Pre-earthquake infrared long wave radiation abnormal information detection method based on autoregressive moving average model

Abstract

The invention discloses a pre-earthquake infrared long wave radiation (OLR) abnormal information detection method based on an autoregressive moving average model time sequence prediction model (ARIMA). According to the method, the precision of predicting the reference background value is obviously superior to that of a classical time sequence prediction model; particularly, the time sequence prediction precision becomes an important evaluation index of prediction performance, the fitting effect of a time sequence prediction model is directly related to the precision of a prediction result, andthe result has an important influence on subsequent infrared OLR abnormal information detection. Moreover, aiming at the problem that a classical time sequence prediction model method has defects inthe aspect of determining upper and lower thresholds, the method provides a relatively reasonable threshold determination strategy, so that an abnormal result detected by utilizing a high-precision reference background value and the reasonable upper and lower thresholds and an obtained conclusion are more reasonable.

Description

Technical field [0001] The invention relates to the field of earthquake prediction, in particular to a method for detecting anomalous information of pre-earthquake infrared infrared long-wave radiation (OLR) based on an autoregressive moving average model (ARIMA). Background technique [0002] Since the former Soviet Union scholars accidentally discovered thermal anomalies before moderate-strong earthquakes when they analyzed satellite infrared remote sensing images in Central Asia, the relationship between infrared thermal anomalies and earthquakes has gradually attracted attention from scholars from all over the world, and is expected to be used as a forecast for short-term earthquakes. . A large number of earthquake cases and statistical analysis show that in the first few days or one month before the occurrence of a major earthquake, long-wave infrared radiation (OLR) in the earthquake area and its vicinity have different degrees of OLR anomalies, which are mainly characteriz...

AI Technical Summary

Problems solved by technology

[0003] Although the commonly used infrared OLR anomaly information detection method is reasonable to a certain extent, it only considers the inherent properties of infrared OLR sequence data, but does not take into account its uncertainty components. Therefore, the accuracy of classical methods for predicting infrared OLR background values ​​is relatively low. Low

In addition, due to the large number of infrared OLR anomaly extraction methods in classical detection methods, the analysis results of the same earthquake are varied and even contradictory, and there is a lack of reasonable unified standards. Especially when determining the threshold value of the background value, the detection results are directly related It is related to the level of prediction accuracy, and it is easy to cause detection errors, and once the detection is wrong, it will have an important impact on the subsequent detection results

Images