时序数据年变信息提取及地震预测指标评估方法研究

Research on the Extraction of Annual Variation Information of Time Series Data and the Evaluation Method of Earthquake Forecasting Index

针对时序数据中的破年变异常,基于S变换时频方法构建信息提取流程,在提取常规短周期破年变信号(ONA)的同时分析背景年变信号(ANA)的演化过程。在此基础上,基于双向非对称阈值策略,结合R值评分及Molchan图表法构建预测指标确定和效能定量评估方法。新疆库尔勒水平摆倾斜北南测项的实际应用结果显示,该测项的ANA信息对台站周边250 km内6级以上地震具有较好的预测效能,ONA信息对200 km范围内5级以上地震的预测效果相对更优。

Aiming at the anomaly of annual variation in time series data,we construct an information extraction process based on the S-transform method.The evolution process of the background annual variation signal(ANA)can also be analyzed in addition to the extraction of conventional short-period annual variation signal(ONA).Based on the bidirectional asymmetric threshold strategy,combined with the R value score and Molchan diagram,we construct a quantitative method for forecasting index determination and performance evaluation.The application results of the NS component of horizontal pendulum tiltmeter in the Korla station,Xinjiang,show that the ANA has better forecasting performance for M≥6 earthquakes within 250 km around the station,and the ONA is relatively better for M≥5 earthquakes within 200 km.