부산대학교 도서관

From 1998 to 2022, Earthquakes have caused more than 850 thousand deaths and US$ 700 billion in economic losses. At this time, there is no reliable precursor to predict them. Many studies have attempted to link Land Surface Temperature (LST) anomalies to earthquake occurrence. This study uses long-term datasets from FengYun-2F/VISSR (Visible and Infrared Spin Scan Radiometer) sensors [1] between 2014 and 2022 to analyze the extent of LST anomalies and their potential link to earthquakes. More than 14,000 land and coastal earthquakes of magnitude higher than M4 have been studied between [+40°, +140°] longitude. Three methods frequently used in the literature, the Standard Deviation (STD), the Interquartile (IQT), and the Z-Score techniques have been implemented to detect LST anomalies on temporal series. Other parameters, such as the confusion matrix and the Receiver Operating Curves (ROC), have also been computed to assess and improve their performance and select the optimum correlation threshold. A positive LST anomaly is typically found before the earthquakes, followed by an LST decrease after the event.