Project Description
Supervisors
Dr Matthew Jones, Environmental Sciences, University of East Anglia – contact me
Professor Timothy Osborn, UEA ENV
Professor Stephen Sitch, University of Exeter Geography
Dr Chantelle Burton, Met Office Hadley Centre
Scientific Background
Megafires, characterised by their extraordinary size, speed, and intensity, are increasingly threatening society, ecosystems, and ecosystem services such as carbon storage (1-4). Recent advances in satellite observations and machine learning provide novel opportunities to study extreme fires on a global scale. In a changing climate, megafire-prone conditions could become more prevalent (3-4). However, the key mechanisms that promote or inhibit megafires are under-studied for most regions globally.
This project addresses critical knowledge gaps by combining novel observations of individual fires globally (5) and climate datasets with machine learning to predict megafire occurrence. The successful candidate will contribute to a ground-breaking efforts to forecast megafire risk and identify land management or policy factors with potential to mitigate that risk.
Research Questions:
- Are megafires becoming more frequent globally, and in which regions?
- Which weather, landscape and land use factors promote or inhibit megafire development?
- Has climate change increased megafire risk, and how could those risks evolve in the future?
Methodology
Supported by the supervisory team, the researcher will:
- Develop a comprehensive global dataset of individual fires, compiling meteorological and landscape variables with potential to influence megafire development, building on the Global Fire Atlas (4).
- Identify megafires: Regionally distinguish between megafires and more ‘typical’ fires with less potential for catastrophic impact.
- Diagnose megafire-prone conditions: Harness machine learning techniques to identify key factors promoting/inhibiting megafire. Disentangle the roles of weather, landscape, and human factors influencing ignition and suppression.
- Analyse regional trends in megafire potential: Study regional trends in observed megafire occurrence (since ~2000s) and megafire-prone weather (since ~1980s), with opportunity to contribute to major reports on the topic (2,4).
Training and Development
Training will maximise future employability in academia and industry:
- Programming and geospatial data analysis using Python/R.
- Machine/deep learning techniques.
- Communication of scientific findings through publications and conferences.
Person Specification
A highly motivated candidate with:
- A degree or equivalent in numerate, computational, or environmental subject areas.
- Experience with programming languages such as Python or R for scientific data analysis is desirable.
Further Information: