Atmospheric, climatic and ecological controls on extreme wildfire years in the northwestern United States
Identification_Information
Data_Quality_Information
Metadata_Reference_Information
Identification Information
Section Index
Citation:
Citation Information:
Originator: Ze'ev Gedalof, David L. Peterson, Nathan J. Mantua
Title: Atmospheric, climatic and ecological controls on extreme wildfire years in the northwestern United States
Geospatial Data Presentation Form: Journal article
Other Citation Details: Chapter from Ph.D. Dissertation, titled "Links between Pacific Basin climatic variability and natural systems of the Pacific Northwest"
Larger Work Citation:
Citation Information:
Originator: College of Forest Resources, University of Washington
Publication Date: Unpublished Material
Title: CLIMET (Climate-Landscape Interactions on a Mountain Ecosystem Transect) Ecoplot Data for North Cascades and Olympic Mountains, Washington
Geospatial Data Presentation Form: Database
Publication Information:
Publication Place: University of Washington, Seattle, WA
Publisher: David L. Peterson
Online Linkage: http://www.cfr.washington.edu/research.fme/climet/
Description:
Abstract:
Wildland fire is an important disturbance agent in forests of the American Northwest. Historical fire suppression efforts have contributed to an accumulation of fuels in many northwestern forests, and may result in more frequent and / or more severe wildfire events. Here we investigate the extent to which atmospheric and climatic variability may contribute to variability in annual area burned on 20 national forests in Washington, Oregon and Idaho. Empirical Orthogonal Function (EOF) analysis was used to identify coherent patterns in area burned by wildfire in the Pacific Northwest. Anomaly fields of 500 hPa height were regressed onto the resulting principal component time series to identify the patterns in atmospheric circulation that are associated with variability in area burned by wildfire. Additionally, cross correlation functions were calculated for the Palmer drought severity index (PDSI) over the year preceding the wildfire season. Parallel analyses based on superposed epoch analysis focused only on the extreme fire years (both large and small), to discriminate the controls on extreme years from the linear responses identified in the regression analyses. Four distinct patterns in area burned were identified, each associated with distinct climatic processes. Extreme wildfire years are forced at least in part by antecedent drought and summertime blocking in the 500 hPa height field. However the response to these forcings is modulated by the ecology of the dominant forest. In more mesic forest types antecedent drought is a necessary precondition for forests to burn, but is not a good predictor of area burned due to the rarity of subsequent ignition. At especially dry locations, summertime blocking events can lead to increases in area burned even in the absence of antecedent drought. At particularly xeric locations summertime cyclones can also lead to increased area burned, probably due to dry lightning storms that bring ignition and strong winds but little precipitation. These results suggest fuels treatments alone may not be effective at reducing area burned under extreme climatic conditions, and furthermore that anthropogenic climate change may have important implications for forest management.
Purpose:
Naturally occurring fires are occasionally allowed to burn, prescribed fires are intentionally set, and fuels are managed to encourage natural fire regimes. Fire exclusion continues at locations where cultural, economic, or ecological values require it. The exclusion of fire over the last century has significantly modified the structure and composition of northwestern forests, and made many types of forest more prone to extreme fire events. Consequently, a better understanding of the factors that contribute to severe, extensive fires is critical for the integration of fire as an ecological process into managed landscapes.
Supplemental Information:
Fire behavior is determined largely by the nature of the fuels, topography, and weather occurring at the site of ignition. Of these factors weather is the most variable over time and the most poorly understood. Furthermore, a substantial proportion of total area burned is likely caused by relatively few fires that occur under extreme weather conditions. The goals of this study are to determine if underlying patterns exist in annual area burned in the northwestern United States, and to determine the extent to which these patterns are associated with mid-tropospheric circulation anomalies and variability in antecedent temperature, precipitation, and drought. This approach is distinct from most previous studies in two respects: (1) we do not treat the area west of the Rocky Mountains as a single coherent unit; and (2) we address large fire seasons, rather than individual large fires.
We divided the study area into multiple regions with coherent fire regimes for several reasons. First, the large contrast in forest types that occur throughout this region is almost certainly a response to different climatic and fire regimes that are in operation. Given these differences it seems probable that there may be more than a single pattern in annual area burned, as well as distinct climatic controls on these patterns. Furthermore, by focusing on extreme wildfire years rather than individual large fires we are better able to assess the role of the atmosphere in forcing wildfires. By averaging over space and time we reduce the role that individual site characteristics play in contributing to fire extent. Lastly, by identifying elements of climate that contribute to extreme wildfire years in specific regions of the American Northwest we provide fire managers with a tool that can potentially be used to anticipate the severity of the upcoming fire season.
