Mapping Garry Oak (Quercus garryana) Using LiDAR and Multispectral Imagery in Greater Victoria

The Garry oak (Quercus garryana) ecosystem is among the most endangered in Canada, yet conservation efforts are hindered by a lack of high-resolution, publicly accessible spatial data. This study addressed this knowledge gap by developing an automated workflow to identify and map individual Garry oak crowns across the Greater Victoria region using a fusion of airborne laser scanning (ALS) and four-band multispectral imagery. This imagery was used to extract structural and spectral metrics to train a Random Forest classifier. Results indicated high model performance with an overall accuracy of 89.7% and a specificity of 96.2%. A sensitivity of 61.0% reflected the challenges of discriminating interlocking crowns within high density forested patches and the spectral diversity of non-native species in an urban environment. However, the model remained robust in open-canopy environments where the unique structural complexity and spectral signature of Garry oak are most distinct. The high specificity of the final output ensures a reliable baseline for land managers and provides a framework for future ground-truthing. This research provides the methodology for the development of the first individual tree level inventory for the region, establishing a non-invasive and scalable approach for monitoring endangered ecosystems, and informing targeted habitat restoration strategies.