Dissertation project: Quantitatively defining monitoring parameter thresholds for volcanic forecasting: an application to Mount Etna using BET_EF. Supervised by Dr Andrew Bell, Dr Mark Naylor and Prof Ian Main
BET_EF is an example of a probabilistic forecasting model that is able to incorporate both aleatoric and epistemic uncertainties. When monitoring data is loaded into the model each monitoring parameter requires an upper and a lower threshold. At present these thresholds are set qualitatively by expert opinion, and it was the aim of this project to find a way to define these quantitatively. This was done by generating synthetic earthquakes and eruptions using Python and finding the point at which earthquake rates increase from the background rate, i.e. the threshold point. Once threshold points were selected for the synthetic data, the method was then applied to data from Mount Etna.
Above you can see graphs of the daily rate of synthetic earthquakes before a synthetic eruption. On the right graph threshold points are selected qualitatively using Python.