Maquis fuel model development to support spatially-explicit fire modeling applications

The availability of accurate data on conditions and spatial distribution of fuels has been shown to be critical in numerous topics of fire studies. In Italy, there is a lack of studies devoted to a systematic analysis of the vegetation characteristics related to fire behaviour, thus involving the use of fuel data and fuel models developed in other countries. In this perspective, the aim of this work is to propose an integrated approach to fuel model development and assessment based on field data collection, fuel classification, and fuel mapping. The methodology was applied to the Mediterranean maquis ecosystem. The survey of the main fuel variables allowed the analysis and the characterization of fuel complexes, classified in four main fuel types. The relative custom fuel models were developed and compared with other fuel model developed for Mediterranean shrublands, showing values similar to standard fuel models but different fire behaviour and spread. Maximum Likelihood algorithm was applied on IKONOS satellite image to map the Mediterranean maquis, evidencing that the use of remotely sensed data at high spatial and spectral resolution provided valuable maps of fuel models. Thus, the maps of fuel models derived by remotely sensed images were used at different resolution as input data in the FARSITE fire area simulator, showing that the fine resolution of fuel maps affects the simulation accuracy, improving the predictive capabilities of the simulator.