A dynamic object-oriented architecture approach to ecosystem modeling and simulation

Abstract

ing on its behalf, a paid-up, nonexclusive, irrevocable worldwide license in said article to reproduce, prepare derivative works, dis-tribute copies to the public, and perform pub-Iicly and display publicly, by or on behalf of the Government.

Full text

,,
*-
ADYNAMIC OBJECT-ORIENTED ARCHITECTURE
APPROACH TO ECOSYSTEM MODELING AND SIMULATION*
Prepared by
Pamela J. Sydelko, J.E. Dolph, and T.N. Taxon
Argonne National Laboratory
Argonne, IL
and
K.A. Majerus
U.S. Army Construction Engineering Research Laboratory
Champaign, IL
for
American Society for Photogrammetry and Remote Sensing (ASPRS)
1999 Annual Conference
May 17-21 <’-,
Portland, Oregon
The submitted manuscript has been created
by the University of Chicago as Operator of
Argonne National Laborato~ (“Argonne”)
under Contract No. W-31-I09-ENG-38 with
the U.S. Department of Energy. The U.S.
Government retains for itself, and others act-
ing on its behalf, apaid-up, nonexclusive,
irrevocable worldwide license in said article
to reproduce, prepare derivative works, dis-
tribute copies to the public, and perform pub-
Iicly and display publicly, by or on behalf of
the Government.
*Work supported by the U.S. Department of Defense, U.S. Department of Energy,
and the U.S. Environmental Protection Agency, under contract W-31-109-Eng-38.

DISCLAIMER
This report was prepared as an account of work sponsored
by an agency of the United States Government. Neither the
United States Government nor any agency thereof, nor any
of their employees, make any warranty, express or implied,
or assumes any legal liability or responsibility for the
accuracy, completeness, or usefulness of any information,
apparatus, product, or process disclosed, or represents that
its use would not infringe privately owned rights. Reference
herein to any specific commercial product, process, or
service by trade name, trademark, manufacturer, or
otherwise does not necessarily constitute or imply its
endorsement, recommendation, or favoring by the United
States Government or any agency thereof. The views and
opinions of authors expressed herein do not necessarily
state or reflect those of the United States Government or
any agency thereof.
,,,, -, .,,.,. . .. .. . .,,..... ,(. ?==-.?7.. ., .,,, -.. -. ...-. .—

DISCLAIMER
Portions of this document may be illegible
in electronic image products. images are
produced from the best available original
document.
-,7 . .
—

ADYNAMIC OBJECT-ORIENTED ARCHITECTURE
APPROACH TO ECOSYSTEM MODELING AND SIMULATION
Pamela J. Sydelko
Environmental Scientist
Argonne National Laboratory
9700 S. Cass Avenue, Bldg 900
Argonne, I] 60439
svdelko@dis.anl. ~ov
Kimberly A. Majerus
Forester
U.S. Army Construction Engineering Research Laboratories
P.O. Box 9005
Champaign, IL 61826
k-maierus @cecer.armv.mil
Jayne E. Dolph
Environmental Computer Scientist
Argonne National Laboratory
9700 S. Cass Avenue, Bldg 900
Argonne, 1160439
dol~h@dis.anl.sov
Thomas N. Taxon
Software Engineer
Argonne National Laboratory
9700 S. Cass Avenue, Bldg 900
Argonne, 1160439
taxon@dis.anl.~ov
ABSTRACT
Modeling and simulation in support of adaptive ecosystem management can be better accomplished
through adynamic, integrated, and flexible approach that incorporates scientific and technological
components into acomprehensive ecosystem-modeling framework. The Integrated Dynamic Landscape
Analysis and Modeling System (IDLAMS) integrates ecological models and decision support techniques,
through ageographic information system (GIS)-based framework. The Strategic Environmental Research
and Development Program (SERDP) sponsored the development of IDLAMS. Initially built upon aGIS
framework, IDLAMS is migrating to an object-oriented (00) architectural framework. An object-oriented
architecture is more flexible and modular. It allows disparate applications and dynamic models to be
integrated in amanner that minimizes (or eliminates) the need to rework or recreate the system as new
models are added to the suite. In addition, an object-oriented design makes it easier to provide run-time
feedback among models, thereby making it amore dynamic tool for exploring and providing insight into
the interactions among ecosystem processes. Finally, an object-oriented design encourages the reuse of
existing technology because 00-IDLAMS is able to integrate disparate models, databases, or applications
executed in their native languages. Reuse is also accomplished through astructured approach to building a
consistent and reusable object library. This reusability can substantially reduce the time and effort needed
to develop future integrated ecosystem simulations.
—.

