Estimation of Available Canopy Fuel of Coppice Oak Stands Using Low-Density Airborne Laser Scanning (LiDAR) Data
Predicting fire hazards and simulating fire intensity require knowledge of fuel conditions. Many aspects of wildfire behavior including the rate of spread and intensity are influenced by the amount of vegetation that fuels the fire. Coppice Oak Forests (COF) are strongly influenced by wildfires. In the present study, we examined the ability of airborne LiDAR data to retrieve available canopy fuel (ACF) of coppice Oak forest in Zagros Mountains, Iran. Two different oak-dominated stands were selected based on the stand density including sparse and dense forests. Systematically, 127 plots were established in the field and ACF was calculated using species-specific allometric equations. An outlier filter was used to remove any outlier pulse from the point clouds. Canopy Height Models (CHM) were generated by subtracting DSM and DTM. Different metrics were calculated from CHMs at the plot locations. Linear regression (LR), Artificial Neural Networks (ANN), Boosted Random Forest (BRF), and K-Nearest Neighbor (KNN) were used for modeling. The result showed that there is a strong correlation between ACF and LIDAR-derived metrics (r2 = 0.74 − 0.79). BRF was the best modeling technique. ACF was estimated more accurately in the sparse stand (r2 = 0.79). LIDAR-based predictions can be used to map ACF over coppice oak forests.
KeywordsLow density LiDAR Stepwise regression Zagros Canopy fuel parameters Canopy height model