Predicting wildfire spread and behaviour in Mediterranean landscapes

Abstract

The use of spatially explicit fire spread models to assess fire propagation and behaviour has several applications for fire management and research. We used the FARSITE simulator to predict the spread of a set of wildfires that occurred along an east–west gradient of the Euro-Mediterranean countries. The main purpose of this work was to evaluate the overall accuracy of the simulator and to quantify the effects of standard vs custom fuel models on fire simulation performance. We also analysed the effects of different fuel models and slope classes on the accuracy of FARSITE predictions. To run the simulations, several input layers describing each study area were acquired, and their effect on simulation outputs was analysed. Site-specific fuel models and canopy inputs were derived either from existing vegetation information and field sampling or through remote-sensing data. The custom fuel models produced an increase in simulation accuracy, and results were nearly unequivocal for all the case studies examined. We suggest that spatially explicit fire spread simulators and custom fuel models specifically developed for the heterogeneous landscapes of Mediterranean ecosystems can help improve fire hazard mapping and optimise fuel management practices across the Euro-Mediterranean region.

Optimizing prescribed fire allocation for managing fire risk in central Catalonia

“…9) could also be compared with existing management plans and historical wildfires to identify particular landscape features that could contribute to the design and refine the location of SMP for fuel treatments in Catalonia (Costa et al, 2011). For instance, recurrent long-distance spreading fire events burning under particular weather conditions provide interesting baseline information to characterize the most frequent synoptic scenarios associated with catastrophic events (Duane et al, 2015;Pereira et al, 2005;Rasilla et al, 2010), and the fire behavior that led to them (Duane et al, 2016;Salis et al, 2016a).…”

“…We used FlamMap for fire spread and behavior modeling (Finney, 2006). FlamMap has been widely used for landscape scale wildlife exposure and risk assessment in studies worldwide, including southern EU Mediterranean countries (Alcasena et al, 2016a;Elia et al, 2014;Jahdi et al, 2016;Mallinis et al, 2016). The landscape input data were constructed with topography, surface fuel and canopy metric grids (Ager et al, 2011).…”

Assessing Wildland Fire Risk Transmission to Communities in Northern Spain

“…ANN performed well and facilitated the generation of a high-resolution ignition probability grid. Understanding how fire weather and geospatial variables associated with anthropic activities can explain fire occurrence has been conducted in previous works [74,75].…”

Generation and evaluation of an ensemble of wildland fire simulations

“…More sophisticated methods exist, for instance some use the distance between the vertices of the fire perimeter [Fujioka(2002), Duff et al(2012)] while others consist in computing scores based on information on the dynamics of the simulated and observed fire surfaces [Filippi et al(2013)]. Evaluation of model per-formance with such scores can be performed on fire cases by running the fire spread simulations using data known in hindsight, such as knowledge regarding fire suppression actions or observed (sometimes corrected) weather data (e.g., [Duff et al(2018), Salis et al(2016)]). Another possibility is to run the simulation based on data that are available at the time of fire start (e.g., [Filippi et al(2014)]).…”