The model parameters control the temporal and spatial resolution of the simulation. Greater computer processing capabilities and time are required for shorter time steps and smaller distance resolutions.
These parameters must be set for each simulation.
The time step is the maximum amount of time that the conditions at a given point are assumed constant so that the position of the fire front can be projected. The positions of all fires will be projected over this time step so that possible mergers can be computed.
A shorter time-step becomes more appropriate with faster fire spread rates. The head of the fire should be used as the determining factor. However, the following table may be a useful starting point:
|
TYPE OF FIRE |
FUELS |
TIMESTEP |
|
surface |
timber |
~30 to 120+ minutes |
|
surface |
brush, dry grass |
~10 to 20 minutes |
|
surface extreme |
all |
~5 to 10 minutes |
The time step is really of secondary importance compared with the spatial resolution of the calculations (as determined by the perimeter and distance resolutions below). The internal time step used by the simulation is constantly changing according to the minimum time required for the fire to spread the distance equaling the distance resolution. The actual time step is thus, only used as a consistent period during which all fires will be projecting to a coincident time before mergers and spotting are computed. Thus, many fires burning in close proximity where spotting is likely should have a shorter time step. This however, will increase processing time.
The visible time step is always a multiple of the actual time step; it cannot be shorter than the time step. The fire front will be drawn to the output screen at these intervals. The visible time step is often set many times longer than the actual time step to avoid unnecessary temporal detail in fire perimeter positions displayed on the monitor. The visible fire perimeter may be color-coded to reflect various frontal fire characteristics (see Output > Fire Perimeter Colors).
The setting for the visible time step is dependent on the purpose of the simulation in terms of graphically representing fire spread and fire behavior. For example, one day visible intervals may provide adequate temporal detail for long simulations (e.g. weeks); these fire perimeter isochrons however will not reveal diurnal patterns of fire behavior or spread. Shorter simulations or those intended to spatially depict varying fire spread rates throughout a burning-period would likely require time steps no longer than several hours.
NOTE: the visible time step can affect the time resolution of the vector file output from the simulation depending on your selections in the Output > Export and Output dialog box.
FARSITE includes primary and secondary visible time steps. The primary visible step must always be set. The secondary visible time step is optional and is set as a multiple of the primary visible step. This is useful for distinguishing fire growth at two meaningful time periods (e.g. hours and days). The secondary visible step will be displayed differently depending on your settings in Output > Fire Perimeter Colors they will be displayed in red if the visible fire perimeter is set to "white" and displayed as "white" if the fire perimeter is color coded by frontal fire characteristics.
The perimeter resolution determines the maximum distance between points used to define the fire perimeter. It is a resolution of a fire front in the direction tangential to the fire perimeter at each point. As fire perimeters expand, new points must be inserted between existing vertices along convex portions to maintain the definition of the fire front. The line segments are lengthening between these vertices. By contrast, segments between vertices of concave portions of a fire front become shorter over time and are eventually removed from the front as the vertices cross-over into previously burned areas (see Mergers and Crossovers in the Technical Documentation).
Setting of the perimeter resolution must consider its "importance" in both a relative and absolute sense. Relative to the resolution of the raster landscape, the perimeter resolution determines the amount of landscape information used in the simulation. In an absolute sense, the perimeter resolution defines the scale of possible curvature and detail of the fire front. Both must be weighed in the context of the simulation purpose.
A coarse perimeter resolution will decrease the use of spatial information in computing fire spread. Less information will remain unused, however, for perimeter resolutions shorter than the raster resolution. Using a perimeter resolution that is several times longer than the raster resolution would allow the fire to skip-over variations in fuels or topography that have a finer scale. This can be acceptable where the perimeter resolution matches a scale of interest much coarser than the raster landscape; for example if FARSITE is used to simulate fires say 30,000ha using a raster resolution 30m resolution.
The perimeter resolution controls the detail of the fire front, both in curvature and in the ability of a fire perimeter to respond to heterogeneities occurring at a fine scale. Finer resolution is necessary to make a spreading fire sensitive to small scale variations in spatial variables (e.g. fuels or topography varying on the scale of a few cells) as well as temporal changes in wind directions. Coarse resolution compared to the raster resolution would limit the potential for fire to spread through narrow corridors of fuel when surrounded by larger areas of non-fuel. Note that there is no gain in meaningful precision when setting the perimeter resolution much lower than about half the native raster resolution.
Setting a meaningful perimeter resolution can be based on the practical knowledge gained through mapping of actual fire fronts. Most fire maps have an implicit resolution, either determined by lack of data or often by the observer's threshold for irrelevant detail. This limit of relevant detail translates directly to a useful perimeter resolution. Until sensitivity analyses have been conducted for known fires or experience suggests otherwise, a logical perimeter resolution would be no more than about twice the raster resolution of 30m.
The distance resolution is the maximum projected spread distance from any perimeter point. This distance cannot be exceeded in a time step before new fuels, weather, and topography data are used to compute the spread rate (see Huygens Principle in the Technical Documentation). It is the resolution in the radial or spreading direction for each point. The values of the distance resolution cannot be greater than the perimeter resolution because of the methods used for detecting and eliminating crossovers. FARSITE automatically adjusts the distance resolution to a working internal value for each fire that cannot exceed this maximum value. This adjustment is necessary for very small fires that require a fine perimeter resolution and thus a fine distance resolution.
The distance resolution is somewhat dependent on the perimeter resolution because it cannot exceed that distance because of the potential for complex crossovers on the fire perimeters (see Mergers and Crossovers in the Technical Documentation). However, the distance resolution can be set shorter than the perimeter resolution. This would give a greater radial than tangential resolution. Considerations for deciding the distance resolution are similar to those pertinent to the settings for the perimeter. The most logical value for distance resolution would be approximately the same as that for the perimeter.