348
raster-based data, especially satellite imagery; a
drawback is the automatic limitation to a specific
resolution of data (Goodchild, 1994).
Spatial models, although computationally demanding, offer the greatest flexibility for parameterization and realistic responses. They rely heavily on the concepts of gradient analysis, that is, the
way in which vegetation structure and composition
steadily change along an array of environmental
conditions, such as altitude, moisture availability,
sheltering, nutrient availability, or disturbance impact. In simulation modeling, gradient analysis
comes into play when a multidimensional gradient
analysis and a site-specific stand inventory are
linked by computer modeling software (Kessell,
1979). Parameterizing a spatially explicit disturbance simulation model involves linking a specific
set of conditions (physiographic, climatic, vegetative) with a set of rules, usually mathematical equations, that describe disturbance behavior under a
range of conditions. Early models were usually
based on a spatially heterogeneous but entirely simulated landscape. Recent advances in GIS-based
data allow for the use of real landscapes, with precise georeferencing of environmental conditions,
which can then be used to simulate any number of
alternative management scenarios (Vasconcelos et
aI., 1994).
A prime example of the model-building process
is provided by models of fire behavior. Many fire
models are based on the fire spread equations of
Rothermel (1972), which predict fire behavior (intensity and rate of spread) with homogeneous and
continuous fuel availability as a function of detailed meteorological information and fuel characteristics (Davis and Burrows, 1994). These rules
of fire behavior are then given spatial dimension
by applying them to a digital elevation model.
Topography controls direction and rate of spread
by affecting the exposure of fuels to wind and the
drying effects of solar radiation (through slope aspect), flame length and degree of flame contact
(through slope steepness), and the presence and extent of contagion barriers and corridors, a function
of the degree of topographic dissection. This landform-based analysis of disturbance propagation
(fire spread) takes into account any existing natural directions of disturbance movement and the
effects of natural barriers. The final element, land
cover information in the form of either detailed
stand inventories or satellite imagery, is incorporated into the spatial data to provide information
as to the types and condition of available fuels.
These data are augmented by information concerning meteorological conditions, postfire sucDynamic Terrestrial Ecosystem Patterns and Processes
cession, fuel accumulation, the effects of disturbance on vegetation, and climatic variation. Fire
ignition frequency and timing may be generated
statistically or explicitly, depending on the
model's intentions.
Simulation modeling of fire behavior is not without several drawbacks. An obvious obstacle is a
need for access to sophisticated computing facilities and skilled programmers. A second obstacle is
a need for detailed data for parameterization, including detailed meteorological data (temperature,
humidity, wind speed, wind direction) and detailed
fuel data (fuel per land area, fuel size distributions,
height of the fuel bed, fuel heat content, particle
density, and moisture content) (Davis and Burrows,
1994). For long-term modeling, as would be required to determine the historical range of variability or to assess the likelihood of a patch dynamic equilibrium, a model must also be coupled
with successional models in order to represent variation in fuel conditions over time. Finally, models
will likely have to be spatially explicit, that is, to
include the effects of patch size, shape, and adjacency. Because of the lack of such detailed data,
many models depend heavily on simplifying assumptions. In most modeling activities, it is not a
lack of technology that prevents effective model
building, but rather a lack of knowledge about the
processes in the landscape and how to best represent these processes. A final drawback is the difficulty of testing models against observations to
discern where assumptions have been made incorrectly. Nonetheless, simulation models are valuable
tools that allow managers to predict the consequences of alternative management scenarios, particularly when there are many and interacting variables of interest.
23.5.4 Performing the Assessment
The process of assessing dynamic pattern in a particular landscape includes, first, the analysis of all
available historical data, followed by the analysis
of observational data from the site. Gaps in this information may be filled by both historical and observational data from similar sites and/or by simulation modeling when no other information is
available. Assessing dynamic pattern is not merely
a matter of describing the patterns of past and current disturbances, along with the successional stage
and functionality of the study area; it uses these
analyses to place the site in its broader spatial and
temporal context. Table 23.1 outlines an appropriate approach, as well as a series of questions that
should be addressed during the assessment process.
raster-based data, especially satellite imagery; a
drawback is the automatic limitation to a specific
resolution of data (Goodchild, 1994).
