Test Of The Predictability Of The Pattern Informatics Method By Long-term Retrospective Study On Large Earthquakes

Until recently,scientists worldwide still remain in the exploratory stage about reliable earthquake prediction,which,however,has been pursuit for ages due to catastrophic loss of life and property caused by devastating earthquake.Study on large earthquakes are not confined to the observation data of the ground surface and earthquake catalogue today.In recent years,research on ionospheric anomalies related to earthquakes,has attracted wide attention.In this paper,firstly learns theoretical knowledge systematically about the Pattern Informatics Method("PI")which have statistical property,and focus on 3 shallow source large earthquakes.Long term retrospective study was carried out on these 3 earthquakes combined with different data.The main research contents and results are as follows:1.Using earthquake catalog from Japan Meteorological Agency,medium-and-long-term retrospective study on M>7.0 earthquakes since 2000 in Japan local area(32.0°-46.0°N,136.0°-148.0°E),including 3·11 Mw9.0 earthquake,were carried out,applied PI method under different models with different parameters.The main variable calculating parameters were respectively the grid size of 0.5° X 0.5° or 1.0°×1.0° and forecasting window lengths from 5 to 10 years.The correlation between "hotspots" and those earthquakes from the continuous forecasting windows were analysed qualitatively,taking earthquakes' tectonic environment into consideration.From the view of forecasting window patterns,the result presented hotspots for most earthquakes,especially Tohoku Mw9.0 earthquake,which is a rare giant trench earthquake sample of PI research,its hotspots were more concentrated and durable.2.Employing the receiver-operating characteristic(ROC)curve test and R score test,medium-and-long-term forecast performance of PI method under different models with different parameters were estimated qualitatively.Those 2 test shows that PI has good medium-and-long-term forecast performance with the relatively long forecasting window length and relatively large space grid size.The result in this paper shows that false and omitted cases were observed for some research target of M>7.0 earthquakes.How to avoid or lower the rate of false and omitted cases is the research emphasis in the future.3.With the property to adapt to different data,PI method was applied to analyse ionospheric anomalous disturbances of earthquakes.With the data from DEMETER electromagnetic satellite(electron number and electron temperature),abnormal pattern evolution characteristics study on the Wenchuan China Ms8.0 earthquake(inland earthquake)and Vanuatu Archipelago Ms7.3 earthquake(sea area earthquake)were carried out,applied modified PI method.Compared with PI method with ionosphere parameters used before by other researchers,different parameters were adopted for the research on two M>7.0 strong earthquakes.In this paper,the retrospective time scale was extended to 1 year.In long-term continuous disturtbance evolution pattern,both large earthquake in ocean plate(subduction)boundary and earthquake of inland showed common characteristics,that is abnormal disturbance is not focused on the epicenter in the spatial distribution of the vertical,but towards the magnetic equator.4.For the same earthquake cases,under the same parameters of time window and space grid size,the time periods of anomalies of the electron number and electron temperature(distinguish from background field),were not synchronous.Under the same physical parameter of the certain earthquake,of abnormal perturbation of different time periods,the patterns were not repeatable.For different earthquakes,there is no fixed length of time from the date of the earthquake to abnormal occurrence.However,witnin 3 months before the earthquake there was an abnormal disturbance that was significantly distinguish from the background field and tended to background value.The abnormal perturbation in each concentrated period is characterized by the uncertainty of the precursory anomaly,the difficulty of qualitative analysis and quantitative identification.The earthquakes case study related to anomalous perturbation of ionosphere in this paper is instructive to the data processing and application of data from China's own electromagnetic satellites which will be launched in the future.5.In conclusion part,based on the previous researches,the differences between applied PI method with earthquake catalog and ionospheric data were analysed contrastively,which including the difference in data,in model parameters,and future prospects.