home > data and maps > derived data > fire hazard

ForestERA Data Layer Details - Fire Hazard (KJ / m²)

Note to data users: Please carefully review the metadata provided with each layer. We request that users consult with the ForestERA project in advance of using these data in publications and/or presentations to ensure that the strengths and limitations of the data are considered.

Description

A number of output layers from the FlamMap fire-modeling program (Finney, in press) can be used in management planning. Many of these layers are useful as surrogates for fire hazards, which are generally defined as the types and amounts of fuels available to feed a fire (Sampson et al., 2000). The two output layers we use most often for this purpose are heat per unit area (.hpa) and crown fire behavior (.cfr).

The heat layer is a prediction of the heat energy that would be produced by a fire as it burned across a particular part of the landscape. Since heat output is primarily related to the amount of fuels that would be burned, this layer is a good indicator of fire hazard. It was intended for use in forest planning as an index of fire hazard, or the danger presented by a burning fire. The units of this layer are in KJ / m2, but this can be easily converted to BTU / ft2 by multiplying by 0.098.

The crown fire 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 ground 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 the fire is spreading through the canopy as well as along the ground. The heat layer is directly related to the crown fire behavior layer. As the fire behavior moves from ground fire to passive crown fire, and again from passive to active crown fire, there is a jump in the amount of heat produced since more fuels are assumed to be burned as the fire moves through the area.

Purpose

These layers were created as part of the ForestERA project to support landscape-scale forest restoration planning efforts by a broad group of stakeholders including federal and state agencies, academic institutions, and non-governmental entities. These data are intended for regional analyses over spatial extents on the order of tens to hundreds of thousands of acres, and were not developed for use at finer spatial scales, although they may be useful for some applications at finer scales.

Development

Fire hazards can be mapped in a variety of ways, but using fire modeling programs, which allow for modeling of fire behavior under specific weather conditions, is a popular methodology. Most fire modeling programs provide a variety of outputs that can help assess how fuels, topography, and weather conditions affect fire behavior, providing more information about fire hazard than simple fuels mapping alone.

We have chosen to use the fire-modeling program FlamMap for our fire hazard modeling (Finney, in press). Although, this application has only been used for modeling over the past few years, it is similar to the widely used FARSITE modeling program (Finney, 1998) and uses the same input layers. However, unlike FARSITE, FlamMap does not model fire spread across the landscape over time. Instead, FlamMap uses the assumption that the entire landscape is burning and predicts the fire behavior across the landscape under a single set of weather conditions. Thus, while FlamMap is not dynamic, it does allow easy comparison of fire behavior over large areas. To do the same with a dynamic fire modeling programs would require many runs of the application.

Fire behavior models require eight inputs that are considered part of a fuel hazard model (Finney, 1998; Keane et al., 1998). We have derived the first three (slope, aspect and elevation) from a 30m resolution Digital Elevation Model (DEM) created by the United States Geological Survey. Four more (canopy closure, mean tree height, crown base height, and crown bulk density) are forest structural characteristics. The canopy cover layer was developed directly by our team from DOQ imagery (Xu et al., in review). The crown bulk density layer was created using metrics (Cruz et al., 2003) applied to the basal area and tree density layers derived from satellite imagery by our team. The mean tree height and crown base height layers were developed by first deriving a quadratic mean diameter layer from our basal area and tree density layers. Using ground data on individual trees, we determined that for most species both mean tree height and mean crown base height are highly correlated with quadratic mean diameter in this region. We used these relationships to develop allometric equations to derive the stand height and crown base height layers (Prather, unpublished data). For reasons explained below, we used the lowest 20th percentile of crown base height, rather than average crown base height, when undertaking fire modeling. The last of the eight input layers is a ground fuels model. This layer is basically a generalized assessment of ground fuels based on dominant vegetation (Anderson, 1982). To create this layer we used a dominant overstory vegetation map created by our team using satellite imagery. Vegetation types were linked to the 13 fuel models developed for the BEHAVE fire modeling program (Burgan and Rothermal, 1984) to create the final layer.

In addition to the layer inputs above, the user must define live and dead fuel moistures, wind direction, and wind speed. We used fuel moistures approximating 95th percentile drought conditions and sustained wind speeds of 30 mph (representative of extreme fire weather) for this example. We used wind blowing from the southwest (225o) for this output, since this is the direction of the prevailing wind during the fire season in this region.

There are a number of possible outputs from FlamMap that could be useful in assessing fire hazard. We have chosen to use the heat per unit area (.hpa) and crown fire behavior (.cfr) outputs as our primary fire hazard assessment layers. These outputs are probably the best representations of the relative danger presented by a fire, as they depend primarily on the types and amounts of fuels available for the fire to burn. However, wind and terrain also play a role, as the fire will burn hotter in areas exposed to the wind. The values of both the heat and crown fire behavior layers depend primarily on crown bulk density. However, crown base height and fuel model play a significant role in determining when a fire transitions from ground to crown and from passive to active crown fire behavior. As a general rule, crown base heights above 2.5m will rarely result in crown fire when using the fuel models typically used in this region.

