Abstract
Long term forecasting of volcanic eruptions can be challenging due to the small amount of data available in the eruption records. The data from volcanoes which share similar physical properties and statistical behaviour (statistical analogues) has been used to help estimate model parameters and forecast future eruptions at a target volcano. This project used a series of hierarchical renewal processes and trend renewal processes, as a way to forecast the next volcanic explosivity index (VEI) greater than or equal to 3 eruption at Mt Taranaki by using the eruption records from Mt Taranaki and different sets of statistical analogue volcanoes. Model averaging was used to combine the posterior distribution of the forecast times from each of the considered models to allow for model uncertainty. Eruption records are likely to be missing data especially in earlier years, therefore analysis is carried out using three different time periods. Using one method of model averaging and the data from the last 10,000 years, we forecast a 50% chance that the next VEI greater than or equal to 3 eruption will occur before the year 2190 with a 95% chance it will occur between the years 2028 and 3039.