Study On Pattern Informatics (PI) Method For Detecting Seismo-ionospheric Disturbances In DEMETER Satellite Data

Earthquakes are the most destructive and serious unexpected natural hazards in the world，numerous earthquakes have occurred in China, meanwhile, earthquake prediction is veryimportant for scientific, social, and economic reasons, while it’s still one of the most difficultscientific challenges. Recently, with the development of the space science and technology,especially after the France electromagnetic satellite DEMETER (Detection of Electro-MagneticEmission Transmitted from Earthquake Regions) launched, seismo-ionospheric phenomena studybecame a new focus in the field of earthquake prediction science. Numerous earthquake casestudy and statistical study have shown ionospheric disturbances prior to earthquakes. Developingnew observation technique, making full use of electromagnetic satellite observation data, tryingnew disturbance detection methods have importance to seismo-ionospheric study.Pattern informatics (PI) method was introduced by Prof Rundle in the field of seismic activity;it is a new way to forecast earthquake based on statistical physics. PI method generally used inearthquake long-term forecasting and it has highest spatial resolution of abnormal regions.The fundamental principle of PI method is as following, the region of interest is divided intomany square boxes with linear dimension, and for each box there is a frequency-time series. Then,the abnormal change of seismic active intensity is extracted from each box and form spatial-tempopattern, after that the probability of occurrence of significant earthquakes in each box is computed,and subtract the background probability. At last we have the boxes with high probability ofearthquake occurrence inferred to as the hot spot.Ionospheric parameters have significant tempo-spatial changes, and it’d difficult to obtainsubtle disturbance form original images prior to earthquakes. In this paper, in order to retrieveionospheric anomalies prior to earthquakes, we process DEMETER data (electron density (Ne)and electron temperature (Te)) with PI method to study11significant earthquake cases happenedin china and around the world between2006-2010.In this study, we have following results:(1) Various anomalies have been found in the PI tempo-spatial image both for Ne and Teprior to earthquakes. The common variation process of NE anomaly is: persistent anomalies appear (45~30days before earthquake) anomalies gradually disappear (25~15days) anomaliesreappear (20~10days) anomalies disappear a few days before earthquake. The range andamplitude of anomalies before each earthquake are different, but it’d usually recover after theearthquake. For Te, the tempo-spatial images differ before and after the earthquake, but thechanges are messy and anomalies prior to earthquakes are not very obvious. This phenomenonmaybe related to the phenomenon of rock rupture experiment. After the earthquake, the anomaliesare gradually disappeared, such variations are observed many ground based observations, this ismaybe caused by the electromagnetic radiation emission form rock deformation. Thus, therelationship between the tempo-spatial evolution image changes and earthquakes are still needfurther study.(2) Earthquake case studies showed that, PI tempo-spatial images changes with the referencetime interval T1-T0and change interval T2-T1; the PI images are dependent on the changeinterval T2-T1. Experimental results show that PI anomalies could be obtained with the properselection of T2-T1and T1-T0.(3) In order to remove the back ground changes in ionosphere, we establish the ionosphericdaily background, and use the differential data (difference between the observation data and dailybackground field data) as the input of PI method to recalculate the earthquake cases. The resultsshow that more obvious anomalies of Ne and Te are obtained campared with the previous studywith the input of original observed data. Some individual case images are still massy.(4) The PI method is originally used in seismic activity long-term probability forecast, whichis based on relationship between the historical seismic activity and the future seismic hazard riskprobability prediction. In this paper, PI method is used to process satellite ionosphere data, therelationship between earthquake activity and ionospheric disturbances are complex. Theearthquake case results show that significant anomalies have been found, and PI method iseffective in retrieving disturbances in ionosphere before the earthquake. The mechanism of thetempo-spatial evolution image changes and earthquakes still need further study.