The preparation processes of large earthquake may start several or dozens of days before its actual occurrence. As scholars have observed the links between stationary GPS measurements of ground deformation and the occurrence of large earthquake, the continuous GPS data can be used as tools to identify such preparation processes. A method based on Martingale theory is proposed to extract the anomalies in continuous GPS data prior to 15 large earthquakes in western North America during 2007–2016. The precursory information of the anomalies for the corresponding earthquake is evaluated by using Molchan^,s error diagram. The results validate that the extracted anomalies have a strong correlation with the short-impending large earthquakes. This demonstrates that changes of ground deformation recorded by GPS contain useful information for earthquake forecasting and the proposed method therefore can be used to improve the prediction of large earthquakes.

中文翻译：

---

连续的GPS数据异常是2007-2016年北美西部15次大地震的先兆

大地震的准备过程可能在实际发生前几天或几十天开始。正如学者们观察到的，固定的GPS地面变形测量与大地震的发生之间的联系，连续的GPS数据可以用作识别这种准备过程的工具。提出了一种基于Mar理论的方法来提取2007-2016年北美西部15次大地震之前的连续GPS数据中的异常。使用Molchan评估相应地震异常的前兆信息^，的错误图。结果证实，提取的异常与短时大地震有很强的相关性。这表明，GPS记录的地面变形变化包含了地震预报有用的信息，因此该方法可用于改善大地震的预报。