Transferability of Airborne LiDAR Data for Canopy Fuel Mapping: Effect of Pulse Density and Model Formulation

Abstract

Canopy fuel characterization is critical to assess fire hazard and potential severity in forest stands. Simulation tools provide useful information for fire prevention planning to reduce wildfire impacts, provided that reliable fuel maps exist at adequate spatial resolution. Free airborne LiDAR data are becoming available in many countries providing an opportunity to improve fuel monitoring at large scales. In this study, models were fitted to estimate canopy base height (CBH), fuel load (CFL) and bulk density (CBD) from airborne LiDAR in a pine stand area where four point-cloud datasets were acquired at different pulse densities. Best models for CBH, CFL and CBD fitted with LiDAR metrics from the 1 p/m2 dataset resulted in an adjusted R2 of 0.88, 0.68 and 0.58, respectively, with RMSE (MAPE) of 1.85 m (18%), 0.16 kg/m2 (14%) and 0.03 kg/m3 (20%). Transferability assessment of fitted models indicated different level of accuracy depending on LiDAR pulse density (both higher and lower than the calibration dataset) and model formulation (linear, power and exponential). Best results were found for exponential models and similar pulse density (1.7 p/m2) compared to lower (0.5 p/m2) or higher return density (4 p/m2). Differences were also observed regarding the canopy fuel attributes.

Keywords

• airborne LiDAR
• canopy base height
• canopy fuel load
• canopy bulk density
• fuel maps
• pulse density