Wildfire Risk to Communities: Spatial datasets of wildfire risk for populated areas in the United States

Metadata:
Identification_Information:
Citation:
Citation_Information:
Originator: Scott, Joe H.
Originator: Brough, April M.
Originator: Gilbertson-Day, Julie W.
Originator: Dillon, Gregory K.
Originator: Moran, Christopher
Publication_Date: 2020
Title:
Wildfire Risk to Communities: Spatial datasets of wildfire risk for populated areas in the United States
Geospatial_Data_Presentation_Form: raster digital data
Publication_Information:
Publication_Place: Fort Collins, CO
Publisher: Forest Service Research Data Archive
Online_Linkage: https://doi.org/10.2737/RDS-2020-0060
Description:
Abstract:
The data included in this publication depict components of wildfire risk specifically for populated areas in the United States. These datasets represent where people live in the United States and the in situ risk from wildfire, i.e., the risk at the location where the adverse effects take place. Related datasets representing components of risk across the entire landscape are available in a separate data publication (Scott et al. 2020, https://doi.org/10.2737/RDS-2020-0016). Likewise, transmitted risk to housing units from the source locations where damaging fires originate will be also be delivered in a separate publication.

Vegetation and wildland fuels data from LANDFIRE 2014 (version 1.4.0) form the foundation for wildfire hazard and risk data included in the Wildfire Risk to Communities datasets. As such, the data presented here reflect wildfire hazard from landscape conditions as of the end of 2014. National wildfire hazard datasets of annual burn probability and fire intensity were generated from the LANDFIRE 2014 data by the USDA Forest Service, Rocky Mountain Research Station (Short et al. 2020) using the large fire simulation system (FSim). These national datasets produced with FSim have a relatively coarse cell size of 270 meters (m). To bring these datasets down to a finer resolution more useful for assessing hazard and risk to communities, we upsampled them to the native 30-m resolution of the LANDFIRE fuel and vegetation data. In this upsampling process, we also spread values of modeled burn probability and intensity into developed areas represented in LANDFIRE fuels data as non-burnable. Additional methodology documentation is provided with the data publication download.

The data products in this publication that represent where people live reflect 2018 estimates of housing unit and population counts from the U.S. Census Bureau, combined with building footprint data from Microsoft (version 1.1), LandScan 2018 where building footprint data were unavailable, USGS building coverage data, and land cover data from LANDFIRE.

The specific raster datasets included in this publication include:

Housing Unit Density (HUDen): HUDen is a nationwide raster of housing-unit density measured in housing units per square kilometer. The HUDen raster was generated using population and housing-unit count and data from the U.S. Census Bureau, building footprint data from Microsoft, and land cover data from LANDFIRE. In Alaska, LandScan 2018 data were used to identify approximate housing unit locations because Microsoft data were not available across the whole state.

Population Density (PopDen): PopDen is a nationwide raster of residential population density measured in persons per square kilometer. The PopDen raster was generated using population count data from the U.S. Census Bureau, building footprint data from Microsoft, and land cover data from LANDFIRE. In Alaska, LandScan 2018 data were used to identify approximate population locations because Microsoft data were not available across the whole state.

Building Coverage (BuildingCover): BuildingCover is a raster of building density measured as the percent cover of buildings within an approximately 5 acre area around each pixel. It includes all buildings and can be used to complement the HUDen raster, which just reflects residential buildings. Building coverage was generated using building footprint data from Microsoft (v1.1), building coverage data from USGS, and land cover data from LANDFIRE. Building Coverage is not available in Alaska because source data were not available across the whole state.

Building Exposure Type (BuildingExposure): Exposure is the spatial coincidence of wildfire likelihood and intensity with communities. The BuildingExposure layer delineates whether buildings at each pixel are directly exposed to wildfire from adjacent wildland vegetation (pixel value of 1), indirectly exposed to wildfire from indirect sources such as embers and home-to-home ignition (pixel values between 0 and 1), or not exposed to wildfire due to distance from direct and indirect ignition sources (pixel value of 0). It is similar to Exposure Type in the companion data publication, RDS-2020-0016, but just where HUDen > 0 or BuildingCover > 0. Pixels where both HUDen and BuildingCover rasters are zero are NoData in the BuildingExposure raster.

Housing Unit Exposure (HUExposure): HUExposure is the expected number of housing units within a pixel potentially exposed to wildfire in a year. This is a long-term annual average and not intended to represent the actual number of housing units exposed in any specific year. It is calculated as the product of wildfire likelihood and housing unit count. Pixels where the HUDen raster is zero are NoData in the HUExposure raster.