Other potential layers that might be useful for fire hazard modeling include a fireline intensity layer, a flame length and several outputs representing the speed and direction of fire spread. The fireline intensity layer is a measure of the heat produced by the fire front per unit time, while the flame length layer estimates flame lengths at the fire front. Both of these layers are very sensitive to fuel model. For example, a grassy fuel will create a more intense fire with longer flame lengths than a pine needle fuel. Therefore, these layers may not be as useful as heat and crown fire activity for identifying areas of high fire hazard.

Accuracy Assessment

There is no accuracy assessment possible for this layer. A measure of uncertainty can be obtained for many, but not all, of the input layers. However, the accuracy of the crown bulk density layer, which is one of the most important input layers, has been documented in the metadata for that layer. The slope, aspect, and elevation layers are derived from a digital elevation model obtained from the USGS, and meet all federal standards for accuracy for terrain data. The canopy cover layer is our own product and its accuracy is documented in the metadata for that layer. The fuel models layer is a direct derivative of our vegetation type layer, for which the accuracy is documented in the vegetation composition metadata. The stand height and crown base height layers are the most uncertain of the layers used in the fire modeling as they depend on the use of a quadratic mean diameter layer created using an equation based on basal area and tree density. The equations we use to derive crown base height and stand height are strong, but because these layers are derived from other predictive layers, errors can accumulate. Please see the metadata for the basal area, tree density, crown base height, and stand height layers for more information.

The primary role of the stand height layer is to determine wind speed at ground level for purposes of determining flame lengths and fire intensity. As long as stand height values are reasonable, this layer will have little impact on the heat output layer. However, the crown base height layer is very important in determining whether a fire will become a crown fire. Areas with crown base height over 2.5 meters have little chance of crowning. In order to account for this, we use a low percentile (20th) of crown base height rather than mean crown base height when running FlamMap. The justification for doing this is outlined in the next section.

Sources of errors

It is important to remember that there are many possible versions of this layer. Entering different values for fuel moistures, wind speed, and wind direction will result in changes to the output layers. In addition, many local factors not taken into account by FlamMap will affect how a fire actually behaves at a given location. Finally, uncertainty in the input layers is likely to be passed on to the fire modeling output. Thus, viewing the output layer as a generalization that is useful for comparison of different locations on the landscape is more realistic than assuming the values for an output layer are correct.

A second important note is that the outputs from this layer were created assuming that crown fire could occur at nearly any location on the landscape. For modeling purposes we used the lowest 20th percentile of crown base height rather than mean crown base height. This is because the fire spreads to the canopy through the lowest canopy fuels. Using a low percentile of crown base height maximizes the area over which the fire can spread into the canopy. Thus, our fire behavior output layers do not give any information about how likely it would be for a crown fire to occur in a given area, but is intended to allow for crown fire to occur over most of the landscape. There are several reasons for developing the layer this way, including the fact that available ground fuel models typically do not adequately predict flame lengths in ponderosa pine areas, crown base heights are highly variable across the landscape and there is tremendous uncertainty in the actual values, and ladder fuels cannot be adequately represented in the fire-modeling program. However, if users are comfortable with the inherent uncertainty in the crown base height layer, the layer of mean crown base heights can be used as an index of the likelihood of a crown fire occurring at any location on the landscape. In reality, very few locations on the landscape would be unlikely to have crown fire under extreme fire weather conditions. Therefore, we feel our fire modeling output layers are reasonable for assessing fire hazard across the landscape.

Use of this layer

We feel the outputs from FlamMap provide a reasonable representation of fire behavior across the landscape. As mentioned above, output values should be treated as approximations of fire behavior that can be interpreted relative to each other and not as absolute predictions of fire behavior. There are a very high number of potential outputs from the FlamMap fire-modeling program depending on assumptions and values entered when running the program. Based on our knowledge and experience with fire modeling, the layers we have produced are good outputs for assessing fire hazard in this region. However, we recognize that others experienced in fire modeling may wish to use their own assumptions, and that other output layers could be used to help guide forest management.

References

Anderson, H. E. 1982. Aids to determining fuel models for estimating fire behavior. USDA Forest Service General Technical Report INT-GTR-122.

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.

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 Fire12: 39-50.

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

Finney, M. A. In press. FlamMap: fire behavior mapping and analysis system. USDA Forest Service General Technical Report.

Keane, R. E., D. G. Long, K. M. Schmidt, S. Mincemoyer, and J. L. Garner. 1998. Mapping fuels for spatial fire simulations using remote sensing and biophysical modeling. In: J.D. Greer, editor. Proceedings of the 7th Forest Service remote sensing applications conference. American Society for Photogrammetry and Remote Sensing, Bethesda, Maryland.

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

Page last updated February 23, 2005