Thermal anomalies detection before 2013 Songyuan earthquake using MODIS LST data

Abstract

Thermal anomaly appears to be a significant precursor of some strong earthquakes. In this study, time series of 8-day MODIS Land Surface Temperature (LST) product are processed and analyzed to locate possible anomalies prior to the Songyuan earthquake (22 November 2013, Jilin Province). In order to reduce the seasonal or annual effects from the LST variations, also to avoid the rainy and cloudy weather in this area, a background data of 8-day LST are derived using averaging MOD11A2 products from 2001 to 2012. Then the 8-day LST data from September 2013 to January 2014 were differenced using the above background. RX anomaly detection algorithm was then used to detect the anomalies. It is found that day-time LST sequence detect some anomalies near the epi-center region, while through the night-time LST images, river change across seasons show obvious anomalies. The study indicates that LST change detection is somewhat effective in this area for earthquake precursor study.