震前电离层TEC异常探测新方法

本文提出一种利用时间序列法(ARIMA模型)进行震前电离层异常探测的新方法.首先,对比分析了该方法与传统探测方法(四分位距法、滑动时窗法)预测TEC背景值的精度.结果表明,时间序列法预测背景值的精度要明显高于传统方法,且预报背景值的平均偏差要比传统方法小2倍左右,说明传统探测方法预测的背景值具有较大系统偏差.为更准确地探测震前电离层扰动,除了得到准确的参考背景值,还需得到更加合理的探测限值,由此本文提出一种更为合理的限差确定策略.最后,以2012年1月10日苏门答腊岛7.2级地震为例,利用该方法分析了其震前电离层的异常扰动情况,并验证了该方法的有效性,实验结果表明:在震前第13天、第8~9天、第1~2天和地震当天电离层均会发生较为明显的异常.而且,其正异常(观测值高于正常值)一般发生在震中以北,距发震时间相对较远;负异常(观测值低于正常值)则在震中各方向均会出现,且距发震时间较近.同时,通过对异常结果分时段统计,发现在发震时刻前,距发震时刻越近的时段发生异常的频率越高,此规律将会对未来更为准确的预报发震时段提供重要参考.

A new method for detection of pre-earthquake ionospheric anomalies

This paper proposed a new method for detection of pre-earthquake ionospheric anomalies using time series analysis based on the Autoregressive Integrated Moving Average (ARIMA) model. Firstly, we compared the precision of this new method with the traditional ones, namely the Inter Quartile Range (IQR) method and the sliding window method, in predicting the TEC reference background values. The results show that the precision of the former is obviously better than the latter, while the average prediction residual errors of the former are twice smaller than the latter. To detect pre-earthquake ionospheric anomalies more accurately, besides precise reference background value, its reasonable error range is also needed. Therefore, this paper put forward a new method to calculate the reference background value's upper and lower bounds. Finally, the earthquake happened in Sumatra on January 10, 2012 was taken as example. We analyzed its pre-earthquake ionospheric anomalies and proved the effectiveness of the new method. The results show that obvious ionospheric anomalies appeared on the 13th, 8th to 9th and 1st to 2nd days before the earthquake as well as several hours during the day when the earthquake happened. Furthermore, positive anomalies (observational values higher than normal ones) generally appeared to the north of the epicenter and are much earlier before the earthquake occurrence, while the negative ones (observation values lower than the normal) occurred in any direction to the epicenter, and close to the moment the earthquake occurrence. Through statistics for the frequency of ionospheric anomalies occurred at different times of the day, we have also found a valuable law that the closer the time of the day before the earthquake moment, the higher the occurrence frequency of the ionospheric anomalies, which is likely to be an important reference to the more accurate earthquake prediction in the future.