BACKGROUND
Military land managers and decision makers are recognizing the need for more strategic and adaptive
approaches to integrated natural resources and ecosystem management. Decision makers face an ever-
increasing challenge to balance maximum flexibility for the mission with amultitude of other land use,
social, political, and economic goals. In addition, these goals encompass environmental requirements for
maintaining ecosystem health and sustainability over the long term. To meet these challenges, simulation
models, geographic information systems (GISS), remote sensing applications, scientific visualization, and
decision analysis techniques are fast becoming tools of the trade. However, when used individually, these
tools are often limited because they evaluate potential impacts for only one ecosystem characteristic at a
time, while holding the remainder of the system static. Effective ecosystem management relies upon
assessing several components of the system simultaneously, in terms of their relationship to each other and
the way that they are affected by land use and land management decisions. The premise of our research is
that integrated natural resource and ecosystem-based management can be better accomplished through the
use of amore dynamic, integrated, and flexible approach that incorporates the appropriate scientific and
technological components into acomprehensive ecosystem-modeling and technology framework.
Other integrated modeling systems have been developed or are being developed for use in land use
planning and natural resource management (Berry, et al., 1996; Maxwell and Costanza, 1995; Bennett, et
al., 1996; Frysinger, et al., 1996; and Fedra, 1996).
IDLAMS
In 1994, the Strategic Environmental Research and Development Program (SERDP) sponsored the
development of the Integrated Dynamic Landscape Analysis and Modeling System (IDLAMS) to address
the military’s need for amodel-integration framework (Li, et al., 1998). Built upon an effort funded by
Fort Riley, Kansas, IDLAMS was developed as amultiagency partnership between the U.S. Department of
Energy’s (DOE’s) Argonne National Laboratory (ANL) and the U.S. Army’s Construction Engineering
Research Laboratories (USACERL). IDLAMS is aprototype computer decision support system that
integrates ecological models and decision support techniques and is based upon aGIS framework.
IDLAMS predicts land conditions (e.g., vegetation, wildlife habitats, and erosion status) by simulating
changes in military land ecosystems for given training intensities and land management practices. Military
land managers can use IDLAMS to help predict the future ecological condition for agiven land use on the
basis of various land management scenarios and levels of training intensity. Land managers can also use
IDLAMS as atool to compare different land management practices and to further determine aset of land
management activities and prescriptions that best suits the needs of aspecific military installation.
In the initial GIS-based IDLAMS, four major models were developed and integrated; avegetation
dynamics model, aset of wildlife habitat suitability models, an erosion model, and ascenario evaluation
module. Brief descriptions of these components and models follow.
Vegetation Dynamics Model: The Vegetation Dynamics Model is the core model for
IDLAMS because the output from this model is the input for all other connected IDLAMS
models. The Vegetation Dynamics model is aspatially explicit, model that incorporates
vegetation changes due to (1) natural succession, (2) land use impacts, and (3) land
management actions.
Wildlife Models: These six submodels represent individual wildlife species and are based on
U.S. Fish and Wildlife Service Habitat Suitability Indices (HSI). Each submodel requires the
user to input either avegetatiordland cover map representing the current condition or a
simulated landcover map generated by the Vegetation Dynamics Model. In some submodels,
additional input maps may be required.
-—--—,- ... ,--, .e,.- .. .. . .. ... . . .. .. . ,.. ,.. .,?.

