5:15 PM - 6:30 PM
[MIS08-P01] Detecting clustered pre-earthquake anomalies of borehole strain network by Receiver Operating Characteristic Curve
★Invited Papers
Keywords:clustered pre-earthquake anomalies, Receiver Operating Characteristic, short-term earthquake forecasting
The preparatory process and the occurrence of shallow earthquakes are usually accompanied by crustal deformation. In recent years, the correlation between monitoring strain anomalies and local major earthquakes has been verified, however, whether these anomalies contain precursory information and their use in forecasting of major earthquakes were not evaluated.
In this study, we evaluate the possibility of strain anomalies containing earthquake precursors by using Receiver Operating Characteristic (ROC) prediction. First, strain network anomalies were extracted in the borehole strain data recorded in Western China during 2010-2017. Then, we considered the clustering characteristics of the anomalies and proposed a new prediction strategy characterized by the number of network anomalies in an anomaly window, N_ano, and the length of alarm window, T_alm, as shown in Fig. 1. We assumed that clusters of network anomalies indicate a probability increase of an impending earthquake, and consequently, the alarm window would be the duration that a possible earthquake would occur. The Area Under the ROC Curve (AUC) between true predicted rate, tpr, and false alarm rate, fpr, is measured to evaluate the efficiency of the prediction strategies.
First, we found that the optimal strategy of short-term forecasts has been established with AUC=0.82, by setting the number of anomalies greater than 7 within 14 days and alarm window at 1 day, as shown in Fig. 2. The results further show the prediction strategy performs significantly better when there are frequent enhanced network anomalies prior to the larger earthquakes surrounding the strain network region. Next, we tested the influence of different proportions of anomalies in an anomaly window under the condition of N_ano=7 and T_alm=1. The results are as shown in Fig. 3, when there were more than half (2N) or a third (3N) of network anomalies in an anomaly window, the detection efficiency was more significant and stable. When there were few anomalies in an anomaly window, the detection efficiency was low and especially when T_ano=8 N_ano, the result seems to be accidental and irregular. This result shows that clusters of anomalies indicate an increased probability of an earthquake, which is also consistent with the physical assumption of the prediction model.
In summary, we discussed whether the prediction strategy can be used in short-term forecasting and further demonstrated the influences of different proportions of anomalies in an anomaly window on short-term earthquake forecasting using AUC values. We confirmed that clusters of network anomalies may contain precursory information related to the earthquakes, and highlighted the potential for short-term earthquake forecasting.
In this study, we evaluate the possibility of strain anomalies containing earthquake precursors by using Receiver Operating Characteristic (ROC) prediction. First, strain network anomalies were extracted in the borehole strain data recorded in Western China during 2010-2017. Then, we considered the clustering characteristics of the anomalies and proposed a new prediction strategy characterized by the number of network anomalies in an anomaly window, N_ano, and the length of alarm window, T_alm, as shown in Fig. 1. We assumed that clusters of network anomalies indicate a probability increase of an impending earthquake, and consequently, the alarm window would be the duration that a possible earthquake would occur. The Area Under the ROC Curve (AUC) between true predicted rate, tpr, and false alarm rate, fpr, is measured to evaluate the efficiency of the prediction strategies.
First, we found that the optimal strategy of short-term forecasts has been established with AUC=0.82, by setting the number of anomalies greater than 7 within 14 days and alarm window at 1 day, as shown in Fig. 2. The results further show the prediction strategy performs significantly better when there are frequent enhanced network anomalies prior to the larger earthquakes surrounding the strain network region. Next, we tested the influence of different proportions of anomalies in an anomaly window under the condition of N_ano=7 and T_alm=1. The results are as shown in Fig. 3, when there were more than half (2N) or a third (3N) of network anomalies in an anomaly window, the detection efficiency was more significant and stable. When there were few anomalies in an anomaly window, the detection efficiency was low and especially when T_ano=8 N_ano, the result seems to be accidental and irregular. This result shows that clusters of anomalies indicate an increased probability of an earthquake, which is also consistent with the physical assumption of the prediction model.
In summary, we discussed whether the prediction strategy can be used in short-term forecasting and further demonstrated the influences of different proportions of anomalies in an anomaly window on short-term earthquake forecasting using AUC values. We confirmed that clusters of network anomalies may contain precursory information related to the earthquakes, and highlighted the potential for short-term earthquake forecasting.