结合PCA和滑动时窗的电离层异常探测方法研究

针对主成分分析（PCA）方法探测电离层异常时探测范围有限且无法区分正负扰动的问题，提出一种结合PCA和滑动时窗的探测方法，并以3次强震为例检测其异常探测能力。结果显示，PCA-滑动时窗方法同时具备了2种方法的探测优势，相较于滑动时窗法，其探测结果更加直观、简明，且不易受空间环境影响，有相对更高的异常探测置信度和可靠度；相较于PCA方法，其能更好地用于长时间时空异常探测，并准确区分出正负异常。另外，通过进一步分析2020年瓦哈卡地震后发现，电离层在震前11~13 d、9 d及4~6 d出现明显异常，且异常位于震中偏南区域。综合4次地震可知，异常主要出现在震中偏南方向，大多呈共轭结构。

Study of Ionospheric Anomaly Detection Method Combining PCA and Sliding Time Window

Considering the limit of the PCA method in detecting ionospheric anomalies, we propose a detection method that combines PCA and sliding time window. We use three strong earthquakes as examples to test its anomaly detection capability. The results show that this method retains the advantages of each of the two methods. Compared with the sliding time window method, its detection results are more intuitive and concise, and have stronger resistance to environmental disturbances as well as relatively higher confidence and reliability of anomaly detection. Compared with the PCA method, it can be better used for long-time anomaly detection and accurately distinguish between positive and negative anomalies. On this basis, this paper analyzes the anomalous perturbation of the pre-earthquake ionosphere for the 2020 Oaxaca earthquake using this method. The study showed that the TEC anomalies increased 13 days before the earthquake, and decreased 12, 11, 9, and 6 to 4 days before the earthquake in the ionospheric TEC, which is located in the southerly region of the epicenter and mainly manifested as negative anomalies. Finally, combining these four earthquakes, it can be found that the ionospheric anomalies mainly appear in the southerly direction of the epicenter, and most of them are conjugate structures.