S2.6: Uncertainty quantification in volcanic phenomena: an essential component for modeling physical processes and for hazard/risk assessment


Alessandro Tadini

Laboratoire Magmas et Volcans, Université Clermont Auvergne, Aubiere Cedex, France

Andrea Bevilacqua

Istituto Nazionale di Geofisica e Vulcanologia, Pisa, Italy

Pablo Tierz

British Geological Survey, The Lyell Centre, Edinburgh, United Kingdom

Mary Grace Bato

NASA Jet Propulsion Laboratory, Pasadena, CA, United States of America

Sebastien Biasse

Earth Observatory of Singapore, Nanyang Technological University, Singapore

Gabrielle Tepp

U.S. Geological Survey, Alaska Volcano Observatory, Anchorage, AK, USA

Samantha Engwell

British Geological Survey, The Lyell Centre, Edinburgh, United Kingdom

Volcanic phenomena are affected by a high degree of uncertainty, both epistemic (i.e. related to incomplete knowledge of the phenomena themselves)and aleatoric (i.e. linked to the physical variability typical of complex natural systems). Uncertainty quantification (UQ) is a fundamental task in hazard and risk assessment (e.g. for emergency management and long-term planning)and is essential for making advances in modeling physical processes. UQ has a significant effect on the solution to many different problems in volcanology, both the inverse problems aimed at the reconstruction of past events and the forward problems aimed at the forecasts of future events.

These problems include:


  • The calculation of eruptive parameters, such as the mass of different volcanicphenomena (fallout, PDC, etc.), the mass flow rate at the eruptive vent/fissure,and the maximum or average plume height. The uncertainty in this case definesthe probability density function of input parameters to the models of volcanicprocesses.
  • The definition of the behavior of the volcano, including the spatial location oferuptive vents, the temporal estimates of eruption onset and duration, and theprobability of different eruptive styles and/or hazardous phenomena.
  • The modeling of volcanic phenomena, especially in those approaches wheregreat simplifications have been introduced to allow the reduction of computationaltimes (e.g. 1-D integral plume models; Gaussian Tephra transport and dispersalmodels; kinetic, integral, or depth-averaged mass flow models). UQ is inthis case crucial to define the limits and the advantages of each model, throughthe comparison with past data.


UQ can be performed with different approaches, including the application of expert judgement techniques, the comparison of different sampling/integration techniques for measuring field data, the employment of different multi-model procedures and modeling benchmarks for numerical simulations, stochastic processes, event trees,and Bayesian networks. In this session we welcome contributions that cover this wide spectrum of UQ of volcanic phenomena, with a specific focus on those studies focused on modeling of physical processes and/or those which provide a direct application of the results to hazard/risk assessments (e.g. hazard or risk maps obtained through approaches that consider all the above mentioned problems).

Scroll to Top