Assessing functional landscape connectivity for disturbance propagation on regional scales—A cost-surface model approach applied to surface fire spread

Leaning on concepts from landscape ecology and functional landscape connectivity, we formulated and developed an operational definition of regional connectivity using a cost-surface modelling approach to assess fire connectivity whereby its structural and spatial components are explicitly isolated. Once the model is calibrated, it allows comparing different scenarios of vegetation composition and moisture contents. The use of commonly available input data and an easy to implement method to code resistance to fire propagation for a given landscape facilitates the application of this approach to other areas of interest. Functional landscape connectivity with regard to fire propagation is expressed through cost surfaces that are computed from a fire friction map and a random set of ignition points. The spatial complexity of the cost surfaces is assumed to be proportional to the landscape connectivity, and its fractal properties are used to measure and describe such spatial complexity. The fractal dimension of a cost-surface serves to assess the regional connectivity in terms of the spatial structure of frictions to fire spread, while the mean value of a cost-surface describes the overall resistance to fire propagation across the landscape in a lumped, non-spatial form. The fire friction map is derived using objective and empirically confirmed techniques enabling to account for the major factors of general fire behaviour. Furthermore, an easy to implement and repeatable method is presented to select the optimum size of random sets of ignition points, implicitly fine-tuned to the spatial variation of the input data. The model was tested on running a number of landscape scenarios based on a NFFL fuel model map. An initial series of runs served to select an optimum number of ignition points and to assess the model sensitivity to fuel moisture. Then, a set of three scenarios of vegetation cover change was devised by replacing a fast fuel model by slower fuels, and the existing network of fuelbreaks was also overlaid. The model performed as expected by quantifying the differential resistance to fire spread implicit to such scenarios. As an overall result, our model indicates that reducing the length scale of the landscape texture has a greater effect preventing fire connectivity than creating large, homogeneous patches of fire resistant vegetation.