An empirical earthquake prediction model

We have monitored seismic activity induced by impoundment of Lake Jocassee in northwest South Carolina for about two years. Low-level shallow activity was recorded. The larger felt events (2.0 ? M_L ? 2.6) were found to be associated with precursory changes in one or more of the following; number of events, t_S/t_p ratio values and radon concentrations in groundwater.The microearthquakes in the precursory period were accurately located in time and space, and their location pattern was used to develop an empirical earthquake prediction model.The precursory period consists of two phases; α-phase or a period of slow (or no) increase in seismicity, and β-phase, a period when the activity increase is more rapid. The main shock was found to be located within a cluster, a “target” area defined by the location of events in the β-phase. There is a general absence of seismic activity in the “target” area in the α-phase. The main shock occurred soon after a period of quiescence in the seismic activity in the β-phase. The magnitude of the shock, M_L is given by: M_L = 2 log D ? 0.07, where D is the duration of the precursory period in days.The model was successfully tested with data for a magnitude 2.3 event on February 23, 1977 which was also accompanied by radon and t_s/t_p anomalies.