An early season wildfire (SUD008) started approximately a kilometre east of Giroux Lake on May 13, 2025, burning for about a week before being declared out, with a total area burned of 30 ha (Figure 1). While actively fought and fairly low severity, this fire’s proximity to our other research sites and ability to quickly mobilize presented an opportunity to assess the burn severity pattern of a fairly rare early season fire in Central Ontario. At the time of wildfire initiation, the water table of peatlands in the vicinity (40 km southeast) were fairly shallow; 0.14 m and 0.11 m below the surface in a shallow (<0.5m) and deep (>0.5m) peatland, respectively.

Figure 1: Wildfire perimeter of the SUD008 wildfire, from the Ministry of Natural Resources Forest Fire Info Map.

Initial scouting was done on the ground, and followed by a coarse assessment using Sentinel-2 satellite imagery on the Google Earth Engine platform. First, a visual inspection was conducted to confirm areas that had appeared to experience the most burn (Figure 2), which was used in conjunction with on-the-ground scouting to establish a study area of interest. Imagery was extracted for pre-fire (2025-04-25 to 2025-05-10) and post-fire periods (2025-05-18 – 2025-15-30) and the Normalized Burn Ratio (NBR) was calculated for each as (NIR – SWIR2)/ (NIR – SWIR2), with the difference Normalized Burn Ratio (dNBR) subsequently computed from the difference of the median images for each period. dNBR values are typically binned (Table 1) for interpretation to fire severity, but for our purposes we used a single threshold of 0.1 to detect any effect of wildfire.

Table 1: Differenced Normalized Burn Ratio (dNBR) thresholds proposed by the European Forest Fire Information Service (EFFIS), from Llorens et al. (2021)
Figure 2: Pre-fire imagery of the SUD008 wildfire area taken 2025-04-25 to 2025-05-10 (left) and post-fire imagery taken 2025-05-20 – 2025-15-30 by Sentinel-2, displayed in natural colour.

On the ground, burn severity was generally insufficient to measure depth of burn (DOB) by traditional means such as the adventitious root method, so field surveys focused on qualitative burn severity assessment (burned, singed, unburned) in four landscape units: upland rock barren, upland forest, peatland margin and peatland middle. Ten peatlands within the fire footprint were selected, largely with peat depths <0.5m due to the lack of fire impact on larger, deeper peatlands. At each peatland, three north to south transects were surveyed with 1 m spacing that extended into the adjacent uplands to capture the continuum of landscape units described previously. In addition to landscape type and burned state, soil depth was measured along with tree density and an approximation of the pre-fire cover. All points were surveyed to sub-meter accuracy using a Juniper Geode (Juniper Systems Inc, USA) so that they could be placed in the context of available geospatial data. Over 500 points were collected in this manner.

Based on the lowest dNBR threshold of 0.1, the area burned was only 11.45 ha, about 38% of the burned area based on the wildfire perimeter.  Despite the large pixel size of the Sentinel-2 imagery, the dNBR of unburned points surveyed was significantly lower (Kruskal-Wallis p = 6.43 x 10-26) than those that were observed to be singed or burned (Figure 3).

Figure 3: Unburned survey points fell in areas of significantly higher (p<1.00 x 10-25) dNBR based on pre- and post-fire Sentinel-2 imagery.

Based on the ground survey transects, peatland margin and middles were generally resistant to burn, with only 4.1 and 2.2% classified as burned, respectively (Figure 6). However, 43.9% of margin points were classified on the ground as singed. Upland forest and rock barrens, on the other hand, were mostly burned, with 78.7 and 64.2% burned, respectively and very few (1.42% and 4.07%) unburned.

Figure 4: Breakdown of burn status (unburned, singed, burned) by landscape unit across thirty transects taken across ten peatlands in the wildfire footprint.

In Part 2, we will look at the use of higher spatial resolution imagery captured by UAV in conjunction with these field surveys as a tool to gauge the spatial distribution of burn from this wildfire.

References

Llorens, Rafael & Sobrino, Jose & Fernández Filgueira, Cristina & Fernández-Alonso, José & Vega, José. (2021). A methodology to estimate forest fires burned areas and burn severity degrees using Sentinel-2 data. Application to the October 2017 fires in the Iberian Peninsula. International Journal of Applied Earth Observation and Geoinformation. 95. 102243. 10.1016/j.jag.2020.102243.