Modelling and mapping forest structure from aerial LiDAR data

Abstract

A methodology for estimating forest structure and fuel variables from airborne LiDAR data is presented, based on the following steps: (1) data pre-treatment to obtain the digital terrain model and the canopy height model; (2) extraction of features or attributes from the LiDAR data for each sampling plot; (3) generation of prediction models for structure variables at plot level; and (4) its extension for the generation of maps of forest variables in larger areas. The methodology was applied to a study area of 4,100 ha in the municipality of Cuenca, where a LiDAR flight was carried out with a nominal density of 4 points/m2. For the generation of the prediction models, stepwise multiple regression was used, taking as reference measurements made in 110 field plots, obtaining values of R2 greater than 0.9 for total biomass, volume and fraction of canopy cover, and above 0.86 for basal area, stem biomass and root biomass. The generated models were applied to the entire work area for the creation of maps of forest structure variables, and the entire procedure was integrated into a specific software application.