Agent-based methods for representing anthropogenic fire in Dynamic Global Vegetation Models
Dr James Millington
Dr T Edwards
No more applications being accepted
Funded PhD Project (Students Worldwide)
Feedbacks between fire, vegetation and climate are complex and vary by biome, but are also influenced and shaped by humans through land use and fire management activities. Overestimates of global burned area by the large majority of Dynamic Global Vegetation Models (DGVMs) in the Fire Model Intercomparison Project are primarily due to inadequate representation of the influence of economic development. To improve performance, models of future global anthropogenic fire will need to consider more than just economic and demographic variables, but also issues such as land tenure, capitalisation and land management approaches and how these vary regionally around the globe.
A prime candidate for structuring these models is agent-based modelling (ABM), given its ability to translate empirical understanding about human behaviour and decision-making into computational representations of individual or groups of people. Although there are several examples of landscape-scale (100–10,000 km2) studies using ABM to understand human-fire interactions, applications at regional and global scales are yet to be forthcoming and pose a novel avenue for investigation. Twin technical challenges remain: 1) identifying appropriate representational approaches to capture key human fire-related behaviours in ABM for different regions around the world that could be applied more broadly, and 2) developing technical methods to scale landscape-scale models to broader extents (e.g. via emulation).
This PhD project will aim to advance both by i) creating landscape-scale ABM for case study locations (e.g. Mediterranean Basin, Tropical Africa), ii) developing methods to emulate results from i) to apply at broader scales. Outcomes will improve understanding about how ABM can contribute to improving representation of anthropogenic fire activity in DGVMs such as being developed elsewhere within the Leverhulme Centre.
The studentship will be supervised by Dr James Millington and Dr Tamsin Edwards, both in the Department of Geography at King’s College London. The student will be part of the Environmental Dynamics (ED) and Geocomputation Research Domains within the department. The ED domain seeks a broader and deeper understanding of Earth’s interacting hydrological, geomorphological, atmospheric and ecological processes, while the Geocomputation domain uses data analytics and simulation to investigate pressing challenges for environment and society.
The student will be supported by scholars working in the new Leverhulme Centre for Wildfires, Environment and Society established in 2019. The Centre is comprised of social and natural scientists from Imperial College, Kings College London, Royal Holloway and University of Reading, working together to create a new, integrative science of wildfire. The aim of the Centre is: to understand what factors govern wildfire regimes, including the sources, frequency, intensity, timing, and spatial pattern of fire; develop ways of predicting fire risks that include new biophysical understanding and account more reflexively for human-environment dynamics; quantify the impacts of fire on natural processes and human systems, including assessing their economic consequences and wider cultural meanings; and develop ideas for living with fire. The student will therefore join a vibrant interdisciplinary research community with a joint vision of producing evidence-based understanding of the human-fire nexus that can help inform policy and practice.
The applicant will have a good undergraduate degree in geography, computer science, environmental sciences or an allied field. They will either have, or be working towards, a masters degree or equivalent in a relevant field. The successful candidate will preferably have experience working with agent-based and/or other spatially explicit simulation modelling, emulation techniques, and associated data and coding tools (e.g. R, Java). They will have experience of writing to high standards, and a willingness to work in interdisciplinary teams.
Applicants should submit:
i) A CV (max 2 A4 sides), including details of two academic references;
ii) A cover letter outlining their qualifications and interest in the studentship (max 2 A4 sides);
iii) A sample of written work up to 2000 words.
iv) If possible, some examples of code (e.g. link to GitHub profile or repository)
These should be sent by email to [Email Address Removed] by 22/07/2019 with “Leverhulme ABM PhD” as the subject. Interviews will take place end of July 2019.
For further information on the project, please contact James Millington ([Email Address Removed]) and Tamsin Edwards ([Email Address Removed])
The studentship will be funded at £17,009 stipend per annum including London allowance (rising in line with annual increments) with fees (either Overseas or HEU as appropriate) paid for three years. There will be support funding for conference attendance. The studentship is expected to be available from October 2019 subject to final Centre contract being signed.