ForestERA - Fire Modeling Tools

home > tools > fire modeling

ForestERA Fire Modeling Tools

ForestERA uses models to assess fire threat across a landscape and inform management decisions about mitigating this threat. We break down "fire threat" into two components: "fire hazard" and "fire risk." Fire hazards are considered to be the fuels available to feed a fire in a particular area, and fire risk is the probability that a fire will ignite in that area (Sampson et al. 1998). We have created models of fire hazard and fire risk for the two million acre focal area of the western Mogollon Plateau. These layers can be used individually, or in combination to assess fire threat. The descriptions that follow are for the Western Mogollon Rim (WMPALA) study area.

Fire Hazard

We used the FlamMap fire behavior program (Finney, unpublished) to create fire hazard maps. This program is a simplified version of the widely used FARSITE fire-modeling program (Finney 1998). The major distinction between the two is that FlamMap does not track fire behavior over time (e.g., spread), but, instead, presents fire behavior under a fixed set of weather conditions, and produces outputs that assume the entire landscape is burning. FlamMap is also spatially explicit and can be run using GIS raster (grid) data layers converted to ascii format in ArcInfo. Thus, FlamMap can be used to quickly create maps of fire behavior across large landscapes.

FlamMap does not provide any accounting for the potential spread of fire across the landscape. However, a fire-spread modeling capability (Finney 2002) will soon be incorporated into FlamMap which will enable evaluation of treatments designed to slow the movement (i.e., maximize the travel time) of fire across the landscape (Finney 2001). The ForestERA toolset will be updated to incorporate this new feature once it is available. In order to incorporate fire spread in our current analyses, we are using a spatial layer representing areas upwind of values at risk.

FlamMap uses the following spatial data: elevation (m), slope (degrees), aspect (degrees), average stand height (m), crown base height (m), crown bulk density (kg/m3), and canopy cover (%), as well as data from ground fuel models. We derived elevation using a mosaic of USGS 30m digital elevation models (DEMs). We derived slope and aspect from the DEM mosaic using GIS software. We generated a canopy cover layer directly from DOQs, as described in Data Layers I: Vegetation Composition and Structure. We calculated crown bulk density from our predicted vegetation layers of tree stem density and basal area using allometric equations (Cruz et al. 2003; Fulé et al 2001b; Keane et al. 1998). To create layers representing crown base height and stand height, we generated our own allometric equations, relating these attributes to the Quadratic Mean Diameter (QMD) of trees. QMD is easily calculated using tree stem density and basal area. The ground fuel model input layer was based on the dominant overstory vegetation layer and the Anderson fuel model classification system (Anderson 1982).

FlamMap also requires the input of wind data and fuel moisture content (%). We obtained this information from historical records of fuel moisture and weather conditions in the region over the last twenty years. For the wind data, we used 30 mph winds from the southwest, which is representative of the prevailing wind direction and the strongest sustained winds in this region during the fire season. Our data for fuel moisture content contains estimates of 1h, 10h, and 100h fuels for each fuel model. We have chosen to use very dry (90th + percentile) fuel moistures (Fulé et al. 2001). Foliar moisture content is set at 100%, which is extremely low and representative of drought conditions.

Two outputs from FlamMap make up our fire hazard layers: crowning behavior and heat per unit area. The crowning behavior layer classifies the behavior of the fire on any given portion of the landscape into active crown fire (crowning), passive crown fire (torching), or surface fire only. Passive crown fire occurs when the fire is spreading on the ground, but some (perhaps even many) of the trees are burning. Active crown fire occurs when fire is spreading through the canopy as well as along the ground. The heat layer is a prediction of the heat energy that would be produced by a fire, and is directly related to the crowning behavior layer. As the fire moves from surface fire to passive crown fire, and from passive to active crown fire, the amount of heat produced increases, because more fuels will burn at each successive level of fire behavior. The heat per unit area output (KJ / m²) is used to scale the threat posed by crown fire. Areas of higher heat output indicate more intense fire activity, and also correspond tightly to areas that can carry crown fire under lower wind speeds.

Fire Risk

Our fire risk layers are based on fire ignitions and can also include their relationship to the frequency of large fires on the landscape. To create our model, we obtained information on all fire ignitions recorded in Arizona between 1986 and 2000 from the US Departments of Agriculture and Interior, and from the National Fire Occurrence Database. From this dataset, we extracted all fires greater than 20 ha (50 acres) in size that occurred during the peak fire season (April - September) across the Arizona portion of our larger study area. Approximately 70% of the ignitions and large fires in our study area occur during these months. We then used a ‘weights of evidence’ analysis to predict the occurrence of large fires based on explanatory data layers (Dickson et al., 2006).