Housing Unit Impact (HUImpact): HUImpact is an index that represents the relative potential impact of fire to housing units at any pixel, if a fire occurs there. It incorporates the general consequences of fire on a home as a function of fire intensity and uses flame length probabilities from wildfire modeling to capture likely intensity of fire. HUImpact does not include the likelihood of fire occurring, and it does not reflect mitigations done to individual structures that would influence susceptibility. It is conceptually similar to Conditional Risk to Potential Structures in the companion data publication, RDS-2020-0016, but also incorporates housing unit count and exposure type. Pixels where the HUDen raster is zero are NoData in the HUImpact raster.

Housing Unit Risk (HURisk): HURisk is an index that integrates all four primary elements of wildfire risk - likelihood, intensity, susceptibility, and exposure - on pixels where housing unit density > 0. It is conceptually similar to Risk to Potential Structures (i.e., Risk to Homes) in the companion data publication, RDS-2020-0016, but also incorporates housing unit count. Pixels where the HUDen raster is zero are NoData in the HURisk raster.
Purpose:
The geospatial data products described and distributed here are part of the Wildfire Risk to Communities project. This project was directed by Congress in the 2018 Consolidated Appropriations Act (i.e., 2018 Omnibus Act, H.R. 1625, Section 210: Wildfire Hazard Severity Mapping) to help U.S. communities understand components of their relative wildfire risk profile, the nature and effects of wildfire risk, and actions communities can take to mitigate risk. These data represent the first time wildfire risk to communities has been mapped nationally with consistent methodology. They provide foundational information for comparing the relative wildfire risk among populated communities in the United States.
Supplemental_Information:
See the Wildfire Risk to Communities website at https://www.wildfirerisk.org for complete project information. The suite of seven raster layers included in this publication are downloadable as zip files by U.S. state. Population Density, Building Coverage, Housing Unit Density, Housing Unit Impact, and Housing Unit Risk are also downloadable as national datasets. National datasets of Housing Unit Exposure and Building Exposure Type are too large for download, but users can request them through the point of contact listed in this metadata document.
Time_Period_of_Content:
Time_Period_Information:
Multiple_Dates/Times:
Single_Date/Time:
Calendar_Date: 2015
Single_Date/Time:
Calendar_Date: 2018
Currentness_Reference:
Ground condition
Status:
Progress: Complete
Maintenance_and_Update_Frequency: As needed
Spatial_Domain:
Description_of_Geographic_Extent:
Conterminous United States, Hawaii, and Alaska
Bounding_Coordinates:
West_Bounding_Coordinate: 165.00000
East_Bounding_Coordinate: -65.00000
North_Bounding_Coordinate: 72.00000
South_Bounding_Coordinate: 18.00000
Keywords:
Theme:
Theme_Keyword_Thesaurus: ISO 19115 Topic Category
Theme_Keyword: environment
Theme_Keyword: geoscientificInformation
Theme_Keyword: society
Theme_Keyword: structure
Theme:
Theme_Keyword_Thesaurus: National Research & Development Taxonomy
Theme_Keyword: Ecology, Ecosystems, & Environment
Theme_Keyword: Fire
Theme_Keyword: Fire detection
Theme_Keyword: Fire ecology
Theme_Keyword: Fire effects on environment
Theme_Keyword: Fire suppression, pre-suppression
Theme_Keyword: Prescribed fire
Theme_Keyword: Environment and People
Theme_Keyword: Forest management
Theme_Keyword: Landscape management
Theme:
Theme_Keyword_Thesaurus: None
Theme_Keyword: burn probability
Theme_Keyword: hazard
Theme_Keyword: fuels management
Theme_Keyword: fire likelihood
Theme_Keyword: fire planning
Theme_Keyword: risk assessment
Theme_Keyword: wildfire hazard potential
Place:
Place_Keyword_Thesaurus: None
Place_Keyword: United States
Place_Keyword: conterminous United States
Place_Keyword: CONUS
Place_Keyword: Alaska
Place_Keyword: Hawaii
Access_Constraints: None
Use_Constraints:
These data were collected using funding from the U.S. Government and can be used without additional permissions or fees. If you use these data in a publication, presentation, or other research product please use the following citation:

Scott, Joe H.; Brough, April M.; Gilbertson-Day, Julie W.; Dillon, Gregory K.; Moran, Christopher. 2020. Wildfire Risk to Communities: Spatial datasets of wildfire risk for populated areas in the United States. Fort Collins, CO: Forest Service Research Data Archive. https://doi.org/10.2737/RDS-2020-0060