Spatial Domain:
Description of Geographic Extent:
Twenty national forests were considered in the analysis, located throughout Washington, Oregon and Idaho. This study region abuts the Canadian border to the North, the Pacific Ocean to the west, the Rocky Mountains to the east, and California and Nevada to the south. The Cascade Mountains unevenly bisect the study area along a north-south axis.
Keywords:
Theme:
Theme Keyword Thesaurus: None
Theme Keyword: wildfire
Theme Keyword: climatic variability
Theme Keyword: top-down controls
Theme Keyword: Pacific Decadal Oscillation
Theme Keyword: Empirical Orthogonal Function analysis
Place:
Place Keyword Thesaurus: None
Place Keyword: Pacific Northwest
Place Keyword: Washington
Place Keyword: Oregon
Place Keyword: Idaho
Place Keyword: Deschutes National Forest
Place Keyword: Ochoco National Forest
Place Keyword: Malheur National Forest
Place Keyword: Umatilla National Forest
Place Keyword: Fremont National Forest
Place Keyword: Wallowa-Whitman National Forest
Place Keyword: Nez Perce National Forest
Place Keyword: Colville National Forest
Place Keyword: Gifford Pinchot National Forest
Place Keyword: Mount Baker-Snoqualmie National Forest
Place Keyword: Mt. Hood National Forest
Place Keyword: Okanogan National Forest
Place Keyword: Olympic National Forest
Place Keyword: Rogue River National Forest
Place Keyword: Siskiyou National Forest
Place Keyword: Siuslaw National Forest
Place Keyword: Umpqua National Forest
Place Keyword: Wenatchee National Forest
Place Keyword: Willamette National Forest
Place Keyword: Winema National Forest
Point of Contact:
Contact Information:
Contact Person Primary:
Contact Person: Ze'ev Gedalof
Contact Organization: Department of Geography, University of Guelph
Contact Address:
Address Type: mailing address
Address: Department of Geography, University of Guelph
City: Guelph
State or Province: Ontario
Postal Code: N1G 2W1
Country: Canada
Contact Voice Telephone: (519) 824 - 4120 ext. 58083
Contact Electronic Mail Address: zgedalof@uoguelph.ca
Data Set Credit:
Review: Philip Mote, Jim Agee, Don McKenzie and one anonymous reviewer.
Funding: Joint Institute for the Study of Atmosphere and Ocean (JISAO) under NOAA Cooperative Agreement No. NA17RJ11232.
Ze'ev Gedalof is also supported by the Natural Sciences and Engineering Research Council of Canada. David Peterson is supported by the USDA Forest Service.
Cross Reference:
Citation Information:
Originator: Ze’ev Gedalof, David L. Peterson, Nathan J. Mantua
Title: Columbia River flow and drought since 1750
Geospatial Data Presentation Form: Journal article
Series Information:
Series Name: Journal of the American Water Resources Association
Other Citation Details: Chapter from Ph.D. Dissertation, titled "Links between Pacific Basin climatic variability and natural systems of the Pacific Northwest"
Larger Work Citation:
Cross Reference:
Citation Information:
Originator: Ze'ev Gedalof, David L. Peterson, Nathan J. Mantua
Publication Date: 2002
Title: A multi-century perspective of variability in the Pacific Decadal Oscillation: new insights from tree rings and coral
Geospatial Data Presentation Form: Journal article
Series Information:
Series Name: Geophysical Research Letters
Issue Identification: Volume 29, No. 24, 2204
Other Citation Details: Chapter from Ph.D. Dissertation, titled "Links between Pacific Basin climatic variability and natural systems of the Pacific Northwest"
Larger Work Citation:
Citation Information:
Originator: College of Forest Resources, University of Washington
Publication Date: Unpublished Material
Title: CLIMET (Climate-Landscape Interactions on a Mountain Ecosystem Transect) Ecoplot Data for North Cascades and Olympic Mountains, Washington
Geospatial Data Presentation Form: Database
Publication Information:
Publication Place: University of Washington, Seattle, WA
Publisher: David L. Peterson
Online Linkage: http://www.cfr.washington.edu/research.fme/climet/
Data Quality Information
Section Index
Lineage:
Methodology:
Methodology Type: Lab
Methodology Description:
We considered several possible methods of subdividing the study area into regions of coherent fire patterns, including cluster analysis, classifications based on vegetation, climate and fire regime, and eigenvector techniques. We chose to use empirical orthogonal function (EOF) analysis because it incorporates most of the advantages of the other techniques considered, but has several key advantages. In particular, EOF analysis is able to identify patterns in area burned that might overlap in space (see Schroeder 1969). EOF analysis is a type of eigenanalysis that identifies structures that explain the maximum amount of variance in a two-dimensional data set. In this application the structure dimension consists of 20 national forests, and the sampling dimension is time, with 48 years. Eigenanalysis of this matrix produces a set of spatial structures in the first dimension (EOFs), and corresponding structures in the sampling dimension (principal components (PCs)). Each PC describes the variability of the associated EOF over time. Each EOF is orthogonal to all other EOFs, and each PC is similarly orthogonal to all other PCs. Each EOF / PC pair has a corresponding eigenvalue that describes the variance explained by the pair.
