Background Field Foundation And Anomaly Extraction Of Pre-seismic Ground Surface Temperature Based On Thermal Infrared Data

Earthquake is a kind of vibration triggered by the tectonic activities inside the earth.It has been a long-heated debate in science as it threatens human life and economy due to its high destructiveness and complexity.Many researchers’ studies show that the surface temperature anomalies generally occur before earthquakes,and it has become a mainstream earthquake research that making use of satellite thermal infrared remote sensing technology to monitor the thermal infrared radiation anomalies before earthquakes.Changes of surface temperature result from a combination of the sun and the atmosphere.It is not only impacted by crustal structural activities but also some non-tectonic information such as altitude,longitude,latitude,human activities.It is unreasonable to directly use the temperature anomaly information on the remote sensing image as an earthquake precursor.Moreover,weakens the correlation between the surface temperature anomalies and tectonic activity is that the absorption of water vapor to electromagnetic wave make the original surface temperature data deficiency seriously.Comparing the temperature anomalies to the normal range,only the surface temperature background field is under normal state can the information of temperature anomalies be extracted.MODIS(Moderate-resolution Imaging Spectroradiometer)surface temperature data is applied to the supplement data of pixels under cloud in this paper,and the methods based on wavelet analysis,singular spectrum analysis and deep analysis are used to establish the surface temperature background field before earthquakes and extract anomalies.The main contents include:(1)Four different temperature products of MODIS data are used to combine with land cover type data,then to establish a linear regression model for each type of surface features.This model can obtain the temperature values after the three different times are normalized to the values.and the three normalized values weighted fusion is carried out as the supplement of the missing area of pixel under cloud.According to the supplementary value experiment in Hetian region of Xinjiang,it is found that the average success rate of supplementary is 97%,and the precision of supplementary rate within 0 ～ 2K is 81.22%.(2)The methods based on wavelet transform,singular spectrum analysis and deep learning are used to construct the background field of surface temperature,then the in situ temperature method is used to extract anomalies.The first method is applied to test the global earthquakes with 8.0 and above magnitude from 2003 to2015.The results show that 68% of the earthquakes were preceded by temperature anomalies.The last method is used to construct the ground surface temperature background field before the earthquake successfully,which can extract abnormal information and can provide a new way of thinking for using thermal infrared remote sensing data for earthquake warning.