Home | Fire | RegCM | Climate | Fire and Climate | Atlas


Fire-Data Sources | Daily Fire-Start Data | Seasonal Cycle |
Monthly Incidence-and-Area Data


 
Fire Data Sources

Data on the incidence and area of fires in the western U.S. come from two major sources: 

  • 1980-2000 monthly gridded data (Westerling et al., 2003)
  • 1986-1996 National Fire Occurrence Data Base (daily point-location (latitude and longitude) fire-start data (Hardy et al., 2001))

The 1980-2000 monthly Westerling et al. (2003) fire data set is based on ~300,000 fire reports of fire starts and area burned from the USFS, BLM, NPS, and BIA. The fire reports were assessed for data quality, combined, and interpolated onto a 1° × 1° grid covering 31o to 49o north latitude and 101o-125oW longitude.

The 1986-1996 National Fire Occurrence Database as described in Hardy et al. (2001) and Schmidt et al. (2002), available at http://www.fs.fed.us/fire/fuelman/ . These data consist of records of the locations of individual fires, the date when each was first reported, the ignition source and, for most records, the ultimate size and the date when the fire was considered to have been controlled (but not necessarily extinguished). A subset of the full data set was extracted containing 332,404 records, including 116,489 fires started by lightning and 197,617 fires started by humans west of 102°W. These data are not without problems, but we screened the raw data set and believe that they are sufficient for our purposes. Further discussion of the nature of these daily fire-start records is provided by Hardy et al. (2001), the review by Westerling et al. (2003), and in an assessment of the quality of such point-location data by Brown et al. (2002).

Data-Set Citations:

 

  Westerling, A.L., A. Gershunov, T.J. Brown, D.R. Cayan, and M.D. Dettinger, 2003: Climate and wildfire in the western United States. Bulletin of the American Meteorological Society, 84:595-604.
     

  Hardy, C.C., K.M. Schmidt, J.P. Menakis, and R.N. Sampson, 2001: Spatial data for national fire planning and fuel management. International Journal of Wildland Fire, 10, 353-372.
     
Image Citations:
 

  Bartlein, P.J., S.W. Hostetler, S.L. Shafer, J.O. Holman, and A.M. Solomon, 2003: The seasonal cycle of wildfire and climate in the western United States, 5th Symposium on Fire and Forest Meteorology, American Meteorological Society, paper P3.9.
     

  Hostetler, S.W., P.J. Bartlein, J.O. Holman, A.M. Solomon, and S.L. Shafer 2003: Using a regional climate model to diagnose climatological and meteorological controls of wildfire in the Western United States, 5th Symposium on Fire and Forest Meteorology, American Meteorological Society, paper P1.3.

(Please cite these when using any of the images below)

 

Fire-Data Sources | Daily Fire-Start Data | Seasonal Cycle |
Monthly Incidence-and-Area Data


 
 
Daily Fire-Start Data (1986-1996)
 
The first-order spatial patterns of the locations of lighting- and human-started fires, and the ultimate sizes of those fires reflect the distributions of forested land across the western U.S. with human-started fires showing additional patterns that reflect the location of urbanized or densely settled rural areas.
 
Lighting- and Human-Started Fires:  Point Locations and Areas
     
 
 
 

Fire-Data Sources | Daily Fire-Start Data | Seasonal Cycle |
Monthly Incidence-and-Area Data


 
 
Seasonal Cycle of Daily Fire-Start Data (1986-1996)
 
The seasonal cycles of the locations of fire reflect the climate-generated deficits of soil moisture (see Fire and Climate pages).  Fires begin earlier in the year in the Southwest, and the locations of the maximum incidence of fire move to the northern Great Basin, Pacific Northwest and Northern Rocky Mtns. as summer progresses, before moving to California as the end of the fire season approaches.  The maps, along with time-series plots, show that the frequency of human-started fires reaches a maximum at the same time of the year as lightning-started fires, but remain high during earlier and later times of the year.
 
Lighting- and Human-Started Fires:  Point Locations and Areas
     
 
 

 

Fire-Start Data by Day of Year
 
The following plots illustrate the total number of fires each individual day over the interval for which data is available, and the total number for each day of the year.  A prominent singularity on 4 July each year can be noted for the human-started fires.
 
 
 
Small-Multiple Plots of Fire Incidence by Month and Year
 
 
Lighting- and Human-Started Fires:  Point Locations and Areas by Month
(Jan 1986-Dec 1996)
     
 
 
 

Fire-Data Sources | Daily Fire-Start Data | Seasonal Cycle |
Monthly Incidence-and-Area Data


 
 
Gridded Monthly Incidence-and-Area Data (1980-2000)
 

The Westerling et al. (2003) area-burned and number-of-fires is plotted below as both "raw" values and as values transformed into “standardized gamma indices” using the mathematical algorithms for computing the standardized precipitation index (SPI, McKee et al., 1993; Hays et al., 1996; Edwards and McKee, 1997; Guttman, 1998, 1999).  The SPI was developed as a method for describing long-term precipitation anomalies and drought indicators in a way that allows comparisons to be made across different climate regions among which the underlying statistical distributions precipitation may vary widely, e.g. from negative-exponential distributions with many zeros to Gaussian-like symmetric distributions.  This same motivation applies to the gridded fire data, where there are frequent occurrences of gridpoints and months with no fires.

The SPI is calculated by fitting a gamma distribution to the raw data, calculating gamma CDF-ordinate (i.e. probability) values for those data, and then calculating Gaussian (normal) PDF-abscissa values for these.  The computed standardized gamma value (SGI) is thus a number with values that range between -3 and +3 that can be interpreted as standardized deviations from the long-term mean for a particular month or season (or z-scores).  The SGI thereby transforms the data onto a common scale that is independent of the data type, underlying distributions, and units of a particular variable, allowing direct comparison and correlation among variables.

 
Westerling et al. (2003) 1-Deg Gridded Data
 
 

Fire-Data Sources | Daily Fire-Start Data | Seasonal Cycle |
Monthly Incidence-and-Area Data


Home | Fire | RegCM | Climate | Fire and Climate | Atlas


   

contact:  bartlein@oregon.edu; steve@coas.oregonstate.edu;  last updated:  25 Jun 2006 02:12 PM