Statistical Analysis of Pre-seismic Ionospheric Electron Density Anomalies Using Ionosonde Data during 1958-2017,Over Japan

Recently, many papers have been published on the earthquake-related ionospheric anomalies. In particular, event studies and statistical studies on Ionospheric electrons using Ionosonde and GPS TEC (total electron content) indicate the pre-seismic anomalies. In Japan region, the previous statistical study during 1998-2010 shows that there is an increase anomaly 1-5 days before the earthquake with M>=6 D<=40 km. In this study, we performed statistical analysis on ionosonde data observed at Kokubunji (35.71N, 139.49E), Japan over 60 years of 1958 2017 in order to investigate the earthquake-related ionospheric anomalies. The ionosonde is operated by the National Institute of Information and Communication Technology, Japan. The studied parameters are NmF2, which is calculated with 1.24×10¹⁰×foF2 and indicates the maximum electron density in the F2 layer, hmF2, which shows the NmF2 altitude, h'F which indicates the apparent altitude of the F2 layer, and foEs which is the critical frequency of the Es layer. For each parameter, 15 days backward median value (M) and its' interquartile range (IQR) were calculated, and we take M-1.5IQR and M+1.5IQR, for the lower and the upper thresholds for anomalies, respectively. The analyzed period is from 1958 to 2017, and we select earthquakes occurred within 1000 km radius from the ionosonde station, and M >= 6 and D <= 40 km. There are 352 earthquakes. To investigate the correlation between ionospheric anomalies and earthquake, we perform the Superposed Epoch Analysis (SEA). We define an anomalous day if the day has anomalies more than 10 hours per a day. Then, we make up the binary time series data on the earthquake-related anomaly. We extract dataset 30 days before and after the earthquake day for the all selected earthquakes and perform SEA. To evaluate the statistical significance, we compute the random test. SEA results show that there is a significant positive anomalies 6-10 days before the earthquakes in NmF2 data. For hmF2 and h'F data, we can find there are no clear correlation. As for foEs, we need further investigation. In our presentation, the SEA results after removing the ionospheric disturbances due to geomagnetic storms and the assessment for the effectiveness in the short-term earthquakes forecast using the Molchan's Error Diagram analyses will be given.