Efficient simulation of wildfire spread on an irregular grid

Abstract

A cell-based wildfire simulator that uses an irregular grid is presented. Cell-based methods are simpler to implement than fire front propagation methods but have traditionally been plagued by fire shape distortion caused by the fire only being able to travel in certain directions. Using an irregular grid randomises the error introduced by the grid, so that the shape of simulated fire spread is independent of the direction of the wind with respect to the underlying grid. The cell-based fire spread simulator is implemented using discrete event simulation, which is a much more efficient computational method than conventional wildfire simulation techniques because computing resources are not used in repeatedly computing small updates to parts of the fire whose dynamics change infrequently, namely those areas of a fire that move slowly. The resulting simulator is comparable in accuracy with traditional fire front propagation schemes but is much faster and can therefore be used as an engine for fire simulation applications that require large numbers of simulations, such as in the role of a risk analysis engine.