Pagami Creek Fire: Before, During, and After

From late August through September 2011, the Pagami Creek fire burned approximately 38,000 hectares of boreal forest in and around the Boundary Waters Canoe Area Wilderness in Minnesota, USA. The fire exhibited a wide range of behaviors across this ecologically diverse landscape where members of our lab, along with US Forest Service collaborators, had been collecting data since 2002 (to map forest conditions) prior to the fire event.

While the fire was still burning, we applied for and received funding from the NSF RAPID program to collect post-event data that would allow us to assess fire intensity, burn
severity, and their impact on forest soils. Field crews deployed in October 2011, near the end of and immediately following the fire event. From these data we are working to determine how the wide range of fire intensity and burn severity influenced the forest: tree mortality, recovery of Carbon, Nitrogen, and Mercury in soils, post-fire forest regeneration, and the extent to which multispectral (Landsat) and hyperspectral (AVIRIS) remote sensing estimates of burn severity reflect our field-based assessments.


Collaborators: Brian SturtevantRandy KulkaShawn FraverPeter Wolter

Forest disturbance

We can observe and describe forest disturbances in a number of ways using satellite imagery, airborne observations, and field measurements. Our work in Upper Midwest and Great Lakes forests considers numerous disturbance types: severe storm winds, insect defoliation, fire, and various partial and clear-cut harvest treatments. We’ve collected a number of field measurement datasets in Minnesota and Wisconsin, especially in areas of spruce budworm outbreak and the Pagami Creek fire. Hyperspectral images from NASA AVIRIS overflights, and multispectral images from Landsat and MODIS, are employed extensively in this work. To identify disturbed areas we reduce the information from multiple spectral bands to calculate vegetation index values for each pixel. Using multiple dates of imagery, these values are compared with nearby undisturbed forest pixels to detect changes over time and subsequently identify the extent and severity of the disturbance.

Lab Members

Matt Garcia