Spatial models, although computationally demanding, offer the greatest flexibility for parameterization and realistic responses. They rely heavily on the concepts of gradient analysis, that is, the
way in which vegetation structure and composition
steadily change along an array of environmental
conditions, such as altitude, moisture availability,
sheltering, nutrient availability, or disturbance impact. In simulation modeling, gradient analysis
comes into play when a multidimensional gradient
analysis and a site-specific stand inventory are
linked by computer modeling software (Kessell,
1979). Parameterizing a spatially explicit disturbance simulation model involves linking a specific
set of conditions (physiographic, climatic, vegetative) with a set of rules, usually mathematical equations, that describe disturbance behavior under a
range of conditions. Early models were usually
based on a spatially heterogeneous but entirely simulated landscape. Recent advances in GIS-based
data allow for the use of real landscapes, with precise georeferencing of environmental conditions,
which can then be used to simulate any number of
alternative management scenarios (Vasconcelos et
aI., 1994).
A prime example of the model-building process
is provided by models of fire behavior. Many fire
models are based on the fire spread equations of
Rothermel (1972), which predict fire behavior (intensity and rate of spread) with homogeneous and
continuous fuel availability as a function of detailed meteorological information and fuel characteristics (Davis and Burrows, 1994). These rules
of fire behavior are then given spatial dimension
by applying them to a digital elevation model.
Topography controls direction and rate of spread
by affecting the exposure of fuels to wind and the
drying effects of solar radiation (through slope aspect), flame length and degree of flame contact
(through slope steepness), and the presence and extent of contagion barriers and corridors, a function
of the degree of topographic dissection. This landform-based analysis of disturbance propagation
(fire spread) takes into account any existing natural directions of disturbance movement and the
effects of natural barriers. The final element, land
cover information in the form of either detailed
stand inventories or satellite imagery, is incorporated into the spatial data to provide information
as to the types and condition of available fuels.
These data are augmented by information concerning meteorological conditions, postfire sucDynamic Terrestrial Ecosystem Patterns and Processes
cession, fuel accumulation, the effects of disturbance on vegetation, and climatic variation. Fire
ignition frequency and timing may be generated
statistically or explicitly, depending on the
model's intentions.
Simulation modeling of fire behavior is not without several drawbacks. An obvious obstacle is a
need for access to sophisticated computing facilities and skilled programmers. A second obstacle is
a need for detailed data for parameterization, including detailed meteorological data (temperature,
humidity, wind speed, wind direction) and detailed
fuel data (fuel per land area, fuel size distributions,
height of the fuel bed, fuel heat content, particle
density, and moisture content) (Davis and Burrows,
1994). For long-term modeling, as would be required to determine the historical range of variability or to assess the likelihood of a patch dynamic equilibrium, a model must also be coupled
with successional models in order to represent variation in fuel conditions over time. Finally, models
will likely have to be spatially explicit, that is, to
include the effects of patch size, shape, and adjacency. Because of the lack of such detailed data,
many models depend heavily on simplifying assumptions. In most modeling activities, it is not a
lack of technology that prevents effective model
building, but rather a lack of knowledge about the
processes in the landscape and how to best represent these processes. A final drawback is the difficulty of testing models against observations to
discern where assumptions have been made incorrectly. Nonetheless, simulation models are valuable
tools that allow managers to predict the consequences of alternative management scenarios, particularly when there are many and interacting variables of interest.
23.5.4 Performing the Assessment
The process of assessing dynamic pattern in a particular landscape includes, first, the analysis of all
available historical data, followed by the analysis
of observational data from the site. Gaps in this information may be filled by both historical and observational data from similar sites and/or by simulation modeling when no other information is
available. Assessing dynamic pattern is not merely
a matter of describing the patterns of past and current disturbances, along with the successional stage
and functionality of the study area; it uses these
analyses to place the site in its broader spatial and
temporal context. Table 23.1 outlines an appropriate approach, as well as a series of questions that
should be addressed during the assessment process.