The datasets presented here are the product of modeling, and as such carry an inherent degree of error and uncertainty. Users are strongly encouraged to read and fully comprehend the metadata and other available documentation prior to data use. No warranty is made by the Originator as to the accuracy, reliability, or completeness of these data for individual use or aggregate use with other data, or for purposes not intended by the Originator. These datasets are intended to provide nationally-consistent information for the purpose of comparing relative wildfire risk among communities nationally or within a state or county. Data included here are not intended to replace locally-calibrated state, regional, or local risk assessments where they exist. It is the responsibility of the user to be familiar with the value, assumptions, and limitations of these national data publications. Managers and planners must evaluate these data according to the scale and requirements specific to their needs. Spatial information may not meet National Map Accuracy Standards. This information may be updated without notification.
Point_of_Contact:
Contact_Information:
Contact_Person_Primary:
Contact_Person: Gregory K. Dillon
Contact_Organization: USDA Forest Service, Fire Modeling Institute (FMI)
Contact_Position: Spatial Fire Analyst
Contact_Address:
Address_Type: mailing and physical
Address: Missoula Fire Sciences Laboratory
Address: 5775 US Hwy 10 W
City: Missoula
State_or_Province: MT
Postal_Code: 59808
Country: USA
Contact_Voice_Telephone: 406-829-6783
Contact_Electronic_Mail_Address: greg.dillon@usda.gov
Data_Set_Credit:
Funding for this project provided by USDA Forest Service, Fire and Aviation Management. Funding also provided by USDA Forest Service, Fire Modeling Institute, which is part of the Rocky Mountain Research Station, Fire, Fuel and Smoke Science Program. Work on dataset development was primarily completed by Pyrologix, LLC under contract with the USDA Forest Service, Fire Modeling Institute.
Cross_Reference:
Citation_Information:
Originator: Short, Karen C.
Originator: Finney, Mark A.
Originator: Vogler, Kevin C.
Originator: Scott, Joe H.
Originator: Gilbertson-Day, Julie W.
Originator: Grenfell, Isaac C.
Publication_Date: 2020
Title:
Spatial dataset of probabilistic wildfire risk components for the United States (270m)
Edition: 2nd
Geospatial_Data_Presentation_Form: raster digital data
Publication_Information:
Publication_Place: Fort Collins, CO
Publisher: Forest Service Research Data Archive
Online_Linkage: https://doi.org/10.2737/RDS-2016-0034-2
Cross_Reference:
Citation_Information:
Originator: Scott, Joe H.
Originator: Gilbertson-Day, Julie W.
Originator: Moran, Christopher
Originator: Dillon, Gregory K.
Originator: Short, Karen C.
Originator: Vogler, Kevin C.
Publication_Date: 2020
Title:
Wildfire Risk to Communities: Spatial datasets of landscape-wide wildfire risk components for the United States
Geospatial_Data_Presentation_Form: raster digital data
Publication_Information:
Publication_Place: Fort Collins, CO
Publisher: Forest Service Research Data Archive
Online_Linkage: https://doi.org/10.2737/RDS-2020-0016
Back to Top
Data_Quality_Information:
Attribute_Accuracy:
Attribute_Accuracy_Report:
Several of the datasets described here are derived from wildfire simulation modeling, and their exact accuracy cannot be measured. These include Building Exposure Type, Housing Unit Exposure, Housing Unit Impact, and Housing Unit Risk. They are intended to be relative measures of wildfire risk for planning purposes. The FSim datasets of burn probability and intensity used as primary inputs were objectively evaluated and calibrated against historic wildfire occurrence statistics within 136 distinct regions of contemporary wildfire activity (pyromes) across the United States (Short, Grenfell, Riley, and Vogler 2020). See Short et al. (2020) for a more detailed description of FSim calibration. Some LANDFIRE fuels and vegetation data used as inputs have also been evaluated for efficacy and calibrated to meet the objectives of LANDFIRE. More information can be found at: https://www.landfire.gov/lf_evaluation.php.

Datasets included here for Housing Unit Density and Population Density were carefully checked to ensure that numbers of homes and people derived from these datasets generally agreed with published population estimates. At a county level, population counts for CONUS and Hawaii derived from our PopDen raster had a mean percent error (MPE) of 7 percent relative to the 2018 ACS and 6.8 relative to the 2018 PEP. The majority of this, however, was due to our bias in selecting the maximum of ACS and PEP when creating the PopDen raster. The bias-corrected error averaged 0.3 percent across all counties in CONUS and Hawaii and was within 2 percent of published ACS and PEP estimates in 92 percent of counties. Our population estimates in Alaska had slightly higher error due to the use of LandScan points instead of Microsoft building points. Across all Alaska counties, the bias-corrected error in population counts derived from PopDen averaged less than 1 percent, with 86 percent of counties falling within 2 percent of the ACS or PEP county estimates.

Accuracy for Housing Unit Density was similar to our Population Density accuracy. Averaged across counties for CONUS and Hawaii, MPE for housing unit counts derived from HUDen averaged 6.2 percent compared to counts derived from ACS numbers, with our bias accounting for 6 percent of that. The error in our housing unit estimates is within 2 percent of numbers derived from both PEP and ACS for 92 percent of all counties in CONUS and Hawaii. In Alaska, the average MPE for housing units across all counties is -11.5 percent. When we adjust for bias in the number of census blocks with population but no LandScan points (a bias of -13.3 percent), the remaining error is 1.7 percent. The error in housing unit estimates is within 2 percent of numbers derived from ACS values for 80 percent of Alaska counties.