Erosion. Model: IDLAMS currently integrates the Revised Universal Soil Loss Equation
(RUSLE) to generate an erosion status map for each current condition or simulated
vegetatiordland cover map input by the user. RUSLE also requires other spatial data
representing various factors affecting erosion.
Scenario Evaluation Module: IDLAMS uses avalue-based decision-analysis process to link
the ecological models with the management needs and user requirements of the resource
manager. This component of IDLAMS requires (1) developing operational definitions for
multiple land use and management objectives, (2) determining and prioritizing alist of
relevant objectives to be used, and (3) defining measures to estimate the extent to which those
objectives have been reached. This information is then used to perform trade-off analysis for
land management alternatives on the basis of the results from the spatially explicit modeling
and to rank the alternatives according to how well they meet the specified objectives.
In 1998, SERDP sponsored an IDLAMS workshop to discuss lessons learned from the IDLAMS
development and to identify fhture research priorities. Workshop participants represented the military user
community, Army research laboratories, and other SERDP model developers. Some of the workshop’s
resultant goals for future research and enhancements to IDLAMS include the following:
●
●
●
●
More flexible and adaptive mechanisms for integrating disparate existing software
applications;
Improved ability to reflect the dynamics of living ecosystems, land uses, and land management
practices;
Capability to support software applications that can operate at multiple spatial and temporal
scales; and
Reduction in the long-term cost of modeling technology by use and reuse of existing data,
models, and system components.
00-IDLAMS .
To accomplish these goals, anew SERDP-funded effort has been undertaken by ANL and USACERL
to evaluate aprototype of anext-generation integration architecture using an object-oriented approach.
This prototype is called Object-Oriented (00)-IDLAMS. The flexibility, dynamics; and interoperability
gained through an object-oriented approach have the potential to provide key technology solutions for
addressing many of the military’s goals and needs for integrated natural resource planning and ecosystem
management.
00-IDLAMS is built upon an object-oriented architecture called the Dynamic Information
Architecture System (DIAS) (Figure 1). This generic architecture was designed and developed at ANL.
DIAS supports distributed, dynamic representation of interlinked environmental processes and behaviors at
variable scales (spatial and temporal) of resolution and aggregation. The generic DIAS development effort
has been sponsored by several organizations and partners so that all users serve to benefit from resources
leveraged through cooperation and DIAS contributions.
The modeling domain of 00-IDLAMS is flexible and is determined by available objects and by the
suite of models and other data processing applications that have been “registered” within 00-IDLAMS to
address specific information processing needs. 00-IDLAMS also employs an object-oriented GeoViewer
module that is designed to navigate within an 00-IDLAMS study arerdframe to create, query, view, and
manipulate objects. The key elements of the 00-IDLAMS architecture are described below.
--
,---: q,:
,. .,,.:.
-%
~
,,. ... , ,. .. , ,>.,.. ....... ., ., :. ;,..+.-..-.,. .. ,.
.(.. ,4. . ,. “,... ... .=.. —7. -_-—_—

,.... ....
,. ..,..,.
“wrmmed- dilizii2
F—————7 .-a%‘“
jEitomal
\..s$!urcac
u
Discrate
Event
!M&ath:
ewsfhtiiy
Ck+ectsto
F-&fcmo~
Figure 1. 00-IDLAMS Conceptual (based upon the Dynamic
Information Architecture System (DIAS))
Key Design Elements of the 00-IDLAMS Architecture
Object-oriented design elements offer the key to flexibility and dynamics in bringing together disparate
data and computer applications for integrated resource management to support the military mission over the
long term.
Entity Objecc Adomain (discipline-specific) object (e.g., grassland, watershed, stream, atmospheric
layer, etc.) encapsulating state/properties and many possible aspects of dynamic object behavior.
Entity objects carry both state variables and behavior:
State variable: Properties of an Entity object and the current value of these properties
Behavion How an Entity object acts and reacts in terms of state changes to an external event.
Analysis Frame: The object representing the area of interest in asimulation. In many cases, the
analysis frame is ageographic boundary of an area.
Object Library: Apermanent collection of Entity objects. These Entity objects are brought into the
analysis frame for use during simulations. When new applications are registered into 00-IDLAMS,
new objects may need to be added or existing library objects may need to be edited. In this way, the
object library will grow, and the potential for reuse increases over time.
Data Import/Export Utilities: Amechanism that supplies Entity objects with the current values for their
state variables. These data-ingestion utilities have been developed to supply the object state variables
from avariety of external data sources.
Process Object: An object providing the means of addressing one or more particular Entity object
behaviors. This object includes aformal description of the software application that represents the
behavior. For the Grassland Entity object example, Process objects can be created to connect the
Entity object behaviors with the software application that can address them. The Process object must
also include any code necessary to deal with how the time scale should be aggregated or disaggregate
on the basis of the context of the simulation (see discussion of Context object).
Discrete Event Simulation Manage~ The Discrete Event Simulation Manager processes events in a
time-ordered queue. Events are created principally by Entity behavior and by user interactions.

Events have atime stamp and may have data associated with them. They are selectively broadcast to
objects that have registered interest in certain Event types at simulation initialization.
GeoViewer: Amodule that provides object-oriented GIS functionality to navigate within an Analysis
Frame and to create, query, view, and manipulate objects within that frame.
Context Objecc The object that specifically lays out the context of the simulation. This object contains
amethod that links up the behavior of Entity objects in the Analysis Frame with their associated
Process objects and specifies any dependencies between them. These dependencies include any code
that needs to be implemented to deal with any discrepancies in the time scale. For example, if two
models in the current suite operate on different time steps, he Context object would include apointer
to the algorithm (coded in the Process object) that aggregates or disaggregate the time scales.
Model Controller Objecc The Model Controller object provides an explicit linkage to external
application source codes and data structures. It is the “wrapper” used to communicate between the
Process object and an external model (or process/subroutine of that model).
Course-of-Action (COA): ACOA object is essentially aflowchart of individual steps constituting a
specific plan or action.’ COAS have many uses and can be used to model procedural or sequential
processes. COAS, like Process objects, are used to represent specific behaviors of Entity, objects.
Spatial Data SeC ASpatial Data Set is an object that carries acomplete geometric specification for a
two-dimensional or three dimensional (2-D or 3-D) spatial partitioning scheme that divides aregion
into 2-D or 3-D cells. It also contains acollection of data elements (attributes) for each cell. The
Spatial Data Set is tied to aProjection object (see below) if aggregation or disaggregation of spatial
scale is needed for the context of the simulation.
Projection Objecc This object is associated with aspatial data set object and is responsible for
implementing any algorithm necessary to deal with multiple scales. The domain expert should be
responsible for providing the correct mechanisms for implementing this disaggregation or aggregation
within the context of the simulation.
Registering External Applications
Fortunately, awealth of existing (legacy) applications are available. These applications have been
developed and have often been validated by scientists with extensive knowledge of their specific subject
domains. It is important to allow for the reuse of these models/applications without expensive reworking or
recoding. Application registration through 00-IDLAMS provides for easy integration of legacy-type
applications (e.g., simulation models, GIS applications, database management systems). Once an
application has been registered with 00-IDLAMS, it can interact, as appropriate, with 00-IDLAMS Entity
objects within asimulation. Applications used within 00-IDLAMS do not directly communicate among
themselves but interact indirectly with one another via Entity objects. This interaction provides for amore
stable environment because the disruption to the suite of models is limited when amodel is added or
removed. Afurther advantage is that 00-IDLAMS external applications are executed in their native
languages (e.g., FORTRAN, C, etc.). 00-IDLAMS can even be used to integrate disparate models,
databases, or applications that can reside on multiple machines across the Internet. Model registration
within 00-IDLAMS is aformalized process that allows model developers to register their models with
00-IDLAMS while retaining full control of their source codes. The model registration procedure consists
of the following five steps:
1. Formally defining the application in terms of
-Attributes/states of domain objects
-Processes represented as behaviors of the domain objects
-Input/output parameters to the processes,
2. Instrumenting amodel source code (flag/delineate individual processes),
3. Writing amodel controller (or wrapper) for the model,

