to
102
Jimmie D. Chew, Christine Stalling, and Kirk Moeller; Rocky Mountain Research
Station, Forest Service, USDA Forestry Sciences Laboratory, P.O. Box 8089, Missoula,
MT 59807.
ABSTRACT: Managers ofpublic landsare increasinglyfaced with makingplanning decisionsfor dynamic
landscapeswith conflicting objectives.A modelingsystemhasbeendesignedto serveas a decisionsupport
systemto help managersand resourcespecialistsintegrate the available knowledgeof vegetationchange
and disturbanceprocesses,and quantify conceptsthat are often difficult to interpretfor specificlandscapes.
The systemis named SIMPPLLE,an acronymtakenfrom "SIMulating vegetationPatterns and Processes
at LandscapescaLEs." SIMPPLLE can be usedto help define and evaluatefuture conditionsat landscape
scales,to identify areas that are moreprone to disturbancesover a given timeframe, to identify the options
for influencing these disturbance processes,and to help design and evaluate different strategiesfor
achievingdesiredfuture conditions.The emphasisin this article is to give an overview of the design of the
system,the types of knowledge integrated,and the type of output produced. The initial validation work
discussedindicatesthat the approachusedfor capturing and integratingprocessknowledgein SIMPPLLE
doespredict realistic results at landscapescales.SIMPPLLEprovides managersa tool to integrate and
interpret conceptsof desiredfuture conditions,range of variability, and the interaction betweenvegetation
patterns and disturbanceprocesses.SIMPPLLE provides a way to help evaluateproposed management
scenarioswithin a future that includesstochasticprocesses. West.J. Appl. For. 19(2):102-108.
Key Words: Disturbanceprocesses,simulationmodels,'landscapemodels,insect outbreaks,wildfIre.
Land managementfor the USDA ForestServiceis a continuIng evolution of designing and applying management
practices in responseto changingdemandsby societyand
an increasedawarenessof ecologicalconcepts.This evolution has grown from an emphasison the effects on individual plant communities to a concern with the cumulative
effects on many individual communitieswithin landscapes
at a range of spatial scales. Managementof landscapes
attemptsto incorporateconceptsexpressedas "desired future conditions," "historic range of variability," "dynamic
disturbanceprocesses,"and "interactionsbetweenprocesses
and vegetationpatterns."A modeling environmentthatcaptures and integratesthe available knowledge of vegetation
change and the processesthat drive the changecan assist
incorporationof theseconcepts.
This article presents a modeling system designed for
simulating vegetationpatternsand processesat a range of
spatial scales.The systemis named SlMPPLLE, an acronym taken from "SIMulating vegetation Patterns and
Processesat LandscapescaLEs." Our primary objective is
NOTE:
Jimmie Chew can be reached at (406) 542-4171;
jchew@fs.fed.us.Copyright @ 2004 by the Society of American Foresters.
WJAF 19(2) 2004
provide an overview of the design of the SIMPPLLE
system, the types of data and expert knowledge incorporated
into the model logic, and the format of output available tousers
for incorporating simulation results into landscape
management planning. Examples of the work used to verifythe
system's performance are presented.
Model Design Criteria
SIMPPLLE is designedto serve as a decision support
systemto help managersand resourcespecialistsquantify
and incorporateconceptsthat are often difficult to interpret
for specific landscapes.Managerscan use the SIMPPLLE
systemto helpdefine and evaluatedesiredfuture conditions
at landscapescales,to identify whatparts of a landscapeare
more prone to disturbanceprocessesover a given time
frame, and to help designand eval:uatedifferent strategies
for achievingdesiredfuture conditions.As with the work by
Baker (1992)on modelinglandscapestructure,this modelis
not intendedto predict preciselywhenand whereprocesses
will occur. Rather,the objectiveis to provide a predictionof
behavioraltrends. The emphasisis on behavioral validity,
not on numerical precision. The relationships between
trends in vegetationconditions and insect activity suchas
the maturing of lodgepole pine (Pinus contorta) and an
increase in mountain pine beetle (Dendroctonusponderosae) activity are more important than the actual acres of
mountainpine beetle activity simulated.The simulatedrelationshipbetweenthe mountainpine beetle activity andfire
processesis more important than the simulated acres of
either process.Spatially explicit output from single silllulations can be provided as possible outcomesand the output
from multiple stochasticsimulationscanbe usedto estimate
the probability of disturbance processesand vegetation
attributes.
The systemis designedto be consistentwith the field
inventories and satellite imagery that exist for the range of
landscapescaleswithin the ForestService.The vegetation
attributes are limited to a dominant speciesor cover type,
size class and structure, and canopy closure. Nonspatial
attributes can co~mefrom queries on vegetationdatabases
maintained by the ForestService.The use of a geographic
information system(GIS) providesthe meansto identify the
setof neighborsfor eachplant communityso thatthe unique
pattern of each landscapecan influence disturbanceprocesses.A variety of commercially available GIS software
packageshave beenused. However,becauseof the goal of
designingSIMPPLLE as a managementtool for the Forest
Service,customizedArcInfo utility functions and ArcView
project files have beendeveloped.
The initial emphasisin systemdevelopmentwas to provide the means to representand integrate the available
knowledge on disturbanceprocessesand vegetationconditions and patterns. Much of the initial knowledge on relationships betweendisturbanceprocessesand betweenprocessesand vegetationpatternhas come from expertopinion.
Rigorousmethodologiesare available for the stepsof quantifying expert opinion (Reynoldsand Holsten 1994). However, for this first version of SIMPPLLE, this information
was gatheredthrough a series of workshops with silviculturists, ecologists,entomologists,and pathologistsfrom the
ForestService. With the system's design,the initial knowledge from both expert opinion and researchresults can be
easily replaced as new information and researchresults
becomeavailable.
General Model Characteristics
SIMPPLLE was designedto be spatially explicit because
of the significanceof the interactionbetweenprocessesand
vegetation patterns (Forman and Godron 1986, Turner
1989). Eachexisting vegetationunit is representedindividually. A probability for eachdisturbanceprocessis calculated for each vegetationunit. Each unit's unique set of
neighborshas an influence on the probability. Simulations
can be made individually or in multiples. Multiple simulations are usedto provide an averagelevel of conditionsand
a range. Simulations can be made with or without fire
suppressionand vegetationtreatments.Changeis simulated
based on either decadeor yearly length time steps.
Model Components
Existing Vegetation
The attributes used to describe a vegetation unit must
addressthree importantcriteria. First, the attributesmustbe
possibleto obtain from availableinventorydata; second,the
attributesmustbe of sufficient detail to enablepredictionof
processprobability; and third, the attributes must contain
enough information to make interpretations for specific
resourcessuchas wildlife habitat.The inventoriesavailable
are often a combinationof data from on-the-groundsurveys
and interpretationfrom aerial photographsor classified satellite imagery. An existing vegetationunit is describedby a
combinationof habitattype (Pfister et al. 1977), dominant
species,size-classand structure, and density. These attributes are consistentwith the hierarchical inventory system used by Northern Regionof the ForestServiceand are
sufficient to use other knowledge that has beendeveloped
such as the hazardrating systemsused for mountainpine
beetle (Amman et al. 1977), or westernspruce budworm
(Choristoneuraoccidentalis)(Carlsonand Wulf 1989).
Potential Vegetation States
Although vegetationdevelopmentis a processof continuously changing species,size class, structure,and density
characteristics,it is often convenientfor modelersto view
the community as making transitions from one state to
another(Kesselland Potter1980).The continuumis divided
into a suitable number of statesbased on the knowledge
availableand the resolutionneededto addressthe management issues.It is assumedthat the likelihood and intensity
of disturbanceprocessescan be associatedwith thesediscretevegetationstatesbasedon the interactionof vegetation
with fuel loadings,life history characteristics,dispersalinteractions,andresourceavailability (Pickettand McDonneil
1989). This approachhas been used in representingboth
succession(Arno et al. 1985)and fIfe ecologyrelationships
(Fischerand Bradley 1987).Eachcombinationof dominant
species, size-class/structure,and density by habitat type
group that can representan existing vegetationunit is identified as a potential vegetation state within SIMPPLLE.
Each potential state storesthe knowledge of what disturbance processescan occur and what the next vegetation
state would be. The collections representa sequenceof
vegetationstateswith processesbeing the agentsfor change
from one stateto anotherwithin a decadeinterval.
Processes
The processesrepresentedin this initial versionare succession,fire, mountain pine beetle in lodgepole pine and
ponderosapine, westernspruce budworm, and root fungi.
Tree regenerationis also treatedas a process.Thereare two
types of knowledge for eachprocess:the knowledgeassociated with the probability of the processoccurring,and the
knowledge associatedwith the processingspreading.The
fire processes,westernsprucebudworm,and mountainpine
beetle all may spreadfrom one unit to another.Most rating
systemsfor insect and diseaseprocessesuse very specific
stand level data (Amman et al. 1977, Carlson and Wulf
WJAF 19(2) 2004
103
104
1989,Stevenset a1. 1980).As a result, severalassumptions
and generalizationswere incorporatedinto the model logic
to work with the level of input data associatedwith the
vegetation attributes at landscapescales. Many of these
assumptionsare based on expert opinion from silviculturists,fIre managers,and ecologists.A significant assumption
associatedwith this representationof knowledgeis that only
the most dominantoutcome of a processis given. Multiple
outcomesfor the sameprocessare not represented.If more
than one outcomeis importantto represent,thena variation
of the processis created.For example,mountainpine beetle
in lodgepole pine is representedas two processes:lightmountainpine beetle and severe-mountainpine beetle.
Treatments
tationattributesandan acreagegoal to let SIMPPLLE selectunits
to treat.
SIMPPLLE System Output
The
system
tion units
system
provides
output
and the entire
provides
occurrence
the
for
processes
probabilities,
the
a process
originated
a unit.
unique
simulation
process
changes
sequence
can be examined
1). The acres
of
attribute
the acres of disturbance
landscape
by time
Different scenarios of treatment applications can be evaluated and compared without having to make changes within
the collection of potential states. Vegetation treatments can
have a combination of impacts: they can change a vegetation state; change the probabilities and types of other processes; or they may change all of these components. For
example, a thinning can change the structure class from
multistory to single story, which also changes the type of
fire process that may occur from stand-replacing to light-
can be produced
severity. Treatments can be used to change the vegetation
p~ttern that can influence probability and spread for some
processes. Treatments in this current version of SIMPPLLE
the entire
include thinning to control density of the plant community,
final harvest practices used for regenerating a new plant
community, and burning treatments used to change species
composition and structure of the community. The user interface is used to build a schedule of treatments to assign to
View.
The
attribute
and process
units
can
also
displayed
specific
potential
vegetation units or it can be used to identify
vege-
can
identifies
the events
emissions
produced
fire
suppression
for individual
value
Light westernspruce budwonn
Root disease
Root disease
Root disease
Standreplacing fire
of
in
origin,
the
the frequency
and
display
an average
from
the multiple
simulations
for each time
be
fIres,
density,
and
step summaries
for
and the high and low
(Table
4).
frequencies
indi-
in Arc-
for individual
"probability"
maps
in
ArcView.
Interpretations
being
added
habitat
(Otus
for various
as reports.
for wildlife
Flammeolus),
arcticus),
resource
Examples
values
are currently
are reports
for
acres
of
species
such as Flammulatted
black-backed
woodpeckers
and potential
old-growth
(Pi-
conditions.
Model Verification
Verification of the model is an ongoingprocess.We are
currentlyevaluatingthe ability of the systemto simulatefIre
behaviorthatis comparableto the large fIre complexesfrom
year 2000in the Bitterroot Valley in Montana.We areusing
FARSITE (Finney 1998)on samplefire eventsto verify the
"type-of-fIre" and "fIre-spread" logic within SIMPPLLE.
The ForestVegetationSimulator(Stage1973,Wykoff et al.
Table 2. Output for the entire landscape from a single simulation showing the acres of disturbance
processes by decade time steps.
Process
Time step 1
Succession
Light westernspruce budwonn
Severewesternspruce budwonn
Light lodgepolepine mountain pine beetle
Severelodgepolepine mountain pine beetle
Ponderosapine mountain pine beetle
Standreplacing fire
Mixed severity fire
Light severityfire
Root disease
350,350
3,969
5,340
975
699
946
59,119
21,330
2,844
9,178
WJAF 19(2) 2004
The
step can be mapped
as
and
the display
for each unique
size-class/size-structure,
landscape
events,
to, smoke
simulations,
3). The time
attributes
project.
of fire
it spreads
(Table
vidual
results
ArcView
process
values
Table
These
and prescribed
For multiple
includes
of each
for individual
number
the units
by wildfIres
units
unit (Table
for the entire
simulation.
a customized
to
a single
and acres
or the attributes
that lists
costs.
species,
disturbance
co ides
Processthat occurred
mapped
can be made
owls
Table 1. Display of output from a single simulation for
an individual vegetation unit showing the vegetation
state and the disturbance process for each time step.
be
Reports
from
processes
step for a single
in a report
state,
or spread
step for the landscape.
2 displays
units
processes
the
their
vegetation
a unit
for each vegetation
by time
vegetaunits,
modeled,
in
within
of each vegetation
are displayed
individual
For individual
disturbance
and whether
The
both
landscape.
Time step 2
Timestep3
387,665
1,936
709
648
444
2,193
32,000
11,066
1,756
16,333
Time
Table 3. Output from
at which each attribute
Species
value
Frequency
PP-DF "'Ii
PP
multiple simulations
occurred.
Size-class Frequency
for an individual
unit showing
the frequency
Frequency
(%)
Density
value
Frequency
(%)
Large
23
2
10
Pole
33
1
90
Medium
24
Succession
4
72
TS
20
Light severity fire
12
value
(%).
90
10
Time step
Process
Succession
vegetation
mean ac
~
Light westernspruce budwonn
Severewesternspruce budwonn
Light lodgepolepine mountain pine beetle
Severelodgepolepine mountain pine beetle
Ponderosapine mountainpine beetle
Standreplacing fire
Mixed severity fire
Light severity fire
Root disease
26,499
196
336
10
0
22
14,560
11,984
2,786
285
1982) will be used on sample plant communitiesto verify
the information in SIMPPLLE's collectionof potentialvegetation states,the time spe~tin a size class,and the resulting
next state. The verification work that has beencompleted
consists of the comparison of past change in an actual
landscape with stochastic simulations of the same landscape,the comparison ?f cycles of disturbanceprocesses
from long-term simulations with how we think the processesinteract, and the comparisonof the simulation of a
relatively small year 2000 fire with the actual event.
The Coram Experimental Forest in northwesternMontana was used as the initial data set to test the SIMPPLLE
system.Coram contains(j,800 ac of mountainousterrain on
the FlatheadNational Forest.The comparisonof the Coram
landscapewas made using timber types delineated in the
early 1930s.For model verification, 10 6-decadestochastic
simulations were made with SIMPPLLE starting with the
1930 vegetation. These simulations take into accountthe
vegetationtreatmentsthat have beenimplementedin connection with researchwork in the ExperimentalForestover
the last 6 decades.The averagefrom thesesimulationsfrom
the 1930 vegetationresulted in a simulated current landscape that has less seedlings/saplingsand more pole and
medium size classesthan found in the actual currentlandscape.The simulationsalsoresultedin more acresfor mixed
western larch (Larix occidentalis) and Douglas-fIr
(Pseudotsugamenziesii)stands than exists in the current
Process
value
~t
Mixed severity fire
Light westernspruce budworm
(%)
Time step 1 minmax ac
step 2
mean ac
Time step 2 minmax ac
20,697-48,643
0-981
0-1,681
49,913
257
134
13
41,389-53,245
31-1,043
0-50
0--0
0-77
1,006-18,376
1,790-15,141
1,159-3,888
0-1,340
0
369
3,484
851
771
886
0-636
0-51
0-0
111-850
1,017-8,248
384-2,022
374-1,278
375-1,906
inventory.The differencein the speciesis attributableto the
difference in the way the inventorieswere assigneda cover
type. The 1930svegetationmapsdelineatedmixed-species
communities. Mixed communities of western larch and
Douglas-fir were common, while communities dominated
by single specieswererare. The currentdelineationis based
on the summarizationof plot datathatdeterminesa plurality
of basal area by species.The differencein the smallersize
classes,seedling/saplings,pole, and medium was determined to be the resultof SIMPPLLE moving the vegetation
units throughthesesize classestoo fast. The initial time in
these size classescomes from a summarizationof Forest
VegetationSimulatorruns on samplestandsthat represent
these forest types by habitat type groups for the entire
Northern Region. The observedgrowth rates from unpublished datacollected in spacingstudiesfor Coram provided
a betterbasis for the time to move throughthe smallersize
classesfor this part of the region.
Fire is the only disturbanceprocess for which records
have beenkept at Coram.Table 5 comparesthe averagefIfe
attributes from the original ten simulations with the fIfe
attributes that actually occurred over the last six decades.
Threeattributesare compared,the numberof fIfe events,the
percentageof fIfe eventssuppressedatlessthan 0.25 ac,and
the total numberof acresburned. Using the past10-yearfIfe
occurrence for the Flathead Forest as the basis for the
probability of a fIfe event resulted in significantly greater
Table 5. Comparison of three fire attributes from the actual occurrence at Coram Experimental
Forest, averages from the original ten simulations, and averages from revised ten simulations.
Actual fires
Original simulation
Revised simulation
WJAF 19(2) 2004
105
.,
number of simulated fIre events. The percentageof these
events that were suppressedat less than 0.25 ac was much
lower than actually achieved.The simulatedburned acres
were gl::eaterthan those actually burned. Changeswere
madeto provide the user flexibility to basethe probability
of fIre events on a more localized area,to be able to adjust
fIre suppressionlogic for differencesin landownershipand
road status,and to provide accessto adjustthe type-of-fire
and fIre-spreadlogic. The resultsof using thesechangesare
shownin the revised-simulationsrow in Table 5. Thereis an
improvement in the numberof fire events and the percent
suppressedat less than 0.25 ac. However,the acresburned
are still abovethe actual.Additional adjustmentsto the fire
logic could get the simulation values closerto the actual.
The application of the systemto numerousother areas
within the Northern Region have provided the opportunity
to continue model verification and fine tuning of its performance. Long-term simulations without fIre suppressionfor
400 years on a 1.5 million-ac area provide the basis for
examining how simulatedcycles for processescompareto
our expectedrelationships betweendisturbanceprocesses.
Figure 1 displaysthe resulting cycles of stand-replacingfire,
mixed-severityfire, and mountainpine beetle in lodgepole
pine from one long-term simulation. Significant levels of
mountainpine beetle activity tend to occur only after periods of minimal fire activity that allow time for lodgepole
pine to mature and becomesusceptible.When a significant
level of mountainpine beetle activity occurs creatingfuels,
it is followed by an increasein the level of fire. Cycles of
mixed-severityfITe are more frequent than the cycles of
stand-replacingfITeand throughtheir reductionof densityin
lodgepolepine standskeep mountainpine beetle activity at
a minimum.
The fITesof 2000 in the Bitterroot Valley of Montana
provide anotheropportunityto verify SIMPPLLE's performance.A comparisonof one of the smallerfITesof 11,475
ac was made with a simulation starting a fITe event in the
sameplant community in which the fITeoccurred.Figure 2
displays the actualfITe and the simulatedfITe.Differences
betweenthe two were the result of inaccuratemapping of
what was typed as nonforest,rock, and assumptionsconcerning fITe suppression.Areas that were identified for
SIMPPLLE as rock were actuallylow-densityforestedareas
that did supporta fire. Areasalongthe wildernessboundary
that SIMPPLLE's suppressionlogic indicated would be
suppressedwere not, but continuedto spread.Both of these
areas need additional emphasis.The need for improved
typing of the vegetationusedfor SIMPPLLEand an expansion of the fire suppressionlogic will be addressedin the
additional work being done on the large fire complexes
from the year 2000. However,the current performanceof
modeling the fire processis consideredadequate.Using
multiple simulations over decadesto create probability
maps for processesand vegetationattributes can involve
700,000
600,000
,
,
.,
500,000
.
.
.
.
,
400,000
~
~
«
300,000
200,000
.
.
.
.
"
"
,
.
.
.
.
"
"
, '
.'
, .
.
.,
,
0
.'
,
,
,
,
,
,
,
,
,
,
'
'
'
.
.
'
'
'
'
1\'
,
",
'
~,
,\,
~
,
1 2
3
..
....
..
~I'"
4
5 6
7
8
9 1011 12131415161718192021
I
~
,
"
.,
,.
.,
"
,.
.,
,.
, ,
.,
.,
...,
"
"
'.
'.
"
'.
"
, ,
..
.'.
.."
,
.
.
.
100 ,000
.
,
,
,"
"
..
,
,.
...,
.
."
, ..,
, "
, ,.
,
,
.
A~'
.
/3~
, , .
, ,
, ,
, ,
, ,
, ,
, ,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
I
..
.'
..
222324252627282930313233343536373839
40
Decade
Figure 1. Cycles of mixed severity fire, stand-replacing fire, and mountain pine beetle from multiple simulations of a historic
representation of a 1,O86,OOO-ac
landscape on the Beaverhead-Deerlodge National Forests.
106
WJAF 19(2) 2004
Figure 2.
