ARIMA Model Analysis of the Regularities of Earthquake Origin Times in the Longmen Mountain Fault Zone

Abstract

Based on earthquake catalog data from the Longmen Mountain fault zone over the past 10 years, we constructed series of earthquake origin time intervals by grouping according to the magnitude (M) and use the ARIMA model for analysis with a 9:1 ratio of fitting-training and prediction-verification data. We found that the series of both M ≥ 2.5 and M ≥ 3.0 showed the variation of nesting with short, medium and long periods. By further predictive verification and comparative analysis, the optimal prediction models for each series were obtained: ARIMA(10,2,1)×(0,1,1)20 direct prediction model for series of M ≥ 2.5, ARIMA(8,2,1)×(0,1,1)40 rolling prediction model for M ≥ 3.0, and ARIMA(1,2,3)×(0,1,1)3 rolling prediction model for M ≥ 4.5. The predicted results suggested that the seismicity of the Longmen Mountain fault zone has a recent gradually weakening trend. This analysis process provides an effective reference and method for studying the time regularities of tectonic earthquake occurrence.