The duration of volcanic eruptions: empirical probabilistic forecasting models based on historic eruption data
The ability to forecast future volcanic eruption durations would greatly benefit emergency response planning. A probabilistic model to forecast the duration of eruptions is presented here. The model relies on past eruptions being a good indicator of future activity. Datasets of historic eruptions fr...
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | unknown |
| Published: |
The Open University
2014
|
| Subjects: | |
| Online Access: | https://dx.doi.org/10.21954/ou.ro.0000f83d http://oro.open.ac.uk/id/eprint/63549 |
| Summary: | The ability to forecast future volcanic eruption durations would greatly benefit emergency response planning. A probabilistic model to forecast the duration of eruptions is presented here. The model relies on past eruptions being a good indicator of future activity. Datasets of historic eruptions from Mt. Etna (flank only), Kilauea, Piton de la Foumaise (PdlF) and Iceland have been compiled through a critical examination of existing literature and careful consideration of uncertainties in reported dates. The eruptions from Mt. Etna, Kilauea and PdlF are all basaltic effusive eruptions, however, the Icelandic dataset is more diverse and seven types of duration have been identified and are assessed independently. These datasets have also enabled an assessment of repose intervals (eruption end to eruption start) to be conducted. Eruption duration and repose interval data are modelled using exponential, Weibull, log-logistic and Burr type XII distributions with parameters found by maximum likelihood estimation. Log-logistic distributions are found to often provide the best-fit to the observed data. Survivor function statistics are applied to the best-fit theoretical distribution of each dataset and used to forecast (a) the probability of an eruption exceeding a given duration, (b) the probability of an on-going eruption (having reached t days) exceeding a specified total duration and (c) the minimum duration associated with a given probability. Eruption duration analyses at individual volcanic systems show systematic variations with time and different time periods have different duration regimes. Comparisons of the erupted volumes associated with these duration regimes show that volume is an important control on eruption duration. Average eruption rates also determine eruption duration and comparisons of data from Kilauea, PdlF and volcanic systems from different regions of Iceland have led to the hypothesis that volcano spreading rate may have an important control on eruption rate and subsequently eruption duration. |
|---|