4. Building (or editing existing) domain Entity objects, and
5. Building Process objects (one for each process delineated in the source code).
00-IDLAMS Prototype
In the original IDLAMS framework (Li, et al., 1998), integration is accomplished through heavy
reliance on astatic GIS structure, which limits flexibility and the potential for truly dynamic interaction
among the models being integrated. To overcome some of these limitations, ANL and USACERL have
undertaken amove toward an object-oriented approach (00-IDLAMS). An 00-IDLAMS prototype is
being developed to demonstrate the key benefits of an object-oriented approach. The study area for the
prototype will be Fort Riley, Kansas, which is the same study area used in the original IDLAMS
application.
The 00-IDLAMS prototype integrates asubset of the original IDLAMS, including the Vegetation
Dynamics Model and the Henslow’s Sparrow Habitat Model (reimplemented as an Environmental Systems
Research Institute (ESRI@)application). For the 00-IDLAMS implementation, the Military Training and
Land Management components of the original Vegetation Dynamics Model were broken out into three
COA objects within 00-IDLAMS (Training, Burning, and Planting). These COAS represent the Fort Riley
land use and land management plans. These plans are inherently procedural in nature and readily lend
themselves to COA implementation. COAS are considered to be models within 00-IDLAMS. The natural
succession processes remain pa-t of the external Vegetation Dynamics Model and are registered with 00-
IDLAMS as the external Vegetation Dynamics Model. Conceptually, the natural functions of vegetation
dynamics will not need to be frequently changed over time and should remain as part of the external model.
On the other hand, the aspects of land use and land management may change frequently over time; this
should not pose any difficulties because COAS can be easily edited.
Figure 2illustrates avery important object-oriented design principle. The models do not talk directly
to one another but are only integrated through their relationships with the Entity objects. This makes it
much easier to replace amodel or integrate additional models without disrupting any other models in the
suite. This modular design principle makes 00-IDLAMS much more flexible than the earlier GIS-based
version.
Fl~ure 2. 00-IDLAMS Prototype Conceptual Design
.- ,—- .. . . . ... .-. —-..—-----. .

Four Entity objects are needed for the currently designed implementation of 00-IDLAMS: Landcover,
Land Management Plan, Land Use Plan, and Henslow’s Sparrow. These objects include state variables and
behavior needed for the models in the 00-IDLAMS suite. More Entity objects can be added if additional
models or applications are added to the 00-IDLAMS suite. Some model input and output parameters are
not held as Entity state variables. Only input or output that will be needed by more than one model must be
state variables. If, however, other models were to be registered that require this input or produce these
variables as output, Entity objects could be easily edited to include them.
This ability to utilize existing objects illustrates the object-oriented design principle of reusability. As
the object library for 00-IDLAMS grows, the potential for reuse grows. As new models are added, it may
be as easy as pulling an existing Entity object from the library and simply editing it to include new
variables and behavior specific to the new model. Another benefit of object-oriented design is the
improved ability to reflect the dynamics among living ecosystems, land uses, and land management
practices, because the current state of these objects is always used by any of the models in the suite.
For the purposes of this paper, details about the individual models are not discussed. Rather, only
model input and output parameters that directly correspond to Entity state variables or behavior are
discussed.
The Landcover Entity object holds six state variables and three behaviors for use in the 00-IDLAMS
suite of models (13gure 3).
Landcover Object
State variables
Landcovermap
Valleysmap
Successionaltime-stepmap
Areasimpactedby trainingmap
Burnedareasmap
Plantedareasmap
Behavior
Implementforest spread process
Implementsuccessionprocess
Implementfire regimeprocess
Figure 3. Landcover Object State Variables and Behaviors
State Variables:
1. Lardcover map -Spatial Data Set object that holds the attributes for landcover categories.
This state variable is used as an input to all the models and is updated by the “update
landcover map” behavior invoked by several models in the following ways:
.The Vegetation Dynamics Model updates this state variable in several ways. It
can change prairie to forest by the forest spread process. The succession process
can change vegetation categories by allowing them to change from one category
to another. The fire regime process controls whether prescribed burning will
change aforested area to prairie.
●The Training COA can update this state variable by setting vegetation categories
to damaged due to training impact.
.-—zr—r—— .-——r-s ,.>.....,. ,,, , . . . . .. . . . . . . .. ...’ . —.—-——

