New paper: Volcano Evacuation with an ABM

New Paper: Estimating Spatio-Temporal Risks from Volcanic Eruptions Using an Agent-Based Model

My colleague, Jumadi, has just had a great paper published on modelling peoples’ risk of harm during an emerging crisis. He developed an agent-based model set on the island of Merapi, Indonesia, that estimates how people will move before and during a volcanic eruption, with their risk changing dynamically as the eruption evolves. The paper is open access and available in the Journal of Artificial Societies and Social Simulation.

Jumadi, J, N. Malleson, S. Carver, and D. Quincey (2020). Estimating Spatio-Temporal Risks from Volcanic Eruptions Using an Agent-Based Model. Journal of Artificial Societies and Social Simulation 23 (2): 2. DOI: 10.18564/jasss.4241

Image illustrating how the volcano hazard changes over time


Managing disasters caused by natural events, especially volcanic crises, requires a range of approaches, including risk modelling and analysis. Risk modelling is commonly conducted at the community/regional scale using GIS. However, people and objects move in response to a crisis, so static approaches cannot capture the dynamics of the risk properly, as they do not accommodate objects’ movements within time and space. The emergence of Agent-Based Modelling makes it possible to model the risk at an individual level as it evolves over space and time. We propose a new approach of Spatio-Temporal Dynamics Model of Risk (STDMR) by integrating multi-criteria evaluation (MCE) within a georeferenced agent-based model, using Mt. Merapi, Indonesia, as a case study. The model makes it possible to simulate the spatio-temporal dynamics of those at risk during a volcanic crisis. Importantly, individual vulnerability is heterogeneous and depends on the characteristics of the individuals concerned. The risk for the individuals is dynamic and changes along with the hazard and their location. The model is able to highlight a small number of high-risk spatio-temporal positions where, due to the behaviour of individuals who are evacuating the volcano and the dynamics of the hazard itself, the overall risk in those times and places is extremely high. These outcomes are extremely relevant for the stakeholders, and the work of coupling an ABM, MCE, and dynamic volcanic hazard is both novel and contextually relevant.