Fire

To calculate Fire Risk, we use the following variables [Copernicus Climate Data Store (CDS)]:

• daily near surface air temperature: daily mean temperature (°C) at ground level (2m).
• daily near surface wind speed: daily mean speed (m/s) at ground level (2m) of wind.
• daily near surface specific humidity: daily mean specific humidity at ground level (2m).
• daily sea level pressure: daily mean pressure (Pa) at ground level, corrected for sea level.
• daily precipitation: total daily precipitation (mm). The first step is to convert specific humidity to relative humidity:

$\text{Relative Humidity} = \frac{E_{pa}}{E_{spa}} \times 100$

where $E_{pa}$ is vapor pressure:

$E_{pa} = \frac{\text{Specific Humidity} \times \text{Sea Level Pressure}}{0.622 + (0.378 \times \text{Specific Humidity})}$

and $E_{spa}$ is saturation vapor pressure:

$E_{spa} = 610.94 \times e^{\frac{17.625 \times \text{Air Temperature}}{234.04 + \text{Air Temperature}}}$ Next we calculate the Fire Weather Index (FWI), using the xclim library. The FWI indicator is a forest fire hazard rating system developed in Canada and used internationally. This index does not measure actual fires, but potential hazard based exclusively on weather conditions. FWI synthesizes in a single value the effect of several daily variables: air temperature, relative humidity (calculated from specific humidity and pressure), wind speed and precipitation. To define annual risk level, we count the number of days when FWI exceeds certain hazard thresholds.

We use the following variables:

• N_extreme: Number of days with FWI ≥ 30 (Extreme Hazard).
• N_very_high: Number of days with 20 ≤ FWI < 30 (Very High Hazard).
• N_high: Number of days with 12 ≤ FWI < 20 (High Hazard). The combination of these counts allows assessment of fire hazard frequency and intensity during the year.

The methodology described for Fire Risk assessment represents the global reference standard for assessing fire hazard based on meteorological data.

It is based entirely on the Canadian Forest Fire Weather Index (FWI) System, a scientifically validated model used operationally in dozens of countries worldwide.

The conversion from specific humidity to relative humidity is a standard preliminary step in meteorology. The formulas used are derivations of fundamental physical equations.

• The Computation of Equivalent Potential Temperature The formula employed for $E_{spa}$ is a version of the Clausius-Clapeyron equation, often presented in the Magnus-Tetens or Bolton approximation form. These are standard formulas used in all meteorological models. The core of our method is FWI. This system was developed by the Canadian Forestry Service.
• Development and structure of the Canadian Forest Fire Weather Index System. Canadian Forestry Service This is the document that describes in detail the entire architecture of the FWI system. It explains how input variables (temperature, relative humidity, wind, rain) are used to calculate the six system components: three fuel moisture codes (FFMC, DMC, DC), two fire behavior indices (ISI, BUI) and the final FWI index, which represents fire intensity at the fire front line.

FWI has been adopted by many fire early warning systems worldwide, including the European system. These systems have validated and sometimes adapted hazard thresholds.

• Comprehensive Monitoring of Wildfires in Europe: The European Forest Fire Information System EFFIS uses FWI as the base indicator for mapping fire hazard throughout Europe. The hazard thresholds we use are consistent with standard ones used by EFFIS and other international systems. The method of counting days exceeding certain FWI thresholds to define annual 'fire season severity' is common practice in climatological research and fire management.
• Wildland Fires and Air Pollution This type of study analyzes how climate change will affect future fire hazard. To do so, they don't just look at maximum FWI value, but examine precisely the metrics we use: how will fire season duration and number of days with FWI above critical thresholds change. Our final risk matrix, which combines frequency of at-risk days, is therefore a logical and scientifically founded representation of fire season severity.