Geo-INQUIRE installation: PVHA – Probabilistic Volcanic Hazard Assessment (TA2-541-1/TA2-541-2)

Hosting team: Dr. Laura Sandri, Dr. Antonio Costa, Dr. Beatriz Martínez Montesinos (INGV Bologna, Italy), Prof. Dr. Arnau Folch (Geociencias Barcelona (GEO3-BCN), CSIC, Spain)

Principal investigator: Paul Jarvis (Earth Sciences New Zealand)

Project title: Probabilistic tephra fall hazard assessment for New Zealand’s most active onshore volcano

Project acronym: TEPHRANZ

Project report ID: C1-TA2-541-1-1 (1^st Call)

Date of visit: 26 May – 13 June 2025

Geo-INQUIRE Virtual Access: The project is ongoing but the results (output of simulations) will be dropped in the Geo-INQUIRE Simulation Data Lake.

Data/Products: In-progress; we will be producing output and results of Fall3D simulations.

Project report:

Previous months-long volcanic eruptions from Mt. Ruapehu in 1945 and 1995-1996 have led to ashfall across the North Island of New Zealand (NZ). During such sustained periods of activity, ashfall forecasts are essential for Civil Defence Emergency Management (CDEM) Groups, infrastructure managers and communities to facilitate decision making around mitigation activities. Currently, short-term (hours-days) forecasts of ashfall in NZ are deterministic, i.e., one or few representative eruptions are simulated. Whilst this enables rapid computation, it does not account for uncertainty in the model inputs, potentially providing decision makers with inaccurate information. In this project, we are exploring a probabilistic approach, whereby large numbers of ashfall simulations are performed for different model input parameters. Although such ensembles are much more computational intensive than a small number of deterministic simulations, recent years have seen significant improvements in both model and access to computing capabilities, potentially enabling probabilistic forecasting on a response timescale.

Taking the 2022 unrest episode of Ruapehu as a case study, we are investigating how daily probabilistic ashfall forecasts could have been produced during the event by integrating a) Ruapehu’s long-term eruption frequency-magnitude relationship, b) the short-term probability of eruption as determined from an existing Bayesian Network, and c) ensemble-based simulations of volcanic ashfall. For the third component, we are using the high-performance computing (HPC) facilities provided by the Geo-INQUIRE Transnational Access to perform large numbers of parallelised ashfall simulations using the Fall3D software. The outputs from our developed workflow are maps of probabilistic ashfall hazard on different days of the 2022 unrest period (Figure 1).

The second component of this project is a feasibility study for the operational use of this workflow in NZ. By exploring different possible implementation configurations, e.g., model resolution, ensemble size, grain size classes, we are assessing what can be achieved with different levels of computational resource. The results from this project will inform future decisions about operational ashfall forecasting capability in NZ.