Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.685549
Title: The potential of exchangeability for understanding magmatic processes and forecasting eruptive activity
Author: Sheldrake , Tom
ISNI:       0000 0004 5915 4565
Awarding Body: University of Bristol
Current Institution: University of Bristol
Date of Award: 2015
Availability of Full Text:
Access through EThOS:
Abstract:
This thesis investigates the strengths, and potential pitfalls, of grouping observations from multiple volcanoes to understand magmatic and volcanic activity, and ultimately, to forecast eruptive activity and volcanic hazards. Crucial to the analysis and discussion is the principle of exchangeability, where volcanoes are analysed on an "all other things being equal" basis. Specifically, this research concerns issues regarding how volcanoes should be grouped together and what techniques can be used to analyse hierarchically structured data. Firstly, the thesis considers this from a statistical perspective, by introducing the methodology of hierarchical Bayes. The results suggest that simply aggregating multiple eruptive records may inaccurately capture the frequency of eruption magnitudes and volcanic phenomena. Secondly, analysis of historical eruptive activily at lava-dome building volcanoes indicate thal volcanoes of the same broad classification can be subdivided based upon empirical evidence, including the pattern of eruptive activity, monitoring observations and petrological data. However, based on further analysis, experts struggle to interpret these data independently in long-term forecasts: an expert elicitation technique is used to forecast repose duration at Soufriere Hills Volcano, with experts unable to identify behaviour that is atypical for a particular volcano. Survival analysis is investigated as a technique to analyse multiple eruptive records. Results indicate that whilst volcanoes may exhibit similar patterns of eruptive activity, they cannot necessarily be fitted using the same statistical model, suggesting it is not correct to simply group datasets that appear similar. Finally, the 0ppOltunity to use exchangeability to understand magmatic processes is discussed, by exploring differences in the pattern of behaviour at volcanoes of the same broad class. This allows the behaviour of lava-dome building volcanoes to be examined using well -established and novel paradigms for magmatic systems.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.685549  DOI: Not available
Share: