Introduction to Vegetation Change
Vegetation Change Models
Silvicultural Growth and Yield Models
FOFEM First Order Fire Effects Model
FireStem Fire Induced Shrub and Tree Mortality Model
FVS Forest Vegetation Simulator
VDDT and TELSA Vegetation Dynamics Development Tool
SIMPPLLE Simulating Patterns and Processes at Landscape Scales
FETM Fire Effects Tradeoff Model
Fragstats
Links To Other Vegetation Models
Fire Models and Tools
Matching Models to the Management Problem

INTRODUCTION TO VEGETATION CHANGE

Just as understanding vegetation change is a critical part of forest management, it also provides the foundational structure of some belief network modeling efforts in CRAFT. Long-term projections of vegetation, fuels and habitat require a dynamic understanding of the ecosystem that incorporates incremental changes tree growth and vegetation change and a wide variety of disturbances. The interaction of plants with each other and their response to disturbance is known as succession, vegetation change or vegetation dynamics.

The word succession originally was used to refer to successive changes in gross vegetation structure and composition that occur following severe disturbance, such as crown fire or the reforestation of cleared fields. Classical succession involves progressive stage-by-stage transitions from treeless ground to complex, multistory old growth. In many forest types, however, this concept of seral vegetational change is not useful. For example, forests that historically burned frequently were sustained by low severity fire that reduced the fuels that would otherwise have led to stand replacing disturbance. Similarly, the small forest gaps associated with windthrow (gap phase dynamics) promote type-resilience at the level of the stand rather than transitional vegetational change. In both cases, individual trees die and are established over time without resulting in an appreciable change in vegetation at the stand level. In that this dynamic is characteristic in many fire-prone forests, many ecologists prefer to use the more inclusive term vegetation dynamics, rather than succession.

Forest planners are usually more concerned with mappable vegetation units, or strata, than with individual trees within a stand. Vegetation modeling is relatively easy when an entire large patch undergoes nearly uniform (seral) successional change over time. However, it is much harder to realistically model the effects of low severity disturbances that sustain vegetation at a finer scale within a stand.

Future vegetation change can be predicted through analyses of pattern and process. One of the easiest ways to predict vegetation change is through transition-state modeling. In this approach, two aerial photos, images or maps of a classified vegetation type are compared over time. Transitional probabilities between states are then calculated and then future patterns are predicted based on past rates of change. As an alternative approach, a mechanistic or causal understanding of vegetation change can help model when and where state transitions are most likely to occur, regardless of historical constraints. Most vegetation-fire models use this latter approach in a variety of ways. Different flame lengths affect the structure and composition of the vegetation in a stand.

Back to Top

VEGETATION CHANGE MODELS

The following models are commonly used in forest management to predict and describe vegetation change. Although many other models are available, the range of models described here suggests how these and similar models can be used in probabilistic effects analysis.

Back to Top

SILVICULTURAL GROWTH AND YIELD MODELS

Description

A wide variety of forest growth and yield models have been developed to forecast and better manage commercial forests. These models are comprised of algorithms that mimic silviculturally useful characteristics of individual trees or stands, such as growth, basal area, density or volume. Most growth and yield models were developed for specific areas and for a specific suite of conditions and tree species. They are especially useful for modeling growth in plantations and in intensively managed forests. An overview of growth and yield simulators that are available for west coast states is available here (Richie 1999).

Applicability to CRAFT

The use of growth and yield models for non-commercial purposes has been limited, although for many purposes these models may be best suited for predicting vegetation change. By themselves, many growth and yield models may be of limited use for predicting vegetation in less intensively managed forests because they rarely incorporate disturbances. However, in combination with other models, such as FOFEM, growth and yield models could be adapted to address a wider range of issues.

Examples of use with CRAFT

• In fire-prone plantations, use growth and yield models in combination with FOFEM to model stand resistance to fire with various thinning strategies.
• In combination with fuel or habitat models, compare alternatives that maximize forest growth for commercial purposes with changes in fire hazard or wildlife habitat.
• Use the deterministic equations in growth and yield models to calculate the range and probabilistic distribution of growth rates due to variation in climate or site productivity.

