Aspen Bibliography


A century of landscape change in the southern Rocky Mountains and foothills of Alberta: Using historical photography to quantify ecological change.

Document Type


Publication Date



Throughout the Rocky Mountain regions of North America fire regimes have been altered towards longer fire return intervals in the 20th century as compared to the 18th and 19th centuries; this has been accompanied by losses of grassland and open canopy woodlands due to encroachment by closed canopy forests. However, we lack information regarding how much grassland or open canopy woodland has been lost, where these losses have occurred, and what has caused them. I examined what we know about fire regimes, and how we incorporate this knowledge into management of forested landscapes. Historical land-based oblique angle photographs taken at the start of the 20th century can provide information regarding what vegetation looked like when humans began to alter fire regimes. Unlike aerial photographs, which are used extensively in remote sensing, the major challenges associated with using historical photographs are that georeferencing methods and image classification procedures are still in their infancy. No one has yet conducted a landscape scale spatial analysis of oblique angle historical photographs to describe vegetation change, and to determine the causes or correlates associated with this change. I developed new methods in GIS and image analysis by using a new software tool (the WSL Monoplotting Tool), and created a new method to extract raster-based data from oblique angle historic photos. I used 137 historical repeat photographs from the Mountain Legacy Project covering 320,000 ha in the southern Alberta Rocky Mountains to measure the vegetation change across the landscape between 1909 and 2008. I found that the majority of the landscape (63.4%) had remained in the same vegetation category, 28% of the landscape was in a later seral stage, and less than 9% was in an earlier state. In 1909 58% of the landscape was non-forest (grasslands, meadows, shrubs, non-vegetated and open canopy woodlands) and 42% in closed canopy forest (conifer, broadleaf deciduous and mixedwood). In 2008 42% was non-forested and 58% forested. The Montane Natural Subregion had the greatest proportion of area undergoing successional advancement, with substantial (but lesser) forward change in the Subalpine, Alpine and Foothills Parkland. We saw that nearly 37% of historical grasslands and 80% of open canopy woodlands converted to more advanced successional types, and this appears to be due to gradual forest advancement from the historic forest edge. The causes of these changes in vegetation structure were related to topography and disturbance history. I then tested the assumption that the historic vegetation structure at the turn of the 20th century was less susceptible to burning at high intensity and over larger areas than the current vegetation structure. I used the Burn-P3 model to compare burn probability, fire intensity and fire size for two different scenarios: a) the baseline scenario (the landscape as of 2014); and b) a historical restoration scenario (landscape restored its historical (1909) vegetation structure). I used a subset of the photographs I used for examining landscape vegetation change to determine what the historic vegetation composition was in the Bob Creek Wildland. I used indicator kriging to create a seamless coverage of historical vegetation structure. I found that the overall mean burn probability was only reduced by 1.2% in the historical restoration scenario. However, many areas of the landscape showed increased burn probability and others showed decreases in burn probability, and these were largely associated with expected changes in rates of spread associated with different vegetation changes between the two scenarios. Increased burn probability was associated with areas that had changed from forest to grassland, and areas with decreased burn probability were associated with portions of the landscape changed to broadleaf deciduous vegetation. When I only considered areas that would burn at an intensity greater than 4,000 kW/m, I found the historical restoration scenario reduced the mean burn probability of the landscape by 44%. The mean fire size was also reduced in the historical restoration scenario. Many parts of the wildland had burn probabilities that were less than 10% what they were in the baseline scenario. If managers were to use the historic state of the Wildland as a restoration target, they would see a net benefit with regard to losses due to intense wildfire. A century of landscape change in the southern Rocky Mountains and Foothills of Alberta: Using historical photography to quantify ecological change