Chapter 8: Database of Unburned Areas within Fire Perimeters

Dataset Overview

This dataset is a compilation of spatial polygons representing the unburned areas (that include fire refugia) within 2,298 fires from 1984 to 2014 within eastern Washington, eastern Oregon, and Idaho (see chapter glossary for definitions of terms). On average, 10% of the area within mapped fire perimeters are unburned areas (Meddens et al. 2018), yet these areas have rarely been mapped and are typically not represented in fire perimeter datasets. Therefore, this dataset is the first comprehensive database of recent unburned areas (including fire refugia) for this region of the Pacific Northwest (figure 8.1).

Data Access:

Figure 8.1


Conservation Applications

Potential conservation applications of this dataset could include the following:

  • The study that produced this dataset addressed a primary concern among natural-resource managers in the context of climate change: as fires have become larger and have burned larger portions of the landscape, fire refugia might be dwindling in size or prevalence. The study found no evidence of such a trend. Fire refugia are important to biodiversity conservation because they can provide shelter to wildlife during fires and habitat after fires, and because they can function as seed sources for plant recolonization of burned areas. Together with other research on fire refugia, this study is improving the understanding of how, where, and why fire refugia occur, which may lead to improved ability to predict the effects of various fuel-management strategies.

Applicable scales for detailed spatial assessments:

  • For conservation applications requiring detailed assessment of spatial variation of the dataset within a geographic boundary (such as a protected area), the following geographic scales may be most appropriate (see appendix 3 for more information).
  • At the scale of: a state park or state wildlife area, a local watershed (12-digit hydrologic unit code [HUC-12]), a Bureau of Land Management (BLM) district, a river watershed (8-digit hydrologic unit code [HUC-8]), an individual county, a national forest, a level-3 ecoregion (e.g. the North Cascades), a single state (Washington, Oregon, or Idaho), a region (multiple states in the Pacific Northwest).
  • Within any of the above geographic scales, the dataset cannot be interpreted for areas that did not burn between 1984 and 2014.

Use of the dataset in conservation applications may be limited by the following considerations:

  • This dataset depicts the locations of unburned islands within past fires but cannot independently be used to predict the locations of future fire refugia. For most unburned islands, it is unknown how likely they are to remain unburned through subsequent fires. Furthermore, the unburned islands depicted in this dataset have not been assessed in terms of their function as habitat. For example, some unburned islands in this dataset could represent bare rock or invasive species and do not provide the ecosystem services typically associated with fire refugia (i.e., wildlife habitat, critical seed sources for recolonization).

Past or current conservation applications:

Chapter 8 Case Study

Dataset citation:

Meddens, A., C. A. Kolden, J. A. Lutz, J. T. Abatzoglou, and A. Hudak. 2018. Spatiotemporal patterns of unburned areas within fi re perimeters in the northwestern United States from 1984 to 2014. Ecosphere 9:e02029.

Dataset documentation link: (open access)

Data access:

The dataset can be downloaded from:

The dataset is not available for interactive online map viewing.

Metadata access:

Metadata can be downloaded from:

Dataset corresponding author:
Arjan Meddens
School of the Environment, Washington State University

Data type category (as defined in the Introduction to this guidebook): Fire

Species or ecosystems represented: This dataset does not represent any individual species or ecosystems.

Units and range of mapped values: Not applicable because this dataset does not depict a range of values. Instead it depicts unburned areas as polygons.

Spatial data type: a vector dataset (polygons)

Data file format(s): Shapefile (.shp)

Geographic coordinate system: North American Datum of 1983

Projected coordinate system:  Universal Transverse Mercator (UTM) Zone 11 North

Spatial extent: Regional

Dataset truncation: The dataset is truncated along the following borders: United States border with Canada; Idaho border with Montana, Wyoming, Utah, and Nevada; Oregon border with Nevada and California.