Back to Top

FOFEM (First Order Fire Effects Model)

Description

The First Order Fire Effects Model (FOFEM) predicts the effects of specified burning parameters to tree mortality, fuels, soils and air quality. FOFEM deterministically calculates mortality by tree species and diameter according to flame length.It has been primarily used to predict the effects from prescribed fire activities. Some common uses include setting fuel moisture limits for conducting prescribed burns or determining the number of acres that can be burned before exceeding particulate emission thresholds. Fuel smoke and soil effects can be calculated by forest type and by generalized fire weather conditions or specified fuel moistures. FOFEM supports an extensive list of tree species arranged by US region.

Applicability to CRAFT

Often the vegetation question of interest is relatively simple; that is, it may not require use of elaborate succession models. FOFEM's equations provide ready access to knowledge that can link fire intensity (i.e., flame length) to tree mortality.

Like other models, FOFEM was not designed to address variation within a plot or spatial differences across a heterogeneous landscape, and as such it produces output values that reflect the deterministic equations that are used. CRAFT users are interested in knowing the range of possible effects that may be possible, not just a single set of specified values that may or may not occur in the future. A broad range of output values can be generated in FOFEM by varying input parameters in response to the categories defined in the user’s belief network. These calculated values can be used in probability tables within CRAFT.

Examples of use with CRAFT

• Calculate the range of tree mortality that would result from a tree diameter or height that can be compared with structural variation in plots.
• Calculate the range of expected tree mortality from possible flame lengths derived from BehavePlus.
• Document variability in fuel consumption according to differences in fuel moisture
• Calculate the range of smoke emissions that could result from variable fuel conditions within a burn unit when the actual fuel loads are largely unknown
• Calculate the range of soil heating that is possible based on variation in initial fuel conditions or fuel moistures for Burned Area Rehabilitation (BAER) assessments
• Use FOFEM in combination with other models within a belief network to address long-term or broad-scale questions.

Website: FOFEM

Back to Top

FireStem (Fire-Induced Shrub and Tree Mortality Model)

Description

FireStem is new software designed to improve predictions of fire-induced tree mortality. Eventually it will be incorporated within BehavePlus to more efficiently link fire behavior parameters with fire effects.

Applicability to CRAFT

Use of FireStem within CRAFT is similar to that of FOFEM (see above section for applicability and examples of possible use). Different results obtained from FOFEM and FireStem help frame the scientific uncertainty of modeling predictions.

Website: FireStem

Back to Top

FVS (Forest Vegetation Simulator)

Description

FVS is a stand-level simulator designed to compare treatment alternatives over time. When used in combination with regional variants of the Fire and Fuel Extension (FFE), FVS provides a useful tool for predicting the effects of treatment prescriptions and disturbances on fuel, stand composition and stand structure over time. The deterministic equations in FVS-FFE predict the effect of different harvesting, fuel treatment and burning scenarios on stand development and potential fire intensity. Simulations are performed on a decadal basis for the entire stand as a unit. FVS can accept both forest stand exam and Forest Inventory and Analysis (FIA) data.

Applicability to CRAFT

In that FVS-FFE is a deterministic model, the software by itself can not be used to perform probabilistic analyses. CRAFT includes a program, called DecNet, that provides an interface between FVS and belief network software.

There are at least two ways that FVS (or any stand level succession model) can incorporate probability. First, for stand-level questions, multiple runs of FVS can be performed that vary fire frequency or flame length according to a probability distribution. Variation in fire weather conditions, and its effect on fire behavior, leads to the greatest uncertainty in modeling fire effects. Second, when the management question requires modeling vegetation across a broad landscape, FVS runs for individual stands (classified polygons or strata in GIS) can be performed separately and compiled to generate a landscape view. A key uncertainty in such landscape modeling is knowing what conditions are like in polygons that have no vegetation data. INFORMS-Most Similar Neighbor analysis assigns the most likely value to these polygons in GIS. In CRAFT, the entire range and distribution of conditions of similar strata is needed, not just the single "most likely" value. In CRAFT, vegetation attributes are assigned to a given strata for probabilistic analysis of a broader landscape question.

Examples of use with CRAFT

• Model the sensitivity of stand dynamics to probabilistic variation in fire weather and fuel moistures.
• Vary fuel conditions within a stand based on the range of possible values based on within-stand sampling.
• Vary vegetation strata attributes for a polygon according to the range of possible conditions derived from sampling that vegetation type.

Website: FVS

Back to Top

VDDT (Vegetation Dynamics Development Tool) &
TELSA (Tool for Exploratory Landscape Scenario Analysis)

Description

The Vegetation Dynamics Development Tool (VDDT) is a framework for modeling vegetation dynamics and disturbances for improving landscape planning. VDDT is adaptable for a wide range of ecosystems and landscapes. Individual users define the successional classes, pathways and transitional probabilities for each "potential vegetation" type. Transitions are calculated on an annual basis.

Compared to other stand-alone vegetation change models, VDDT excels in its ability to integrate multiple influential factors within a single framework. By allowing users to individually design successional relationships among vegetation classes, VDDT provides flexibility for improved place-based modeling. An additional strength of VDDT is that it allows modelers to analyze the sensitivity of their successional model to its embedded assumptions.

Unlike FVS or FETM, VDDT does not use actual fuel or stand data, and so fire intensity cannot be calculated based on local fuel conditions and fire weather variability. Fire occurrence probabilities are assigned for each vegetation (polygon) unit according to the successional state of the pixel, not based on its neighbors (i.e., VDDT has no fire spread component). VDDT is not spatially explicit by itself, however, TELSA extends VDDT's successional networks to the landscape scale allowing planners to assess how varied management actions and disturbances affect broader scale vegetation outcomes.

Applicability to CRAFT

VDDT can be embedded within a CRAFT temporal network that has a successional and disturbance component. The disturbance factors and some relationships that users create within VDDT can be mirrored by CRAFT, but successional pathway diagrams in VDDT are not analogous to belief networks. Belief networks can have no cycles without explicit time steps; that is, a belief network should not have feedback loops as are typical of VDDT successional transition diagrams.

At present, no software is available to perform multiple iterations of VDDT for building conditional probability tables in CRAFT. VDDT's use of probability may allow modelers to integrate probabilistic VDDT output with CRAFT, depending on the broader problem addressed within CRAFT.

Examples of use with CRAFT

• Calculate a range of landscape successional outcomes using different transition probabilities.
• Calculate a range of landscape successional outcomes using different probabilities of disturbance (e.g., fire, pathogens, etc.)

Websites: VDDT and TELSA

Back to Top

FETM (Fire Effects Tradeoff Model)

Description

FETM is a landscape disturbance model designed for forest planning using different management strategies. It is non-spatial in the sense that it calculates results for fuel condition classes (FCCs) or for the landscape as a whole rather than for individual vegetation units on the ground. FETM tracks the areal extent of 188 unique FCCs that result from different treatments, prescribed fire, wildfire and successional change. Unlike many other landscape vegetation models, FETM calculates fire behavior using real stand and fuel data, topography and weather. Fuel changes over time increase the acres burned. Importantly, FETM also is capable of calculating fire intensity, which is important in forests that are prone to fires of mixed severity. FETM incorporates the fuel consumption and emissions algorithms of CONSUME. FETM calculates seven air quality parameters and how they differ according to varied treatments over the long-term. FETM calculates the net present value of the costs and benefits that surround different management alternatives.

Applicability to CRAFT

FETM is a useful tool for modeling changes in fuel and emissions to address management problems at the landscape-scale. In that FETM's output presents summary data by vegetation types across management alternatives, its use in a belief network in CRAFT is appropriate to address problems at that scale. In addition, some stand-level questions can be addressed with FETM or CONSUME.

At present, no software is available to perform multiple iterations of FETM or CONSUME for building conditional probability tables in CRAFT. However, a range of conditions (i.e., user-defined iterations) can be run to improve a belief net's breadth of analysis.

