Evaluation of short-term probabilistic eruption forecasting at Whakaari, New Zealand

Abstract

Phreatic explosions at volcanoes are difficult to forecast but can be locally devastating, as illustrated by the deadly 2019 Whakaari (New Zealand) eruption. Quantifying eruption likelihood is essential for risk calculations that underpin volcano access decisions and disaster response. But estimating eruption probabilities is notoriously difficult for sudden onset eruptions. Here, we describe two retrospectively developed models for short-term (48 h) probabilistic forecasting of phreatic eruptions at Whakaari. The models are based on a pseudo-prospective analysis of seven Whakaari eruptions whose precursors were identified by time series feature engineering of continuous seismic data. The first model, an optimized warning system, could anticipate six out of seven eruptions at the cost of 14 warning days each year. While a warning is in effect, the probability of eruption is about 8% in 48 h, which is about 126 times higher than outside the warning. The second model used isotonic calibration to translate the output of the forecast model onto a probability scale. When applied pseudo-prospectively in the 48 h prior to the December 2019 eruption, it indicated an eruption probability up to 400 times higher than the background. Finally, we quantified the accuracy of these seismic data-driven forecasts, alongside an observatory expert elicitation that used multiple data sources. To do this, we used a forecast skill score that was benchmarked against the average rate of eruptions at Whakaari between 2011 and 2019. This exercise highlights the conditions under which the three different forecasting approaches perform well and where potential improvements could be made.

Volcanic Early Warning Using Shannon Entropy: Multiple Cases of Study

“…As Kilburn (2018) indicated ground deformation, volcanic seismicity (mainly VT events) and their associated energy release could be considered one of the most reliable pre‐eruptive evidences. As is broadly known, scientific community has been using the energy of the seismic signal to forecast volcanic eruptions (Boué et al., 2016; Ortiz et al., 2003; Power et al., 2013) and recently some variations of the use of the energy have obtained promising results (Caudron et al., 2021; Dempsey et al., 2020, 2022). Despite of several successfully results, some limitations were found.…”

Using Template Matching to Detect Hidden Fluid Release Episodes Beneath Crater Lakes in Ruapehu, Copahue, and Kawah Ijen Volcanoes

“…To isolate a continuous tremor signal, we needed to remove signatures from regional and volcano‐tectonic (VT) earthquakes. To do this, we applied an outlier detection algorithm to the raw waveform data prior to calculating RSAM, MF, and HF (Dempsey et al., 2022). Wave trains from earthquakes perturb the average velocity record at stations above the background.…”

“…However, contamination from tele seismic waves from large‐magnitude long‐duration subduction earthquakes still occurred. To remove these, we applied a two‐window (20 min), moving minimum (Dempsey et al., 2022).…”

“…High values of DSAR have been inferred to correlate with high gas levels in the edifice, suggesting either reduced fluid motion and/or trapping that has led to a gas-accumulation. With recent advances in seismic data processing, we have been able to leverage these large, continuous data streams for latent patterns that may be precursors to eruptions (Ardid et al, 2022;Dempsey et al, 2020Dempsey et al, , 2022.…”

“…A normalized time series value of 1 (=10 0 ) indicates the mean signal strength whereas a value of 100 (=10 2 ) indicates a strong signal that is two standard deviations above the mean. Dempsey et al (2022) have shown that such normalization improves the discrimination of precursors and accuracy when forecasting eruptions. Alternative normalization techniques based on station distance could be implemented (e.g., Aki & Koyanagi, 1981;Thompson & West, 2010) but were not tested here.…”