Short, Karen C.; Grenfell, Isaac C.; Riley, Karin L.; Vogler, Kevin C. 2020. Pyromes of the conterminous United States. Fort Collins, CO: Forest Service Research Data Archive. https://doi.org/10.2737/RDS-2020-0020

Short, Karen C.; Finney, Mark A.; Vogler, Kevin C.; Scott, Joe H.; Gilbertson-Day, Julie W.; Grenfell, Isaac C. 2020. Spatial datasets of probabilistic wildfire risk components for the United States (270m). 2nd Edition. Fort Collins, CO: Forest Service Research Data Archive. https://doi.org/10.2737/RDS-2016-0034-2
Quantitative_Attribute_Accuracy_Assessment:
Attribute_Accuracy_Value: Unknown
Attribute_Accuracy_Explanation:
Quantitative accuracy cannot be evaluated.
Logical_Consistency_Report:
Pixel values in these Wildfire Risk to Communities datasets should be within the following ranges:
Housing Unit Density: Integer values between 0 and 65,582.
Population Density: Integer values between 0 and 159,332.
Building Cover: Integer values between 0 and 100.
Building Exposure Type: Floating point values between 0 and 1.
Housing Unit Exposure: Floating point values between 0 and 0.17245.
Housing Unit Impact: Integer values between 0 and 2,000,000,000.
Housing Unit Risk: Integer values between 0 and 12,591,000.
Completeness_Report:
For the HUDen, PopDen, and Building Cover datasets: All pixels that are part of the land and water of the United States have valid non-negative values.
For the HUExposure, HUImpact, and HURisk datasets: All pixels in areas in populated areas (where HUDen > 0) have valid non-negative values. Pixels outside of populated areas are NoData.
For the Building Exposure Type dataset: All pixels in areas with buildings (where HUDen > 0 or Building Cover > 0) have valid non-negative values. Pixels outside of these areas are NoData.
Lineage:
Source_Information:
Source_Citation:
Citation_Information:
Originator: Short, Karen C.
Originator: Finney, Mark A.
Originator: Vogler, Kevin C.
Originator: Scott, Joe H.
Originator: Gilbertson-Day, Julie W.
Originator: Grenfell, Isaac C.
Publication_Date: 2020
Title:
Spatial dataset of probabilistic wildfire risk components for the United States (270m)
Edition: 2nd
Geospatial_Data_Presentation_Form: raster digital data
Publication_Information:
Publication_Place: Fort Collins, CO
Publisher: Forest Service Research Data Archive
Online_Linkage: https://doi.org/10.2737/RDS-2016-0034-2
Type_of_Source_Media: Online
Source_Time_Period_of_Content:
Time_Period_Information:
Single_Date/Time:
Calendar_Date: 20150101
Source_Currentness_Reference:
Ground Condition
Source_Citation_Abbreviation:
FSim BP and FLPs (FLP1, FLP2, FLP3, FLP4, FLP5, FLP6)
Source_Contribution:
Burn probability (BP) and/or flame-length probabilities (FLPs) modeled with FSim were primary spatial inputs to exposure and risk datasets presented here. BP provided information about the overall probability of any 270-meter pixel experiencing a large fire of any intensity. FLPs provided information about the conditional probability of particular fire intensity levels (i.e., likelihood of a particular intensity level, given a fire) for every 270-meter pixel.
Source_Information:
Source_Citation:
Citation_Information:
Originator: LANDFIRE, U.S. Department of the Interior, Geological Survey
Publication_Date: 2017
Title:
LANDFIRE 1.4.0 40 Scott and Burgan Fire Behavior Fuel Models layer
Edition: 1.4.0
Geospatial_Data_Presentation_Form: raster digital data
Other_Citation_Details:
Scott, Joe H.; Burgan, Robert E. 2005. Standard fire behavior fuel models: a comprehensive set for use with Rothermel's surface fire spread model. Gen. Tech. Rep. RMRS-GTR-153. Fort Collins, CO: U.S. Department of Agriculture, Forest Service, Rocky Mountain Research Station. 72 p. https://doi.org/10.2737/rmrs-gtr-153
Online_Linkage: https://landfire.cr.usgs.gov/viewer/
Online_Linkage: https://www.landfire.gov/fuel.php
Type_of_Source_Media: Online
Source_Time_Period_of_Content:
Time_Period_Information:
Single_Date/Time:
Calendar_Date: 20150101
Source_Currentness_Reference:
Ground Condition
Source_Citation_Abbreviation:
LANDFIRE FBFM40
Source_Contribution:
The LANDFIRE Fire Behavior Fuel Models layer was a primary input to the FSim BP and FLP datasets. It was also used to determine habitable land cover when producing housing unit and population density rasters.
Source_Information:
Source_Citation:
Citation_Information:
Originator: Short, Karen C.
Publication_Date: 2017
Title:
Spatial wildfire occurrence data for the United States, 1992-2015 [FPA_FOD_20170508]
Edition: 4th
Geospatial_Data_Presentation_Form: vector digital data
Publication_Information:
Publication_Place: Fort Collins, CO
Publisher: Forest Service Research Data Archive
Other_Citation_Details:
Spatial wildfire occurrence Additional information is available in: Short, Karen C. 2014. A spatial database of wildfires in the United States, 1992-2011. Earth Systems Science Data 6:1-27. https://doi.org/10.5194/essd-6-1-2014
Online_Linkage: https://doi.org/10.2737/RDS-2013-0009.4
Type_of_Source_Media: Online
Source_Time_Period_of_Content:
Time_Period_Information:
Range_of_Dates/Times:
Beginning_Date: 19920101
Ending_Date: 20151231
Source_Currentness_Reference:
Observed
Source_Citation_Abbreviation:
FPA FOD
Source_Contribution:
It was used in the process of creating the burn probability (BP) and flame length probability (FLP) rasters.
Source_Information:
Source_Citation:
Citation_Information:
Originator: U.S. Census Bureau
