Empirical recurrence rate time series for volcanism: Application to Avachinsky volcano, Russia

Abstract

The volcanic activity of Avachinsky is assumed to follow a Poisson process and is fingerprinted with a sequence of empirical recurrence rate time series. The last 5 time series are used as a prediction set to check the predictive ability of the candidate model produced by time series modeling techniques. The model is used to forecast future recurrence rates that, in turn, are used to develop the mean function of the volcanic process, which is required to evaluate the probability of future eruptions. At the model validation stage, the candidate model forecasts a mean number of 2.31 eruptions for the prediction set that are close to the actual number of events, which are 2 eruptions. For a full scaled forecast, the model concludes a total of 2.37 new events for the next 25 years for the Avachinsky volcano.