Using stochastic modelling to unravel the eruptive behaviour of the active central volcanoes of São Miguel (Azores)

Abstract Island volcanoes often present incomplete or poorly preserved eruptive records, generating large uncertainties about their past behaviour. São Miguel is one such case, where three active central volcanoes (Sete Cidades, Fogo, and Furnas) have erupted multiple times in the last millennia. However, most eruptive events remain undated or are poorly reconstructed. Here, we estimate inter-event times (IETs), the time intervals between successive eruptive events, using different statistical distributions and compute the probabilities of an eruption occurring in the next 50 yr for each central volcano. To do so, we generated synthetic catalogues by randomly assigning stratigraphic-consistent ages to catalogues of basaltic s.l. and trachytic s.l. events (BT catalogues) and trachytic s.l. events (T catalogues). The datasets were tested for stationarity and several renewal models were applied to evaluate the model that best describes the occurrence of events through time. The results revealed that Sete Cidades data are non-stationary, suggesting an incomplete eruptive record or changes in eruption frequency over time. Mean IETs values are shorter for Sete Cidades (154 yr for BT catalogues; 205 yr for T catalogues) than for Fogo (546 yr for BT; 909 yr for T), and Furnas (318 yr for T). Mean probabilities of an eruption within the next 50 years are higher at Sete Cidades (21% for BT; 17% for T), followed by Furnas (13% for T) and Fogo (8% for BT; 5% for T). This analysis accounts for data uncertainties, providing important clues for forecasting eruptions in volcanoes where the eruptive record may be incomplete.