Publication_Date: 2019
Title:
2018 5-year American Community Survey
Geospatial_Data_Presentation_Form: tabular digital data
Online_Linkage: https://www2.census.gov/programs-surveys/acs/summary_file/2018/data/
Type_of_Source_Media: Online
Source_Time_Period_of_Content:
Time_Period_Information:
Range_of_Dates/Times:
Beginning_Date: 2014
Ending_Date: 2018
Source_Currentness_Reference:
Ground Condition
Source_Citation_Abbreviation:
2018 5-year ACS
Source_Contribution:
2018 5-year ACS population counts for census block groups were used along with other data sources to estimate the 2018 population in each census block. These block level estimates were used to produce the HUden and POPden datasets.
Source_Information:
Source_Citation:
Citation_Information:
Originator: U.S. Census Bureau
Publication_Date: 2018
Title:
2018 Population Estimates Program, County Population Totals: 2010-2018
Edition: 2018
Geospatial_Data_Presentation_Form: tabular digital data
Online_Linkage: https://www.census.gov/programs-surveys/popest/data/data-sets.2018.html
Type_of_Source_Media: Online
Source_Time_Period_of_Content:
Time_Period_Information:
Range_of_Dates/Times:
Beginning_Date: 2010
Ending_Date: 2018
Source_Currentness_Reference:
Ground Condition
Source_Citation_Abbreviation:
2018 PEP
Source_Contribution:
2018 PEP population estimates for counties were used along with other data sources to estimate the 2018 population in each census block. These block level estimates were used to produce the HUden and POPden datasets.
Source_Information:
Source_Citation:
Citation_Information:
Originator: Microsoft
Publication_Date: Unknown
Title:
USBuildingFootprints
Edition: 1.1
Geospatial_Data_Presentation_Form: vector digital data
Online_Linkage: https://www.microsoft.com/en-us/maps/building-footprints
Online_Linkage: https://github.com/Microsoft/USBuildingFootprints
Type_of_Source_Media: Online
Source_Time_Period_of_Content:
Time_Period_Information:
Single_Date/Time:
Calendar_Date: 2019
Source_Currentness_Reference:
Observed
Source_Citation_Abbreviation:
Microsoft building footprints
Source_Contribution:
Building footprint data were first processed to remove buildings unlikely to be residential housing units. The locations of remaining building points were used to spatially distribute population within each census block. Building footprint data were also a primary input to the building coverage raster.
Source_Information:
Source_Citation:
Citation_Information:
Originator: Heris, Mehdi P.
Originator: Foks, Nathan Leon
Originator: Bagstad, Ken
Originator: Troy, Austin
Originator: Ancona, Zachary H.
Publication_Date: 2020
Title:
A national dataset of rasterized building footprints for the U.S.
Geospatial_Data_Presentation_Form: raster digital data
Publication_Information:
Publisher: U.S. Geological Survey
Online_Linkage: https://doi.org/10.5066/P9J2Y1WG
Type_of_Source_Media: Online
Source_Time_Period_of_Content:
Time_Period_Information:
Single_Date/Time:
Calendar_Date: 2018
Source_Currentness_Reference:
Publication date
Source_Citation_Abbreviation:
USGS building coverage data
Source_Contribution:
This national dataset of building coverage created from version 1.0 of the Microsoft Building footprints dataset provided the starting point for our building coverage raster. Due to known errors in version 1.0 of the Microsoft data, there are swaths of missing data in this USGS dataset. There are also areas where this dataset correctly shows building coverage that is missing in version 1.1 of the Microsoft data. So we used this USGS dataset and Microsoft building footprints version 1.1. to capture as much information on building locations as possible.
Source_Information:
Source_Citation:
Citation_Information:
Originator: Rose, Amy N.
Originator: McKee, Jacob J.
Originator: Urban, Marie L.
Originator: Bright, Eddie A.
Publication_Date: 2017
Title:
LandScan 2017
Geospatial_Data_Presentation_Form: raster digital data
Publication_Information:
Publication_Place: Oak Ridge, TN
Publisher: Oak Ridge National Laboratory
Online_Linkage: https://landscan.ornl.gov/
Type_of_Source_Media: Online
Source_Time_Period_of_Content:
Time_Period_Information:
Single_Date/Time:
Calendar_Date: 2017
Source_Currentness_Reference:
Publication Date
Source_Citation_Abbreviation:
LandScan
Source_Contribution:
Populated pixel centroids from LandScan were used in place of Microsoft building footprint centroids in Alaska, because the Microsoft data were not consistently available across the entire state. LandScan data were also used to check and validate the HUden and POPden rasters.
Process_Step:
Process_Description:
The Wildfire Risk to Communities (WRC) datasets are based on wildfire simulation modeling. Given the relatively short time available for analysis and production of the WRC datasets, the methods for this project were designed to leverage the existing national wildfire simulation data from Short et al. (2020) without further local calibration or modeling work. To make the WRC data most useful to communities, we implemented a process to downscale the national data from their native 270-meter (m) cell size to the native 30-m resolution of the nationally available LANDFIRE fuels and vegetation data. Through this process, we also used geospatial smoothing techniques to account for wildfire hazard in parts of communities adjacent to wildland vegetation that may have indirect exposure to wildland fire. The overall process was as follows:

1. Produce annual burn probability (BP; wildfire likelihood), flame length probabilities (FLPs; intensity), conditional risk to potential structures (CRPS), risk to potential structures (RPS), and exposure type rasters at 30-m resolution for all lands in the U.S. Methods for these datasets are described in data publication RDS-2020-0016.

2. Produce the population density raster (PopDen) with the following general steps: 1) estimate 2018 census block population count using both the 2018 American Community Survey (ACS) and Population Estimates Program (PEP) population estimates; 2) allocate that population evenly to qualifying building centroids within the block, producing a new population-per-centroid attribute; then 3) convert the population-per-centroid attribute of the point features to a 30-m density raster using a pair of 200m-radius moving window operations and the LANDFIRE FBFM40 raster to define habitable land cover. PopDen is calculated in units of persons per square kilometer (sq km).

3. Produce the housing unit density raster (HUDen) with the following general steps: 1) Calculate the ratio of persons per household for each U.S. county or equivalent (“county ratio”), using county-level population and housing unit numbers published in the 2018 ACS; 2) divide the population-per-centroid attribute of the building centroids used for PopDen by the county ratio of the county containing each building centroid to produce a new housing-units-per-centroid attribute; then 3) convert the housing-units-per-centroid attribute of the point features to a 30-m density raster using a pair of 200m-radius moving window operations and the LANDFIRE FBFM40 raster to define habitable land cover. HUDen is calculated in units of housing units per sq km. Note: Census blocks with no building centroids have a HUDen value of zero, even if population was present.

4. Produce the building coverage raster with the following general steps: 1) determine the total building footprint area from either the published USGS data, small building centroids (building footprints 900 sq m or smaller), or large building polygons (building footprints larger than 900 sq m); then 2) convert footprint area to the 30-m resolution BuildingCover raster using a pair of 75m-radius moving window operations and the LANDFIRE FBFM40 raster to define habitable land area. BuildingCover is calculated as the percent of habitable land covered by buildings.

5. Produce the building exposure type raster by masking the exposure type raster published with RDS-2020-0016 to areas where HUDen > 0 or BuildingCover > 0. The building exposure type is “direct” (pixel value of 1) if the underlying land cover is considered burnable in the LANDFIRE FBFM40 raster. The exposure type is “indirect” (pixel value between 0 and 1) if two conditions are met: 1) the land cover is nonburnable urban, agricultural, or bare ground, and 2) burn probability raster is non-zero. Finally, the exposure type is “nonexposed” (pixel value of 0) if the underlying land cover is nonburnable and the burn probability raster is zero.

6. Produce the housing unit exposure raster (HUExposure) with the following general steps: 1) convert the HUDen raster to a 30-m raster of housing unit count by multiplying by 0.0009 (sq km per pixel); then 2) multiply housing unit count by the burn probability raster from RDS-2020-0016.

7. Produce the housing unit impact raster (HUImpact) with the following general steps: 1) convert the HUDen raster to a 30-m raster of housing unit count by multiplying by 0.0009; 2) create a conditional housing unit risk raster (cHURisk) by calculating the sum at each pixel of FLPi * RFi, where i is fire intensity level 1 – 6 and FLP and RF are the flame length probability and response function values for each intensity level, then multiplying that sum by housing unit count; then 3) multiply the cHURisk raster by building exposure type to capture the decay in potential wildfire housing unit impact with increasing distance away from the edge of wildland fuels. Note: Response function (RF) values used in this calculation were: FIL1: 25; FIL2: 40; FIL3: 55; FIL4: 70; FIL5: 85; FIL6: 100).

