Forest fires threaten a large part of the world's forests, communities, and industrial plants, triggering technological accidents (Natechs). Forest fire modelling with respect to contributing spatial parameters is one of the well-known ways not only to predict the fire occurrence in forests, but also to assess the risk of forest-fire-induced Natechs. This study is a review of methods based on geospatial information system (GIS) for modelling forest fires and their potential Natechs that have been implemented all over the world. The present study conducts a systematic literature review of the methods used for forest fire susceptibility, hazard, and risk assessment, while dividing them into four general categories: (a) statistical and data-driven models; (b) machine learning models; (c) multi-criteria decision-making models, and (d) ensemble models. In addition, some forest fire detection techniques using satellite imagery are reviewed. A comparison is also conducted to highlight the research gaps and required future research. The results of the present research assist decision makers to select the most appropriate techniques according to specific forest conditions. Results show that data-driven approaches are the most frequently applied methods while ensemble approaches are more accurate.

On February 2007, a massive fire in a propane de-asphalting unit in an oil refinery in Texas, USA happened due to liquid propane release from a cracked pipe in a control station injuring four people, damaging extensive equipment, causing significant business interruption, and resulting in more than $50 million losses. The accident was triggered by a natural hazard: freezing of piping at a control station caused an inlet pipe elbow to crack, which in turn, led to the release of high-pressure liquid propane which was rapidly ignited. In addition, there were two near-miss events due to potential domino effects. In fact, the accident could reasonably have resulted in much more severe consequences due to the exposure of large butane storage spheres and chlorine containers, increasing the possibility of a catastrophic domino effect. This paper develops a Natech (natural hazard triggering technological disasters) risk assessment methodology that relies upon Bayesian network capabilities and takes into account the potential Natech domino effects. The methodology is implemented in the intended refinery and mathematically graphically represents the dynamic cause–effect relations between units involved in the scenario, and handles uncertainties among the interactions. In addition, the methodology can provide a risk value for the entire scenario that can be used further for risk-based decision making.