Towards physics-informed machine learning for wildfire simulations
A spatial temporal conditional autoregressive model
G.B. Bultema (TU Delft - Technology, Policy and Management)
O. Kammouh – Graduation committee member (TU Delft - System Engineering)
P. H.A.J.M.van Gelder – Graduation committee member (TU Delft - Safety and Security Science)
María Nogal – Graduation committee member (TU Delft - Integral Design & Management)
Xiao Liu – Graduation committee member (Georgia Institute of Technology)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
Wildfire simulations have become increasingly important as their frequency and severity increases, posing a threat to communities and resulting in billions in damages. However, current wildfire models face a trade-off between accuracy and computational complexity. Current wildfire models can be physics-based, but computationally expensive, or based on empirical data, which allows for better computational speed, but decreases the physical basis. A model combining the computational speed of empirical models with the physics understanding of physics-based models to increase the accuracy of wildfire simulations is desired. This master thesis explores the design of such a model by answering the question: ’How can a near real-time wildfire simulation be designed using physics-informed machine learning?’....