Statistical analysis of an ionospheric parameter as a base for earthquake prediction

This paper is related to the use of ionospheric density variations to tentatively predict earthquakes. The results of this statistical analysis are presented as a function of various parameters. The ion density was recorded by the low-altitude satellite DEMETER during more than 6 years, and a search for anomalies was automatically conducted with the complete data set. In a second time, a software checked if each anomaly could correspond to an earthquake. The search was conducted at less than 1500 km from the anomaly positions, and until 15 days after the anomaly time. The earthquakes have been classified depending on their magnitude, depth, and position (below the sea or inland). This attempt to predict earthquakes of course generates a lot of false alarms and wrong detections. Nevertheless, it is shown that the number of good detections increases with the magnitude of the earthquakes. In average the number of perturbations is higher the day of the earthquake, and then smoothly decreases for the days before. Earthquakes below the sea are better detected. There are seismic areas close to the South Atlantic Magnetic Anomaly and at high latitudes where the number of natural perturbations is too important to expect a high number of good detections. Finally, when there are several perturbations corresponding to a single earthquake, it is possible to combine their positions to have a better estimation of the location of the future epicenter. However, uncertainties about the time and the magnitude are large.