Pre-Fire Forest Structure Characteristics
A comparison of LiDAR-derived pre-fire forest structure characteristics for state and private coniferous forests of the Pacific Northwest
Affiliation: Department of Geography, Portland State University, Portland, Oregon USA
Email: aldyer@pdx.edu / dyeralec6@gmail.com
Keywords: remote sensing, forest structure, LiDAR, burn severity
Wildfire activity in the Pacific Northwest has been on the rise in the recent decades due to the area’s warming climate (Perry et al., 2008) but more importantly due to fire exclusion and logging practices since European colonization since the 20th century (Reilly et al., 2021). A more complete understanding of how forest structure influences burn severity may provide insight into both active wildfire management techniques as well as pre-fire timber practices.
This project is a comparison between burn severity and Light Detection and Ranging (LiDAR)-derived forest structure metrics in both state and timber forests to discover forest structure characteristics that lead to increased burn severities. A portion of the 2020 Riverside wildfire in Oregon’s western Cascades will be the study area for this project (Figure 1). The tree species in this lower elevation section of the west Cascades mainly include Douglas-fir (Pseudotsuga menziesii) and Western Hemlock (Tsuga heterophylla). It is hypothesized that dense timber stands in privately owned lands will burn at the highest severity due to the large amount of thin, fire-prone trees. Contrastingly, even in state forests with a similar canopy density, burn severity may be lower if the trees are older and have grown fire-resistant bark.
Figure 1. LiDAR point cloud visualizing elevation by a multipart color scheme.
LiDAR data was acquired from the Clackamol project (Grid #45122-A3) from the Oregon Department of Geology and Mineral Industries (DOGAMI) that was collected in 2013. The LiDAR point cloud (Figure 2) was utilized to create continuous images derivatives of canopy density and canopy height (Figure 3). With an average point spacing of 1.14 feet, the spatial resolution of the output images was rounded up to 1 meter, which is over four times higher than the point spacing.
Canopy density was calculated by first converting the point cloud to a raster format where each cell value is equal to the point density (i.e., number of points) within each cell. This was completed for both the first returns (above ground) and the last returns (on ground). Canopy density was calculated by normalizing the above ground point density by the overall (on ground and above ground) point density. The resulting canopy density image ranges from 0.0 to 1.0, where 0.0 represents no canopy and 1.0 very dense canopy.
Canopy height is simply calculated by subtracting the digital elevation model (DEM) from the digital surface model (DSM) to get the height of the objects (in this case vegetation) above the ground. The DEM and DSM are given from the DOGAMI project.
Additionally, the metric used to characterize burn severity was the Composite Burn Index (CBI) developed by Key & Benson (2006) and created in Google Earth Engine (GEE) based on Parks et al. (2019) (Figure 3).
Figure 2. LiDAR point cloud visualizing elevation by a multipart color scheme.
Figure 3. Image layers used in analysis including four LiDAR-derived rasters and a composite burn index raster.
How the distribution of CBI varies by canopy density and canopy height between state and private forests is shown in Figure 4. Here, canopy density is classified into low (75%), and canopy height is classified based on approximate age into young (82 ft). Canopy density shows little change in CBI based on class and management, contradicting the original hypothesis. However, the CBI does vary by height class for the privately owned forests with higher burn severity for the younger, shorter trees and lower burn severity for mature and old growth trees.
Figure 4. Comparison of the composite burn index (CBI) for state and private managed forests based on canopy density (A) and canopy height (B).
In conclusion, LiDAR-derived products provide the ability to characterize forest structure across a landscape with a high spatial resolution and study how pre-fire conditions influence burn severity. While canopy density showed little variation in the CBI, canopy height played an important role in how the privately owned forests burned. Results for private timber forests showed an increased burn severity for thin, young stands and a considerable decrease in burn severity for thick, mature stands. This makes sense as mature trees have a thicker bark layer and no lower hanging limbs, which allows them to better resist burning. Additionally, the state forests burned evenly, although severely, with little to no influence on forest structure. Wildfire management approaches may be tailored to better protect young timber stands from burning while allowing fires to burn through state forests where conditions may not be controllable.
