Publications

Shah, M; Qureshi, RU; Khan, NG; Ehsan, M; Yan, JG (2021). Artificial Neural Network based thermal anomalies associated with earthquakes in Pakistan from MODIS LST. JOURNAL OF ATMOSPHERIC AND SOLAR-TERRESTRIAL PHYSICS, 215, 105568.

Abstract
There are several reports about thermal anomalies associated with the impending earthquakes (EQs); among which the underlying mechanism is linked with the main shock with long, intermediate and short-term precursors. In this paper, we analyzed thermal anomalies from Moderate Resolution Imaging Spectroradiometer (MODIS) based Land Surface Temperature (LST) of three different magnitudes and shallow depth EQs in Pakistan. Our focus is to investigate the thermal anomalies by the statistical approach and Artificial Neural Network (ANN) in spatial and temporal LST values within 10 days before and after the main shock as short-term precursors. After implementing statistical and the ANN approach, LST revealed that thermal anomalies occurred within 1-10 days before the main shock of M-w > 6.0 EQs. However, a low intensity LST anomaly is also recorded within 20-25 days before the main shock of M-w 5.2 EQ. We study that LST anomalies are magnitude and depth dependent and it is more likely to occur before EQ of M-w > 6.0 and shallow depth within 10 days before/after the main shock day. The results depict that M-w 5.2 EQ anomaly is not clearly associated to the main shock, as it locates outside the window of 1-10 days before/after the main shock. Similarly, two out of three events caused post-seismic thermal anomalies of less magnitude as compared to the pre-seismic thermal anomalies. The preseismic LST anomalies occur with high intensity before shallow depth and large magnitude EQs than post seismic LST anomalies. The LST anomalies occurred before all the EQs suggesting it to be a reliable precursor of short to intermediate interval associated with an impending EQ.

DOI:
10.1016/j.jastp.2021.105568

ISSN:
1364-6826