When the flame goes out: an exploration of landscape change using repeat photography related to Indigenous burning in Kananaskis Country, Alberta

• Main Author: Frederickson, Maya
• Other Authors: Higgs, Eric
• Format: Thesis
• Language: English
• Published: 2022
• Subjects: Indigenous burning; Repeat photography; Mountain Legacy Project; Rocky Mountains; Landscape change; Fire regime; Canada; Nakoda
• Online Access: http://hdl.handle.net/1828/13930

Fire exclusion has defined 20th century forestry practices in North America and produced many unintended consequences. In the Canadian Rocky Mountains, the removal of fire from the landscape caused significant landscape changes over the past century. Mountain forests are now more uniform in stand composition and structure, and understorey diversity is reduced. These changes mean that forests are now more susceptible to high-intensity, difficult-to-control wildfires. Re-introducing Indigenous led historical burning patterns modeled on traditional burning techniques can be a restoration technique for these highly altered ecosystems. Indigenous fire regimes that emphasized regular, low-intensity burning created forests that had less fuel build up and were not as susceptible to dangerous wildfires. In order to effectively re-introduce historical fire regimes onto the Canadian mountain landscapes, it is essential to understand the history of human management of landscapes with fire. This project uses new methods of oblique image analysis that build on recent developments in oblique image analysis to examine the historical management of a portion of the traditional territory of the Stoney Nakoda Nation that overlaps present day Kananaskis Country in Alberta, Canada. While it is difficult to capture low-intensity Indigenous burns using traditional fire reconstruction methods, in-depth analysis of historical photos taken before the introduction of fire suppression laws may reveal new insights into historical fire regimes. Images were classified using machine learning software and compared to images classified by a human to verify the accuracy of the machine learning software. A case study of georeferencing images was also conducted, with the landcover estimates generated by georeferenced images compared to oblique estimates. Spatial signatures of Indigenous burning were identified and applied to repeat image sets to look for visual evidence of Indigenous burning on the landscape. The results from this study provide a ...