8. Produce the housing unit risk raster (HURisk) with the following general steps: 1) calculate the cHURisk raster as described in the previous step; and 2) multiply cHURisk by the burn probability raster from RDS-2020-0016.

Additional methodology documentation is provided with the data publication download.
Process_Date: 2020
Back to Top
Entity_and_Attribute_Information:
Overview_Description:
Entity_and_Attribute_Overview:
Below is a description of the files included in this data publication.

DATA FILES - STATE (7)

Georeferenced TIFF files are provided for each U.S. state and the District of Columbia. Alaska is split into North and South because of large file sizes. For each state there are seven raster datasets that are provided as a single downloadable zip file:

1. \Data\*STATE*\BuildingCover_*STATE*.tif: Continuous integer values of percent cover of buildings with a 30-m pixel size. Values are between 0 and 100.

2. \Data\*STATE*\BuildingExposure_*STATE*.tif: Continuous floating point values of exposure type that depict the type of wildfire exposure a housing unit would experience with a 30-m pixel size. Values are between 0 and 1. Referred to in the Wildfire Risk to Communities web application as Exposure Type. A value of 1 is "direct" exposure. Values between 0 and 1 represent "indirect" exposure, with highver values representing closer proximity to directly exposed areas (i.e., areas of burnable wildland vegetation). A value of 0 represents "nonexposed" areas that have nonburnable land cover and are more than 1530 m (approx. 1 mile) from burnable wildland vegetation.

3. \Data\*STATE*\HUDen_*STATE*.tif: Continuous integer values of 2018 housing unit density with a 30-m pixel size. Values are between 0 and 65,582. Units are housing units per square kilometer (sq km). Housing units are any structures that have residential population associated with them and include primary single-family and multi-family residential buildings, secondary or seasonal homes, and facilities such as prisons, dormitories, barracks, and nursing homes.

4. \Data\*STATE*\HUExposure_*STATE*.tif: Continuous floating point values of housing unit exposure with a 30-m pixel size. Values are between 0 and 0.17245. Units are the expected number of housing units per pixel exposed to wildfire in a year. This is a long-term average and not intended to represent the actual number of housing units exposed in any specific year. Calculated as the product of wildfire likelihood and housing unit count.

5. \Data\*STATE*\HUImpact_*STATE*.tif: Continuous integer values of housing unit impact with a 30-m pixel size. Values are between 0 and 2,000,000,000. It is a unitless index.

6. \Data\*STATE*\HURisk_*STATE*.tif: Continuous integer values of housing unit risk with a 30-m pixel size. Values are between 0 and 12,591,000. It is a unitless index.

7. \Data\*STATE*\PopDen_*STATE*.tif: Continuous integer values of 2018 population density with a 30-m pixel size. Values are between 0 and 159,332. Units are people per sq km.


DATA FILES - CONUS (5)

Georeferenced TIFF files are provided as mosaics for the conterminous U.S. (CONUS) in a separate downloadable zip file for each of the following themes:

1. \Data\CONUS\BuildingCover_CONUS.tif: Continuous integer values of percent cover of buildings with a 30-m pixel size. Values are between 0 and 100.

2. \Data\CONUS\HUDen_CONUS.tif: Continuous integer values of 2018 housing unit density with a 30-m pixel size. Values are between 0 and 65,582. Units are housing units per sq km. Housing units are any structures that have residential population associated with them and include single-family and multi-family residential buildings as well as facilities such as prisons, dormitories, barracks, and nursing homes.

3. \Data\CONUS\HUImpact_CONUS.tif: Continuous integer values of housing unit impact with a 30-m pixel size. Values are between 0 and 2,000,000,000. It is a unitless index.

4. \Data\CONUS\HURisk_CONUS.tif: Continuous integer values of housing unit risk with a 30-m pixel size. Values are between 0 and 12,591,000. It is a unitless index.

5. \Data\CONUS\PopDen_CONUS.tif: Continuous integer values of 2018 population density with a 30-m pixel size. Values are between 0 and 159,332. Units are people per sq km.


(Associated .OVR files for all \Data files are included and contain pyramids that allow the raster datasets to draw more quickly in GIS software. Associated .XML files contain dataset-specific FGDC-CSDGM metadata containing a description of the content, quality, and other characteristics of the data. Other associated files include: .TFW, .AUX.XML, .VAT.CPG, and .VAT.DBF, and contain spatial reference, statistics, and attribute information.)


SUPPLEMENTAL FILES (2)

1. \Supplements\WRC_PopulatedAreas_GISDataSymbology.xlsx: Microsoft Excel file with suggested class definitions and colors for displaying the raster datasets in GIS software.