Comparison of the actual Blodgett Trail Head fire
with the simulated fire.
hundredsof fire events.The degreeof accuracyconsidered
appropriatefor landscapeplanningis not the samerequired
for planning fire suppressionactivities on an ongoing fire.
Model Application
The fIrst version of SIMPPLLE was delivered to the
Northern Regionin January1997.To accountfor variability
within the habitattype groupsin the region,SIMPPLLE has
beenstructuredto provide a Westsideand Eastsideoption.
The system has been applied to a number of landscapes
within Montana and Idaho for the Northern Regionand the
Bureau of Land Managementat scalesfrom 26,800 to 1.8
million ac. Its use has ranged from project planning to
landscapeassessment,
and analysis of the managementsituation prior to forest plan revision. Each level of analysis
involves different uses of the system.The comparisonof
mountainpine beetle activity in lodgepolepine for a number
of alternativesin a landscapeon the HelenaNational Forest
(Figure 3) is typical of its use at the projectplanning level.
The potential for providing desired vegetationconditions
can be evaluated spatially over time. The change in vegetation attributes as a result of both treatmentsand disturbanceprocesseswere mappedby decadesfor a number of
Figure 3. Acres of mountain pine beetle activity in lodgepole
pine for alternatives on the Poorman Landscape, Helena National Forest. All levels are the average of five simulations.
Alternatives include both treatments and fire suppression. The
"no-suppression"
level does not include fire suppression or
treatments. The "with-suppression"
level includes fire suppression but no treatments.
managementalternativesfor a landscapeon the Kootenai
National Forest. The vegetation attributes selected were
thoseusedto identify potential old-growth conditions.
Additional versions of SIMPPLLE are currently being
developedf9r use in a study that comparesvarious models
for evaluatingfuel treatmentsat landscapescales(Weise et
al. 2000). Within this application,SIMPPLLE is used with
the MAGIS optimization and schedulingmodel(Zurring et
al. 1995) to quantify risks from disturbanceand schedule
fuel treatmentsat landscapescales(Jonesand Chew 1999).
Versions of SIMPPLLE are being developedfor Yosemite
National ParkandAngelesNational Forestin California,the
Kenai Peninsula in Alaska, Gila National Forest in New
Mexico, Conecuh National Forest in Alabama, HuronManisteeNational Forestin Michigan, and the Blackwater
StateForest/EglinAir ForceBasein Florida. The mixture of
ownershipsin theseareasdisplaysthe ability to use SIMPPLLE for lands other than National Forests.
Discussion
SIMPPLLE provides a modeling tool for managersto
integrateand interpretconceptssuchas desiredfuture conditions, range of variability, and the interaction between
vegetationpatternsand disturbanceprocesses.SIMPPLLE
offers an environmentin which the knowledge developed
by scientistsand managerscan be integratedinto the quantification of potential vegetative conditions, disturbance
processprobabilities,and the logic for the interaction between processesand vegetationpatterns.SIMPPLLE provides a way to help evaluateproposedmanagementscenarios within a future that includes stochasticprocesses.Proposedschedulesof managementactivities may not be possible whenthe likely occurrencesof numerousdisturbance
processesare considered. Without the consideration of
likely disturbanceprocesses,effects of no action alternatives are often underestimated.
Stochasticsimulations with SIMPPLLE can help in designing managementstrategiesby quantifying what processesmay have a higher occurrenceon the landscape,or
what parts of the landscapeare more prone to disturbance
processes.Doesoneusea managementstrategythat focuses
action in those areas that have the highest likelihood of
severedisturbanceevents with the intent of reducing the
disturbanceevents?Doesoneusea strategyof treatmentsto
createa vegetationpatternthat reducesprocessspread?Or
doesone use a strategyof putting investmentsin management actions on those parts of the landscapethat have a
lower likelihood of significant change to minimize the
chanceof losing investments?
In the initial versionsof SIMPPLLE, the emphasisis on
the ability of the systemdesignto captureour knowledgeof
vegetationchange and the interaction between vegetation
patternsanddisturbanceprocessesat differentspatialscales.
Future work will place an emphasis on improving the
knowledge within the system.The initial validation work
with Coram ExperimentalForestin NW Montanaindicates
that the approachusedfor capturingand integratingprocess
WJAF 19(2) 2004
107
knowledge in SIMPPLLE does predict realistic results at
landscapescales.
Complete documentationof the systemis under developmentas a generaltechnical report, and additional documentationand examplesof use canbe found on the website
www.fs.fed.us/rm/missoula/4151/SIMPPLLE.
