Despite being a natural ecological process, wildfires are dramatic events that, accelerated by global change, could negatively affect ecosystem services depending on their severity level. However, because of data processing constraints, fire severity has been mostly neglected in risk analysis (especially at regional levels). Indeed, previous studies addressing fire severity focused mainly on analyzing single fire events, preventing the projection of the results over large areas. Although, building and projecting robust models of fire severity to integrate into risk analysis is of main importance to best anticipate decisions. Here, taking advantage of free data-processing platforms, such as Google Earth Engine, we use more than 1000 fire records from Western Italy and Southern France in the years 2004–2017, to assess the performance of random forest models predicting the relativized delta normalized burn ratio (rdNBR) used as proxy of fire severity. Furthermore, we explore the explanatory capacity and meaning of several variables related to topography, vegetation, and burning conditions. To show the potentialities of this approach for operational purposes, we projected the model for one of the regions (Sardinia) within the study area. Results showed that machine learning algorithms explain up to 75% of the variability in rdNBR, with variables related to vegetation amount and topography being the most important. These results highlight the potential usefulness of these tools for mapping fire severity in risk assessments.

Global change is increasing concerns regarding fire risk, especially for the wildland–urban interface (WUI). Indeed, different cross-national funded projects aim to develop new methodologies and technologies for assessing and reducing fire risk at the WUI. Traditional approaches focused on multi-criteria decision analysis usually neglecting the interaction of multiple fine-scale components. Here, within the framework of the Italy–France Maritime INTERMED project, we develop a new fire risk assessment methodology following the IPCC guidelines, i.e., considering the three main risk determinants: hazard, exposure and vulnerability. Each single risk determinant integrates multiple components which were estimated using high-resolution data from the European Spatial Agency (ESA), field data, fire spread simulators and local administrations’ geo-data. Hazard components run at the local scale, involving weather and fuel moisture conditions, and also ignition patterns, whereas fire exposure components integrate both percolation and spot potentials (fire pathways and micro-scale land flammability) and postfrontal combustion (residential fuels). Finally, vulnerability components include the type of house and its configuration but also the existence of protection devices. Preliminary results suggest that this methodology might help fire risk planning in WUIs at both landscape and community scales.

The study of wildfire’s causes and ignition patterns, which in the Mediterranean area, are mainly directly or indirectly linked to human activities, is essential for designing effective wildfire risks reduction measures and to mitigate the related impacts on ecosystems and livelihoods

Wildfire ignition patterns are a key factor in fire regimes and, thus, increasing knowledge about where and why wildfires occur is essential to implement more effective prevention strategies. Furthermore, spatial predictions of potential fire ignition patterns might help to better allocate the economic efforts for prevention and to develop wildfire policies. Despite this, few studies addressed ignition patterns within the analysis of fire regimes in detail. This is often due to the scarcity and coarse resolution of data, e.g., ignition causes aggregated by municipalities or provinces that prevent low-scale studies or management actions. In this work, we propose a two-level approach combining spatial coarse-scale models of fire causes with high-resolution ignition prediction maps to overcome the aforementioned limitations. In the framework of the Italy–France Marittimo MED-Star project, two datasets relating to Corse (France) were obtained and combined: (1) fire causes from Promethée and (2) the spatial ignition points of Goliat database. The random forest algorithm was used for modelling the probability of fire causes based on socioeconomic and environmental factors (retrieved from Copernicus LMS and INSEE’s databases). Later, a recommendation system algorithm was applied to calibrate an ignition prediction model based on local conditions at a finer scale. Finally, both models were spatially projected over the territory and then, the probability of each cause (resampled at ~50 m) was multiplied by the ignition prediction map to obtain a high-resolution likelihood for each cause. This work aimed to be the first step to find which algorithms give better results in processing data on wildfire ignition points and causes, together with project fire ignition prediction maps. Results suggest that more extensive and detailed databases could improve result accuracy and that agencies’ data collection on fire ignition points and causes could lead to a better understanding of fire ignition patterns and their variety.

The Mediterranean region is considered one of the “hot spots” of climate change, recording a warming 20% greater than the global average, with a substantial reduction in precipitations