5:15 PM - 6:45 PM
[SCG50-P15] Detection of atmospheric radon concentration anomalies related to earthquakes using a statistical time series model
Keywords:radon, time series analysis, statistical model
Radon(222Rn) is a radioacrive noble gas and has a half-life of about 3.8 days.Many studies have reported the phenomenon that atmospheric radon concentration changes before or after earthquakes. It is due to changes in crustal conditions caused by earthquakes, which affect radon from the subsurface. Atmospheric radon concentration variations include seasonal variations, and it should be addressed to discuss the relationship with earthquakes. In conventional studies, seasonality has been removed by extracting normal variations from the normal period set in the analyzed data. However, the results should depend on the setting of the normal period (Huang et al., 2024).
In this study, the SARIMA-GARCH model was applied to atmospheric radon concentration variation data measured at Fukushima Medical University (N37.7°, E140.5°) to quantify anomalies. The SARIMA-GARCH model consists of the SARIMA model, which is one of the linear time series models including seasonality, and the GARCH model, which assumes standard deviation variations. This SARIMA-GARCH model can reduce seasonality more flexibly than the conventional method, and there is no ambiguity in the model selection. Therefore, more appropriate results may be obtained with the model.
The standard deviation variation obtained by the model is corresponding with the results of previous studies that analyzed the data, and standard deviation tends to increase before or after an earthquake. In the M-D space consisting of magnitude (M) and epicentral distance (D), the earthquakes with a maximum peak of standard deviation before the earthquake distribute in the region with a large epicentral distance (D). It is suggested that the variation of radon concentration in the atmosphere before the earthquake may differ depending on the epicentral distance(D).
In this study, the SARIMA-GARCH model was applied to atmospheric radon concentration variation data measured at Fukushima Medical University (N37.7°, E140.5°) to quantify anomalies. The SARIMA-GARCH model consists of the SARIMA model, which is one of the linear time series models including seasonality, and the GARCH model, which assumes standard deviation variations. This SARIMA-GARCH model can reduce seasonality more flexibly than the conventional method, and there is no ambiguity in the model selection. Therefore, more appropriate results may be obtained with the model.
The standard deviation variation obtained by the model is corresponding with the results of previous studies that analyzed the data, and standard deviation tends to increase before or after an earthquake. In the M-D space consisting of magnitude (M) and epicentral distance (D), the earthquakes with a maximum peak of standard deviation before the earthquake distribute in the region with a large epicentral distance (D). It is suggested that the variation of radon concentration in the atmosphere before the earthquake may differ depending on the epicentral distance(D).