Overcoming data scarcity for probabilistic eruption forecasting at data-limited volcanoes
Data scarcity is one of the main sources of uncertainty in eruption forecasting. Due to multiple factors, such as the location or the poor preservation of deposits, numerous volcanoes have incomplete records of eruptions. In this thesis, I explore statistical approaches to overcome data scarcity whe...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis-Doctor of Philosophy |
| Language: | English |
| Published: |
Nanyang Technological University
2023
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/166192 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | Data scarcity is one of the main sources of uncertainty in eruption forecasting. Due to multiple factors, such as the location or the poor preservation of deposits, numerous volcanoes have incomplete records of eruptions. In this thesis, I explore statistical approaches to overcome data scarcity when forecasting eruptions at data-limited volcanoes. Firstly, I analysed first recorded eruptions from Holocene volcanoes to provide relative completeness dates while discussing the factors causing eruption under-recording. Secondly, I estimated the average recurrence interval of first recorded eruptions at a regional level and the annual probability of eruption at individual potentially active volcanoes from several regions of Asia-Pacific. Lastly, I identified analogues and estimated the f-M relationship for Melimoyu (Chile) by applying a quantitative and objective approach based on hierarchical clustering and probabilistic modelling. This thesis provides insights into the challenges of forecasting eruptions at under-studied volcanoes and presents numerical solutions for overcoming data scarcity. |
|---|