Literature Cited
ARNo, S.F., D.G. SIMMERMAN,AND R.E. KEANE. 1985. Forest succession
on four habitat types in western Montana. USDA For. Serv. Gen. Tech.
Rep. 1NT-177. 74 p.
AMMAN, G.D., M.D. MCGREGOR,D.B. CAHILL, AND W.H. KLEIN. 1977.
Guidelines for reducing losses of lodgepole pine to the mountain pine
beetle in unmanaged stands in the Rocky Mountains. USDA For. Servo
Gen. Tech. Rep. 1NT-36. 19 p.
BAKER,W.L. 1992. Effects of settlement and fife suppression on landscape
structure. Ecology 73(5):1879-1887.
CARLSON,C.E. AND N.W. WULF. 1989. Silvicultural strategies to reduce
stand and forest susceptibility to the western spruce budworm. USDA
For. Servo Agric. Handb. No. 676. Washington, DC. 31 p.
FINNEY, M.A. 1998. FARSITE: Fire area simulator-model development
and evaluation. USDA For. Servo Res. Pap. RMRS-RP-4. 47 p.
FIsCHER,W.C., ANDA.F. BRADLEY,1987. Fire ecology of western Montana
forest habitat types. USDA For. Servo Gen. Tech. Rep. 1NT-223. 95 p.
FORMAN,R.T.T., AND M. GoDRON. 1986. Landscape ecology. John Wiley
& Sons, New York, NY. 620 p.
JONES,J.G., AND J.D. CHEw. 1999. Applying simulation and optimization
to evaluate the effectiveness of fuel treatments for different fuel
conditions at landscapes scales. P. 89-96 in Proc. from the joint fire
science conference and workshop, Vol. II, Neuenschwander, L.F., and
K.C. Ryan (eds.). Univ. of Idaho, Moscow, ID
108
WJAF 19(2) 2004
KESSELL,S.R., ANDM.W. FOrrER. 1980. A quantitative succession model
for nine Montana forest communities. Environ. Manage. 4(3):227-246.
PFISTER,R.D., B.L. KOLALCHIK,S.F. ARNo, ANDR.C. PREsBY.1977. Forest
habitat types of Montana. USDA For. Servo Gen. Tech. Rep.
GTR-INT-34. 174 p.
PICKETT,S.T.A., AND M.J. McDoNNElL. 1989. Changing perspectives of
community dynamics: A theory of successional forces. Trees.
4(8):241-245.
REYNOLDS,K.M., AND E.H. HOLSTEN.1994. Relative importance of risk
factors for spruce beetle outbreaks. Can. J. For. Res. 24:2089-2095.
STAGE,A.R. 1973. Prognosis model for stand development. USDA For.
Servo Gen. Tech. Rep. INT-137. 32 p.
STEVENS,R.E., W.F. MCCAMBRIDGE,AND C.B. EDMINSTER.1980. Risk
rating guide for mountain pine beetle in Black Hills ponderosa pine.
USDA For. Servo Res. Note RM-385. 2 p.
TuRNER,M.G. 1989. Landscape ecology: The effect of pattern on process.
Annu. Rev. Ecol. Syst. 20:171-97.
USDA FOREST SERVICE. 1987. FSH 2409.17 Silvicultural practices
handbook, supplement No.6, regional stocking guides. Northern
Regional Office, Missoula, MT. 78 p.
WEISE, D.R., R. KiMBERLIN, M. ARBAUGH, J. CHEW, G. JONES, J.
MERGENIAN, M. WIITALA, R. KEANE, M. SCHAAF, AND J. VAN
WAGTENDONK.2000. A risk-based comparison of potential fuel
treatment trade-off models. P. 96-102 in Proc. from the joint fire
science conference and workshop, Vol. 11,Neuenschwander, L.F., and
K.C. Ryan (eds). Univ. of Idaho, Moscow, ID.
WYKOFF,W.R., N.L. CROOKSTON,
ANDA.R. STAGE. 1982. User's guide to
the stand prognosis model. USDA For. Serv. Gen. Tech. Rep. INT -133.
113 p.
ZUURING,H.R., W.L. WOOD, ANDJ.G. JONES.1995. Overview of MAGIS:
A mu1ti-resourceanalysis and geographic information system. USDA
For. Servo Res. Note INT-RN-427. 6 p.