Outputs

The two spatial state variables whose values change as a result of modeled processes are land cover type (LCT) and time since last fire (TSF). The model produces two new layers for each time step of the simulated period, providing a sequence of LCT and TSF maps for each run of the model (Figure 1).

The model generates other spatial variables derived from the simulation of fire disturbance. For a single run the spatial variable recurrence accounts for the number of times in the hold period that each pixel burnt (Figure 2). The recurrence is independently explained indeed for the two spread pattern types: wind driven (Figure 3) and topographic (Figure 4).

The recurrence maps are used to assess the probability of burning at least once in the simulated time period. Three probability of burning maps are based on the three types of recurrence maps (Figure 5).

The MEDFIRE model incorporates two spatial-explicit fire suppression processes in order to not all targeted fire sizes are effectively burnt. The first fire fighting strategy concerns opportunities generated in areas where spread rate is low enough to fire be suppressed (fire spread suppression mimics overall fire fighting capacity when using low intensity fire conditions). The second fire suppression strategy is based on fire fighting opportunities derived from very low fuel loads present in areas recently burnt. Both processes are implemented introducing pre-specified thresholds: the suppression starts whenever spread rate or TSF are lower than a spread-rate-threshold (in percentage) or a TSF-threshold (in years). The spatial variable effectiveness shows which areas of fire footprints are effectively burnt contrary to fire suppressed areas. The effectiveness is assessed for three increasing thresholds. Two effectiveness maps are generated one for each fire fighting strategy (Figure 6, 7 and 8).

Given the ignitions points the variable perimeter expansion allows to spatially understand the spread process of each single simulated fire (Figure 9). The spatial variable accounts for the time step needed to fire reaches by spread each cell.

Figures

Figure 1: Sequence of the dynamic state variables: land cover type (up row) and time since last fire (down row). Maps on the left represent the initial conditions (similar for all model runs), the maps on the middle are 10 years after and the ones on the right are from 20 years after the initial time.

Figure 2: Fire recurrence in a 20-year time period (1 model run).

Figure 3: Fire recurrence in a 20-year time period (1 model run) for a topographic fire.

Figure 4: Fire recurrence in a 20-year time period (1 model run) for a wind-driven fire.

Figure 5: Map A represents the probability of being burnt (based on 50 runs) at least one time in a 20-year period. Maps B and C represent the probability based on topographic and wind-mediated fires respectively.

Figure 6:Effectiveness in a fire footprint when fire fighting strategy is based on opportunities generated by areas recently burnt.

Figure 7: Effectiveness of fire fighting strategies in a topographic spread type fire.

Figure 8: Effectiveness of fire suppression in a wind-mediated spread type fire.

Figure 9: Blue points mark the ignition points of a set of fires. The figure shows the fire perimeter expansion for that set of fires in the south-west region of Catalonia