Time period represented: Historic / recent (1984 - 2014)

Methods overview:

Fire perimeters for 2,298 fires from 1984 to 2014 were obtained from the Monitoring Trends in Burn Severity (MTBS) database, which includes all fires of at least 405 hectares in the western United States. To detect unburned islands, pairs of Landsat scenes were compared: an immediate post-fire scene was compared with a scene from one year earlier from the same season, and a one-year-post-fire scene was compared to a pre-fire scene from the same season. A classification tree approach was applied to separate burned from unburned areas (fully described in Meddens et al. (2016), which demonstrated 89% classification accuracy when compared with field observations). Additional adjustments were made to account for the seasonal timing of vegetation greenness and to remove single isolated pixels that were identified as being fire refugia. The resulting gridded (raster) dataset was then transformed to a polygon shapefile dataset. For more information, please consult the dataset citation listed in section 2 of this chapter.

Major input data sources for this dataset included: Remote-sensing data (Landsat)

The mapped values of the dataset may be interpreted as follows:

Polygons represent areas of the landscape that were unburned areas within larger fire perimeters, where fire refugia may exist.

Representations of key concepts in climate-change ecology:

Over recent decades, the size of fires in the western United States and the overall area burned each year has been increasing, attributable in part to climate change (Meddens et al., 2018). Thus, changes in fire patterns are a way in which climate-change exposure is manifested.

To the degree that fire refugia are maintained in constant locations from one fire to another (i.e., if they are determined by landscape characteristics such as topography or water availability), they may represent areas of resistance to changing fire dynamics. They may also play a role in resilience to disturbance events (e.g., fires) because they may provide seed sources from which plant species can recolonize surrounding burned areas and they may provide post-fire animal habitat. For these reasons, areas with lower densities of fire refugia might be more sensitive to changes in fire patterns. It should be noted, however, that this dataset catalogs refugia from individual fires but does not depict longer-term spatial patterns in the likelihood of refugia to exist (i.e., to persist from one fire to another).

This dataset involves the following assumptions, simplifications, and caveats:

The input data (Landsat scenes) used to produce this dataset had a spatial resolution of 30 m. The analysis removed single isolated pixels from being considered fire refugia. Therefore, fire refugia smaller than two Landsat pixels (1,800 m2) could not be detected, although such small fire refugia might play ecologically important roles in post-fire ecosystem recovery. The analysis that produced this dataset did not differentiate between unburned patches that function as habitat for wildlife from those that do not (e.g., bare rock).

Quantification of uncertainty:

The classification tree approach developed by Meddens et al. (2016) and used to produce this dataset had a classification accuracy of 89% for the 19 wildfires studied by Meddens et al. (2016). The dataset described in this chapter and in Meddens et al. (2018) depicts refugia for a much larger number of fires (2,298 as opposed to 19); thus, it should not be assumed to have an equivalent classification accuracy.

Field verification:

This dataset was produced using an approach developed by Meddens et al. (2016), based on a study that analyzed field data from 19 wildfires. Field observations were used to calibrate and improve the model used to detect fire refugia from Landsat data.

Prior to dataset publication, peer review was conducted by external review (at least two anonymous reviewers, each from a different institution).

Meddens, A. J. H., C. A. Kolden, and J. A. Lutz. 2016. Detecting unburned areas within wildfire perimeters using Landsat and ancillary data across the northwestern United States. Remote Sensing of Environment 186:275–285.

Meddens, A., C. A. Kolden, J. A. Lutz, J. T. Abatzoglou, and A. Hudak. 2018. Spatiotemporal patterns of unburned areas within fi re perimeters in the northwestern United States from 1984 to 2014. Ecosphere 9:e02029.

Steenvoorden, J., A. Meddens, A. Martinez, L. Foster, and W. Kissling. 2019. The potential importance of unburned islands as refugia for the persistence of wildlife species in fire-prone ecosystems. Ecology and Evolution 9(15): 2045-7758.