Comparison and integration of lidar and photogrammetric point clouds for mapping pre-fire forest structure

Abstract

Lidar is an established tool for mapping forest structure, but its sparse spatial and temporal coverage often preclude its use in studying forest disturbance. In contrast, aerial imagery has been and continues to be regularly collected in many regions, and advances in stereo image matching have automated the creation of dense photogrammetric point clouds, which can also be used to map forest structure when paired with an accurate digital terrain model. As part of a study of the physical and ecological impacts of the 2012 High Park fire in Colorado, we generated a photogrammetric point cloud from pre-fire aerial imagery collected in 2008 and combined it with a digital terrain model generated from a 2013 post-fire lidar collection to produce canopy metrics commonly used in modelling of forest structure. We explore the correlation structure between the lidar and photogrammetry-derived canopy metrics, and the relationships between those metrics and forest structure attributes measured at unburned plots in the vicinity of the burn scar. Most corresponding lidar- and photogrammetry- derived canopy metrics had strong linear relationships between them (median r = 0.82), and metrics from both datasets yielded similar root mean square errors in multiple regression models of aboveground biomass (29.3% and 31.0%), basal area (29.8% and 27.7%), and several other forest structure attributes. We found the source of the canopy metrics (lidar or photogrammetry) to be a non-significant factor in some models of forest structure, suggesting that these datasets may be interchangeable in particular cases. Models derived from pre-fire aerial imagery were combined with burn severity information to quantify loss of biomass (2.31 ± 0.001 Tg) and examine possible relationships between forest structure and burn severity. These applications illustrate how the broad spatial and temporal coverage of aerial imagery and growing coverage of lidar can be utilized to improve understanding of changes in forest structure, including assessments of forest carbon flux.