Using unmanned aerial vehicles and wildlife camera trapping for the early assessment of road reclamation effects in Northern Ontario

Abstract

The forest industry has been performing forest road reclamation to regenerate forests after harvest, to maintain wildlife habitat and to limit forest access by the public, all to help maintain large, contiguous wild forests. International forest certification bodies like the Forest Stewardship Council (FSC) and Sustainable Forestry Initiative (SFI) use third-party accreditation to ensure that forest management best practices are followed, wood products are harvested sustainably, and sensitive habitat and wildlife species are protected. To access new resources, previously disturbed areas must be returned to similar, natural forest conditions, which includes the reclamation of access roads. Deactivating access roads inherently creates challenges for monitoring site rejuvenation by removing the very pathways needed to monitor those areas. This study attempted to develop a new method for inexpensive, repeatable, large-scale monitoring and measurement of regenerating forest areas considered inaccessible from the ground. This study was done using off-the-shelf unmanned aerial vehicles, or drones. Recent technological advancements has allowed for the application of random photo sampling methods to orthorectified mosaic tiles derived from drone imagery. The goal was to use automated software, and accurately classify percent vegetation cover (VC) on deactivated logging roads and correlate reclamation ‘success’ to wildlife species presence/absence. Two conventional image classification software packages (ERDAS Imagine and eCognition Developer) were tested to assess VC values on road reclamation treatments with overall orthomosaic classification accuracies reaching 97.86%. The largest influence on road regeneration VC was found to be road ecotype. Lack of significant difference in the photo sampling results suggests the previously applied reclamation efforts in this study area were unsuccessful. The wildlife monitoring efforts found a significant difference in the use of different road treatments and ecosite by species, with active/non-reclaimed roads having the lowest species presence and activity.