Fire severity and ecosystem impacts immediately following an extreme fire event in northern Minnesota
The Pagami Creek Fire, which was started by lightning in the Boundary Waters Canoe Area Wilderness (BWCAW), about 14 miles east of Ely, Minnesota on August 18, 2011, grew to be the largest fire in the BWCAW since 1894 (Figure 1). It burned over 38,000 hectares.
Developing an understanding of the effects of fire on soil carbon (C), nitrogen (N), and mercury (Hg) pools is particularly important to inform management approaches and policies that consider the effects of fire on forest ecosystems.
What makes the Pagami Creek Fire a truly exceptional research opportunity is the highly detailed spatial data on pre-fire forest composition, structure, and disturbance history, including unique data such as NASA hyperspectral imagery, forest plot and biogeochemical cycling data, and spruce budworm disturbance assessments---see, for example, Figure 2 (a, from Wolter and Townsend 2011; b, from Wolter et al. 2009). This set of pre-fire data has been assembled by our research team over the past decade through a series of remote sensing and disturbance ecology research projects within the region. However, many of the indicators of fire effects, such as needle and leaf coloration and soil chemistry, are ephemeral and therefore must be assessed immediately following the disturbance. The timing of the Pagami Creek Fire (autumn) therefore required a rapid assessment of fire effects before the onset of winter.
The key questions that are typically difficult to address once the immediate effects of fire have disappeared are
To what extent do remotely-sensed (hyperspectral) fire severity estimates reflect field-based severity indices in both the overstory and the understory?
How do overstory and understory fire severities interact to influence soil carbon (C), nitrogen (N), and mercury (Hg) loss immediately after a fire in comparison with those in the first growing season after the fire?
To address these questions we collected on-the-ground burn severity, soil, and vegetation measurements in late fall of 2011. One hundred plots were established across a range of fire severities and forest types to measure overstory and forest floor fire severity and collect soil samples. We produced an initial estimate of fire severity using the relative difference normalized burn ratio applied to pre-burn and post-burn Landsat TM images to optimize our sampling effort (Figure 3).
We also acquired remotely sensed images from coordinated over-flights of NASA’s Airborne Visible/Infrared Imaging Spectrometer (AVIRIS). The AVIRIS sensor is used for imaging spectroscopy – known as hyperspectral imagery because of the high number and resolution of the “bands” it detects within the light reflectance profile (Figure 4). Such resolution makes hyperspectral imagery ideal for mapping the chemical composition of the canopy and exposed soil. By comparison, Landsat TM applies multispectral imaging with fewer bands, each with lower resolution (indicated by the bars in figure 4). We will use the 6-m resolution AVIRIS images taken post burn (see left insets, figure 4) in combination with our field data to precisely model and map fire burn severity for the complete extent of the Pagami Creek Fire.
We will revisit the burn area in early summer 2012 and repeat our fall 2011 soil measurements. This will allow us to assess immediate after-fire losses of C, N, and Hg from the system, as well as the considerable losses that occur in the winter through spring months immediately after the fire. Both field and image data will be analyzed with respect to the extensive pre-fire data that we have previously mapped for the study area.
Accurate field estimates of disturbance allow us to provide a foundation for a high-profile, long-term research program to investigate spatial interactions of landscape conditions, fire behavior, resulting impacts, and ecosystem recovery. Thus, we will be able to evaluate the extent to which pre-burn attributes constrain fire behavior and severity in time and space. Our field measurements will assess how fire severity influences C, N, and Hg soil pools, while the combination of pre and post-fire remote sensing will allow us to scale these relationships up to determine how much C, N, and Hg was emitted during the fire. The research will facilitate a more complete examination of spatial feedbacks underlying forest landscape structure, fire disturbance, and future patterns of ecosystem recovery. We will establish a baseline of initial impacts at high resolution and large extent – including the initial loss of soil elements after the 2011-2012 winter – for which we have an unprecedented level of detailed pre-fire forest conditions. This baseline dataset will set the stage for several major programs of investigation that dovetail with ongoing research by our team (figure 8).
- Philip A. Townsend, University of Wisconsin, Madison - Department of Forest and Wildlife Ecology; Associate Professor
- Brian Sturtevant, US Forest Service, Northern Research Station; Research Ecologist
- Randy Kolka, US Forest Service - Northern Research Station; Team Leader and Research Soil Scientist
- Peter Wolter, Iowa State University, Natural Resource Ecology and Management; Assistant Professor
- Shawn Fraver, US Forest Service, Northern Research Station; Research Ecologist
- Eric Gustafson, US Forest Service, Northern Research Station; Project Leader and Research Landscape Ecologist
- Jay Charney, US Forest Service, Northern Research Station; Participating Research Meteorologist
- Lee E. Frelich, University of Minnesota, Center for Forest Ecology; Director and Research Associate
- Bruce Anderson, US Forest Service, Superior National Forest; Forest Monitoring Coordinator
This project was funded by a RAPID grant from the National Science Foundation’s (NSF) Ecosystem Program, the National Fire Plan and the US Forest Service.
Last Modified: 01/24/2012