Impact of LiDAR pulse density on forest fuels metrics derived using LadderFuelsR

Reliable forest canopy metrics derived from LiDAR are essential for assessing landscape fire hazard and implementing effective wildfire prevention strategies. However, nationwide LiDAR datasets typically feature low-to-moderate pulse densities, which limit their accuracy in estimating such fuel properties. In this study, we evaluated how low-resolution LiDAR impacts forest vertical structure at the individual tree level by systematically thinning high-resolution LiDAR data to simulate typical pulse densities found in nationwide surveys. The study area encompassed diverse Mediterranean forests in Spain. Key fire hazard metrics, including leaf area density (LAD), leaf area index (LAI), canopy base height (CBH), fuel layer depth, and interlayer distances, were derived using the LadderFuelsR package at the tree level. Four fuel models, each linked to standard fuel model classifications, were identified and analyzed to evaluate the classification shifts across thinning levels and to quantify the rates of change in key fuel properties. Our results showed that thinning causes a significant bias in fire hazard estimation. The CBH and the distance between the layers increased with thinning. In contrast, the fuel layer depth, height, and total and understory LAI decreased. However, fuel models respond differently to pulse thinning depending on their forest structure. Accordingly, thinning affected trees with open crowns and high understory biomass less because of the higher pulse density in the lower crown regions than in those with closed crowns and lower biomass. For example, the understory layer remained more stable in trees with open crowns and a near-ground fuel structure than in those with compact, taller crowns (≥10 pulses/m 2 vs. ≥100 pulses/m 2 ). Similarly, the crown properties exhibited higher stability in open-canopy fuel types than in dense canopies. For instance, CBH and inter-layer distances stabilized at ≥25 pulses/m 2 for open, low crowns but required ≥50–100 pulses/m 2 for dense, tall canopies. Likely, canopy depth stabilized at ≥2–5 pulses/m 2 in open-canopy trees but required ≥25–50 pulses/m 2 in denser forests. Moreover, not all fuel metrics responded uniformly to pulse thinning. Height-based metrics were less affected than crown- and distance-related metrics, whereas the LAI was the most sensitive, declining steadily with lower pulse densities. Finally, we aggregated the tree-level data by median values before estimating the rates of change-masked intra-variability, particularly in highly heterogeneous fuel models. This study highlights the need for tailored LiDAR pulse-density thresholds in nationwide surveys to ensure a balance between data costs and reliability to support forest management and wildfire risk mitigation. • Low-resolution nationwide LiDAR bias fire hazard estimations. • Different fuel models exhibited varying degrees of sensitivity to pulse thinning. • Height-based metrics were less affected than crown and distance-related metrics. • The near loss of the understory layer resulted in an underestimation of fire hazard. • The canopy base height and the distance between fuel layers were overestimated.