water temperature at 174 stations in the area of the North-South Seismic Zone before the Ludian Ms6.5 earthquake on August 3,2014.The analysis results show that 10 water level and 7 water temperature data had the high frequency anomalies before the earthquake,and the points of abnormal information were concentrated in the tectonic belts of the central Yunnan sub-block in the southwest.The high-frequency anomalies migrated to the epicenter over time.Based on the analysis of crustal structure and tectonic stress near the earthquake epicenter,as well as the dynamic evolution in a larger area in this region,we found that the spatial distribution of abnormal information had a good consistency with the crustal deformation of the Sichuan-Yunnan region.This result not only shows that the PDF can effectively reflect the regional tectonic stress field,but also has certain reliability and applicability to extract high-frequency anomaly information from fluid observation data.