Examples of use with CRAFT

• Describe how fuel consumption differs in response to variable fuel moisture or humidity.
• Determine variation in the smoke emissions that could result from different proportions of sound versus rotten fuel.
• Consider fuel in response to a range of possible time periods between harvesting and fuel treatment.

Website: FETM

Back to Top

SIMPPLLE (SIMulating Patterns and Processes at Landscape scaLEs)

Description

SIMPPLLE is designed to characterize the range of possible successional conditions of a broad landscape. SIMPPLE provides less detail that that provided by FVS, as it it landscape model, rather than a stand-level model. Vegetation change is modeled probabilistically along different pathways that allow probabilistic changes between classified vegetation types. Vegetation classification can include both structural and compositional attributes. Growth and disturbance dynamics can be modeled decadally or yearly.

SIMPPLLE is spatially explicit in that individual polygons are assigned values and vegetation change is affected by that of adjacent units. Individual simulations can be used to describe the most likely, worst case and most optimistic scenario, given the specified set of assumptions.

Applicability to CRAFT

Probabilistic output from SIMPPLLE can be used to inform belief networks in CRAFT to address broader management questions. SIMPPLLE's emphasis on probability is largely consistent with that of CRAFT, although planners may want to probabilistically model more parameters than what SIMPPLLE permits. For example, SIMPPLLE's use of fire weather is rudimentary compared to fire models such as FARSITE or FlamMap, and fire weather is among the most important probabilistic variables for predicting vegetation change.

As with other landscape models, it is difficult to decide what sized area is appropriate for analysis. Recent megafire events demonstrate that an area's "fireshed" can be much larger than historic record suggest. By calculating the conditional probability of fire risk in CRAFT, multiple runs of SIMPPLLE can be performed, to document how these assumptions influence future vegetation conditions. The sensitivity of SIMPPLLE output to other variables embedded in the software, such as state transitions, can not be readily analyzed.

At present, no linking software is available to perform multiple iterations of SIMPPLE for building conditional probability tables in CRAFT. As in other vegetation models, a range of conditions can be run to expand a belief net's breadth of analysis.

Examples of use with CRAFT

• Model the range of vegetation conditions and disturbances that are possible and likely for a landscape.
• Use SIMPPLLE's GIS output to inform vegetation layers for use in FARSITE to model multiple fire spread scenarios over a range of fire weather conditions.

Website:

SIMPPLLE

Back to Top

FRAGSTATS

Description

FRAGSTATS differs from other software discussed in this section in that it is not a succession model and it does not directly model fire effects. Instead, it is a useful tool for describing landscape features that result from other landscape modeling runs. It provides a wide assortment of metrics including patch size, connectivity, edge effects, point pattern analysis, and nearest neighbor analysis.

Applicability to CRAFT

Landscape patterns are an important component of many forest management problems. Spatial patterns control the movement of both desirable and undesirable species and the movement of fire and pathogens through the landscape. Edge effects can influence the quality of species habitat and through changes in microclimate, a site's susceptibility to fire. In projects that include such concerns, the output of spatially-explicit fire or succession models can be analyzed in FRAGSTATS to determine the consequences of management actions.

At present, no software is available to perform multiple iterations of FRAGSTATS for building conditional probability tables in CRAFT. As in other models, typical, optimistic and worse-case scenarios can be modeled to a expand a belief net's breadth of analysis.

Examples of use with CRAFT

• Calculate modeled severe fire effects on patch size with different management alternatives.
• Calculate differences in fire spread rate near the wildland urban interface with different patch and connectivity patterns.
• Identify and characterize "fire refugia" in which overstory vegetation is most likely to persist over time using FRAGSTATS in combination with multiple runs of FARSITE.
• Calculate wildlife habitat connectivity with different alternatives and disturbance scenarios.
• Assess invasive species risk due to contagion hazards with different management and disturbance scenarios.

Website: FRAGSTATS

Back to Top

LINKS

Registry for Ecological Models: http://eco.wiz.uni-kassel.de/ecobas.html