2. \Supplements\WRC_PopulatedAreas_Methods.pdf: Portable Document Format file containing detailed descriptions of the data products included in this publication and the methods used to create them.
Entity_and_Attribute_Detail_Citation:
Scott, Joe H.; Thompson, Matthew P.; Calkin, David E. 2013. A wildfire risk assessment framework for land and resource management. Gen. Tech. Rep. RMRS-GTR-315. U.S. Department of Agriculture, Forest Service, Rocky Mountain Research Station. 83 p. https://doi.org/10.2737/rmrs-gtr-315

Short, Karen C.; Finney, Mark A.; Vogler, Kevin C.; Scott, Joe H.; Gilbertson-Day, Julie W.; Grenfell, Isaac C. 2020. Spatial datasets of probabilistic wildfire risk components for the United States (270m). 2nd Edition. Fort Collins, CO: Forest Service Research Data Archive. https://doi.org/10.2737/RDS-2016-0034-2
Back to Top
Distribution_Information:
Distributor:
Contact_Information:
Contact_Organization_Primary:
Contact_Organization: USDA Forest Service, Research and Development
Contact_Position: Research Data Archivist
Contact_Address:
Address_Type: mailing and physical
Address: 240 West Prospect Road
City: Fort Collins
State_or_Province: CO
Postal_Code: 80526
Country: USA
Contact_Voice_Telephone: see Contact Instructions
Contact Instructions: This contact information was current as of December 2020. For current information see Contact Us page on: https://doi.org/10.2737/RDS.
Resource_Description: RDS-2020-0060
Distribution_Liability:
Metadata documents have been reviewed for accuracy and completeness. Unless otherwise stated, all data and related materials are considered to satisfy the quality standards relative to the purpose for which the data were collected. However, neither the author, the Archive, nor any part of the federal government can assure the reliability or suitability of these data for a particular purpose. The act of distribution shall not constitute any such warranty, and no responsibility is assumed for a user's application of these data or related materials.

The metadata, data, or related materials may be updated without notification. If a user believes errors are present in the metadata, data or related materials, please use the information in (1) Identification Information: Point of Contact, (2) Metadata Reference: Metadata Contact, or (3) Distribution Information: Distributor to notify the author or the Archive of the issues.
Standard_Order_Process:
Digital_Form:
Digital_Transfer_Information:
Format_Name: TIFF
Format_Version_Number: 2020
Format_Information_Content:
Building Exposure Type and HUExposure: 32 Bit floating point; BuildingCover, HUDen, HUImpact, HURisk, and PopDen: 32 Bit unsigned integer; LZW compression; pyramids: levels 5, Nearest Neighbor resampling
File_Decompression_Technique: Files zipped using the zipfile module in Python 2.7.16 (.ZIP file format version 6.3.6). (We recommend putting all *.zip files into a folder named Data before unzipping.)
Digital_Transfer_Option:
Online_Option:
Computer_Contact_Information:
Network_Address:
Network_Resource_Name: https://doi.org/10.2737/RDS-2020-0060
Digital_Form:
Digital_Transfer_Information:
Format_Name: XLSX
Format_Version_Number: see Format Specification
Format_Information_Content:
Microsoft Excel Open XML spreadsheet file
File_Decompression_Technique: Files zipped using 7-Zip 19.0
Digital_Transfer_Option:
Online_Option:
Computer_Contact_Information:
Network_Address:
Network_Resource_Name: https://doi.org/10.2737/RDS-2020-0060
Digital_Form:
Digital_Transfer_Information:
Format_Name: PDF
Format_Version_Number: see Format Specification
Format_Information_Content:
Portable Document Format file
File_Decompression_Technique: Files zipped using 7-Zip 19.0
Digital_Transfer_Option:
Online_Option:
Computer_Contact_Information:
Network_Address:
Network_Resource_Name: https://doi.org/10.2737/RDS-2020-0060
Fees: None
Custom_Order_Process:
National datasets of Housing Unit Exposure and Building Exposure Type are too large for download, but users can request them through the point of contact listed in this metadata document.
Back to Top
Metadata_Reference_Information:
Metadata_Date: 20201221
Metadata_Contact:
Contact_Information:
Contact_Person_Primary:
Contact_Person: Gregory K. Dillon
Contact_Organization: USDA Forest Service, Fire Modeling Institute (FMI)
Contact_Position: Spatial Fire Analyst
Contact_Address:
Address_Type: mailing and physical
Address: Missoula Fire Sciences Laboratory
Address: 5775 US Hwy 10 W
City: Missoula
State_or_Province: MT
Postal_Code: 59808
Country: USA
Contact_Voice_Telephone: 406-829-6783
Contact_Electronic_Mail_Address: greg.dillon@usda.gov
Metadata_Standard_Name: FGDC Content Standard for Digital Geospatial Metadata
Metadata_Standard_Version: FGDC-STD-001.1-1999
Back to Top