The atmospheric circulation anomalies associated with each EOF were determined by regressing the 500 hPa anomaly field onto the associated PC time series. Suppose P is a vector of the PC time series with n observations, and H is an n × m matrix of 500 hPa height anomalies for a given month of the fire season, where m indicates the number of gridpoints for which 500 hPa height anomalies were considered. The vector K can be calculated as:
K = n^-1 PH
K will be of length m, and can be mapped to show the anomaly field associated with a one-standard-deviation perturbation in P. These maps are commonly called spatial regressions, or map projections, because a time-varying spatial field is being regressed (or projected) onto a time-varying vector. Maps of K were developed for each month of the fire season (May - September). The correlation between 500 hPa height at each gridpoint in K and the relevant BAI was also calculated, and regions where the association is significant (a<0.05) are shown on the figures as dark shaded regions. Although some spurious results are expected, the strong spatial autocorrelation in the atmosphere means that most of the regions identified in this analysis represent important "centers of action", which are contributing to the increase in area burned downstream. The variance in P that can be explained by K can be estimated by calculating k, the expansion coefficient time series of K:
k = HK
which provides a vector describing the variability in K over time. The correlation between k and P provides an estimate of the variation in P that can be linearly explained by k, but should be interpreted cautiously as K was initially defined using P.
This analytical approach assumes that there is a linear association between atmospheric variability and wildfire activity. It is possible however that extreme wildfire years may occur under a much narrower set of climatic conditions than less extreme years. In this scenario the linear association could appear quite weak, and the underlying forcing pattern might be missed altogether. In order to assess this scenario we used superposed epoch analysis (SEA) to characterize the 500 hPa height anomalies during the smallest and largest fire years. The five years with the largest area burned and the five years with the smallest area burned were identified. The mean and standard deviation in the 500 hPa height field for the large and small wildfire years were calculated. A two-sample t-test was used to identify regions where 500 hPa height anomalies are significantly different between the epochs. For display purposes, the small-fire-years composite was subtracted from the large-fire-years composite to emphasize the difference in circulation between the two epochs. Additionally, these map composites were compared to the spatial regression maps for consistency and magnitude.
The role of antecedent climate conditions in preconditioning forests for large wildfire years was explored using cross-correlation analysis. The correlation between the regional BAI (and PCs), and total precipitation, mean temperature, and PDSI, was calculated for each month of the fire season (May to September) as well as the 12 preceding months. These results provide some insight into the relative roles of seasonal-scale climatic variability and shorter-term (presumably synoptic-scale) processes in driving large wildfire years. Similar to the analysis of 500 hPa described above this cross-correlation assumes a linear association between area burned and antecedent climate. For comparative purposes we also calculated composite maps showing the mean PDSI for the most extreme wildfire years.
Methodology Citation:
Citation Information:
Originator: Schroeder, M.J.
Publication Date: 1969
Title: Critical Fire Weather Patterns in the Conterminous United States
Series Information:
Issue Identification: Technical Report WB 8
Publication Information:
Publication Place: Silver Spring, MD
Publisher: Weather Bureau, U.S. Department of Commercer, Environmental Science Services Administration
Metadata Reference Information
Section Index
Metadata Date: 6/15/2004
Metadata Review Date: 6/28/2004
Metadata Future Review Date:
Metadata Contact:
Contact Information:
Contact Organization Primary:
Contact Organization: Fire and Mountain Ecology Lab, College of Forest Resources
Contact Address:
Address Type: mailing address
Address: University of Washington, Box 352100
City: Seattle
State or Province: WA
Postal Code: 98195-2100
Country: USA
Metadata Standard Name: FGDC Biological Data Profile of the Content Standard for Digital Geospatial Metadata
Metadata Standard Version: FGDC-STD-001.1-1999
Metadata Time Convention: local time
Metadata Access Constraints: None
Metadata Use Constraints: None
Metadata Security Information:
Metadata Security Classification System: None
Metadata Security Classification: Unclassified