Unlocking the potential of Airborne LiDAR for direct assessment of fuel bulk density and load distributions for wildfire hazard mapping

Large-scale mapping of fuel load and fuel vertical distribution is essential for assessing fire danger, setting strategic goals and actions, and determining long-term resource needs. The Airborne LiDAR system can fulfil such goal by accurately capturing the three-dimensional arrangement of vegetation at regional and national scales. We developed a novel method to estimate multiple metrics of fuel load and vertical bulk density distribution for any type of vegetation. The approach uses Beer-Lambert law for inverting the ALS point cloud into vertical plant area density profiles, which are converted into vertical bulk density distribution profiles using species-specific plant traits. The approach is evaluated by comparing ALS-based vegetation profiles and fuel metrics with field-based data from southeastern France, Spain, and Portugal for a range of vegetation types. ALS-based and field-based vertical vegetation profiles were consistent. The range of values of fuel load metrics was also consistent with field data. Good correlations and low bias were attained for simple stratified structure with R² of 0.6, 0.42 and 0.68 and bias of -5 %, -2 % and -3.3 % for canopy base height, canopy fuel load, and canopy bulk density respectively. However, correlations were low for complex vertical structures. The use of species-specific plant traits appeared relevant by lowering the deviation between field and ALS-based values for most species. Our field-independent fuel metric estimation shows comparable performance to results in the literature based on classification approaches trained on field metrics, highlighting the generality of our direct approach. We demonstrated how our approach is more relevant than field data for defining vertical vegetation strata in complex forest structures. We showed an application of the methods by mapping multiple metrics at regional scale (6343 km²) such as canopy base height, fuel strata gap, and canopy and understory fuel loads. Our approach is adequate for feeding next generation models of wildfire risk assessment systems, enhanced by more flexible and accurate fuel data than the existing fuel typologies.