S1.10: Volcano monitoring and eruption forecasting in the presence of uncertainty
Convener(s)
Andrew Bell
University of Edinburgh, United Kingdom
a.bell@ed.ac.uk
Laura Sandri
INGV, Italy
laura.sandri@ingv.it
Mark Bebbington
Massey University, New Zealand
m.bebbington@massey.ac.nz
Geological, geophysical and geochemical monitoring data provide the best insights we have into the status of a volcanic system. However, forecasts of the timing, location, size, and style of eruption based on these data are fundamentally uncertain. A statistical approach is required to work with them, and information useful to decision makers. Forecast uncertainty arises for a number of reasons.The physical and chemical processes controlling eruptive behaviour are inherently stochastic. Monitoring data is limited, ambiguous, and erroneous. Geological records are incomplete. And our models that relate changes in any of these to the likelihood, timing, and nature of future activity are wrong. Consequently,more reliable and useful quantitative forecasting will require developments in a range of statistical methods and understanding.This session is looking for contributions that address statistical issues in volcano monitoring and eruption forecasting. Topics could include: optimization of monitoring networks (for single volcanoes or across volcanic regions) to provide most useful forecasting information; approaches to deal with an absence of baseline monitoring data; forecasting changes in eruption style or the end of eruption; adjusting forecasts to account for missing data; the integration of ‘physics-based’ and empirical forecasting models; and tools to allow better decisions to made on the basis of uncertain forecasts.
Core connection between the proposed session and societal risk mitigation:Eruption forecasting can be a key component of risk management strategies,allowing timely measures to reduce societal risk, such as evacuations or land use and infrastructure planning. However, forecasts are uncertain, and decision making under these conditions is challenging. As a community of scientists and risk managers, in order to make better decisions, we need improved understanding of the nature of eruption forecasting methods, the data on which they are based,and their uncertainties.
This session is sponsored by the IAVCEI Commission on Statistics In Volcanology.