‘Weights of evidence’ is a Bayesian method of predicting events, based on known locations of event occurrence. The approach was originally developed for medical diagnoses (Spiegelhalter & Knill-Jones 1984), but has been expanded to the prediction and spatial analysis of many other patterns in occurrence data. Weights of evidence models use the location of known occurrence points to determine correlations with a set of categorical input maps (Bonham-Carter et al. 1989). For our analyses, we related ignitions to 7 spatial data layers, or input maps, that included information on physiographic, biotic, climatic, spatial, and human factors likely to influence the ignition of large fires. We created unique input maps for elevation, topographic roughness, aspect score, ponderosa pine vegetation, precipitation, road density, and spatial domain (Dickson et al., 2006).

We used the Arc-SDM (Kemp et al. 2001) spatial data modeler extension to ArcView to conduct all Weights of Evidence calculations and analyses. By determining relationships between past large fires and these landscape-level variables we created a fire risk layer that shows the relative chance of a large fire occurring in the future at any given location on the landscape. The analytical procedure also allows for the determination of uncertainty in the fire risk layer, and a separate layer can be created showing this uncertainty. The resolution of these data allows for analysis at a minimum of 1 km2 so our fire risk layer is of more course resolution than most of the other data layers we have created. Both layers can be resampled to 90m resolution, to match other data. However, the value in each cell (pixel) of those resampled layers is a value representative of the 1 km² area around the cell, and not the cell itself. Thus, any interpretation of the resampled layers should be undertaken with this in mind.

References

Alexander, M. E. 1988. Help with making crown fire hazard assessments. Pp 147-156. Symposium and Workshop on protecting people and homes from wildfire in the interior West. USDA Forest Service General Technical Report INT-GTR-251.

Andrews, P. L. 1986. BEHAVE: Fire behavior prediction and fuel modeling system - BURN subsystem. USDA Forest Service General Technical Report INT-GTR-194.

Andrews, P. L. and L. P. Queen. 2001. Fire modeling and information system technology. International Journal of Wildland Fire 10: 343-352.

Brown, J. K. 1978. Weight and densities of crowns of Rocky Mountain conifers. USDA Forest Service Research Paper INT-RS-197.

Burgan, R. E., R. W. Klaver, and J. M. Klaver. 1998. Fuel models and fire potential from satellite and surface observations. International Journal of Wildland Fire 8: 159-170.

Burgan, R. E. and R. C. Rothermal. 1984. BEHAVE: Fire behavior prediction and fuel modeling system - FUEL subsystem. USDA Forest Service General Technical Report INT-GTR-167.

Cardille, J. A., S. J. Ventura, and M. G. Turner. 2001. Environmental and social factors influencing wildfires in the upper Midwest, United States. Ecological Applications 11: 111-127.

Chou, Y. H., R. A. Minnich, and R. A. Chase. 1993. Mapping probability of fire occurrence in San Jacinto Mountains, California, USA. Environmental Management 17: 129-140.

Cruz, M. G., M. E. Alexander, and R. H. Wakimoto. 2003. Assessing canopy fuel stratum characteristics in crown fire prone fuel types of western North America. International Journal of Wildland Fire 12: 39-50.

Dickson, B. G., J. W. Prather, Y. Xu, H. M. Hampton, E. N. Aumack, and T. D. Sisk. 2006. Mapping the probability of large fire occurrence in northern Arizona, USA. Landscape Ecology: 21(5):747-761.

Dubayah, R. O, and J. B. Drake. 2000. Lidar remote sensing for forestry. Journal of Forestry 98: 44-46.

Finney, M. E. 1998. FARSITE: Fire area simulator - model development and evaluation. USDA Forest Service Research Paper RMRS-RP-4.

Fule, P. Z., C. McHugh, T. A. Heinlein, and W. W. Covington. 2002. Potential fire behavior is reduced following forest restoration treatments. Pp 28 -35 in Ponderosa pine ecosystems restoration and conservation: steps toward stewardship (R. K. Vance, C. B. Edminster, W. W. Covington, and J. A. Blake, eds.). USDA Forest Service Conference Proceedings RMRS-P-22.

Keane, R.E., D. G. Long, K. M. Schmidt, S. Mincemoyer, J. L. Garner. 1998. Mapping fuels for spatial fire simulations using remote sensing and biophysical modeling. In: J.D. Greer, editor. Proceedings of the seventh Forest Service remote sensing applications conference. American Society for Photogrammetry and Remote Sensing, Bethesda, Maryland.

Pew, K. L., and C. P. S. Larsen. 2001. GIS analysis of spatial and temporal patterns of human-caused wildfires in the temperate rain forest of Vancouver Island, Canada. Forest Ecology and Management 140: 1-18.

Sampson, R. N., R. D. Atkinson, and J. W. Lewis. 2000. Mapping Wildfire Hazards and Risks. The Haworth Press, New York.

Scott, J. H. 1999. NEXXUS: A system for assessing crown fire hazard. Forest Management Notes 59: 20-24.

ForestERA Fire Modeling Tools

Fire Hazard

Fire Risk

References

See also