.The Planting COA can update this state variable by changing damaged areas to
early or late successional grasslands, depending on what is planted (erosion
control or native species).
.The Burning COA can update the Iandcover map state variable because burning
changes forest to late successional prairie. In addition, because burning also
accelerates the successional time-step for an area (see below), it can change the
Iandcover state if the time-step of an area is almost ready to change to the next
category.
2. Valleys map –Spatial Data Set object that defines where valleys exist. This variable is used
in the forest spread process of the Vegetation Dynamics Model and the Burning COA.
3. Successional time-step map –Spatial Data Set object that holds the successional time-step
attribute. Succession within the Vegetation Dynamics Model proceeds from damaged
gmslands, to early successional grasslands, to late successional grasslands, to forest. The
number of years spent in each stage is aVegetation Dynamic Model input parameter. For
example, for the Fort Riley implementation, the time needed to change from early to late
successional may be 15 years. Because each cell in the spatial data set has an initial
successional time-step assigned to it, if an early successional cell has asuccessional time-step
of 14 years, it will change the next year into late successional. Processes within the
Vegetation Dynamics model as well as all three COAS can update the successional time-step
map state variable either by moving it forward or setting it back.
4. Areas impacted by training map –Spatial Data Set object that holds the output from the
Training COA (see below). These areas are used to determine where to plant in the Planting
COA.
5. Burned areas map –Spatial Data Set object of burned areas. This map is an output of the
Burning COA. The burned areas map is used by the Henslow’s Sparrow object, because
recently burned areas are considered poor habitat in the Henslow’s Sparrow habitat model. In
addition, the burned areas map for each year can be saved for viewing after the simulations.
It can also be used to create animated visualizations that show the burning regime over time.
6. Planted areas map -Spatial Data Set object of planted areas. This map is an output from the
Planting COA. Currently, the planted areas map is not used directly by another model.
However, because the history of areas planted maybe of interest to land managers, it is saved
as astate variable for visualization purposes. As for the burned areas map, the planted areas
map can also be viewed and used in animations.
1. “Implement Forest Spread Process” –This behavior will implement the forest spread process
of the Vegetation Dynamics model. The three processes of the Vegetation Dynamics model
have been registered as separate processes in 00-IDLAMS. This demonstrates the
modularity of the system. If at some time, the Vegetation Model needs to be updated with a
new forest spread algorithm, but all ot’her processes of the model remain valid, only this
process will need to be changed to call another algorithm.
2. “Implement Succession Process” –This behavior implements the succession process of the
Vegetation Dynamics model.
3. “Implement Fire Regime” –The fire regime process of the Vegetation Dynamics Model is
invoked by this behavior.
-r----- . . .. ,. -+’77-7?-7-7-7W. ..>,. , .,, .,-, .- .— ..- .—.
. .,,.

The Land Management Plan Entity object holds one state variable and two behaviors (Figure 4).
l!!!!
..
.
Land Management Plan Object
State variables
Landmanagementobjectives
Behavior
ImplementBurningCOA
-Decidewhereto bum
-Distributeburnedacres
-Outputupdatedkmdcover
-Outputupdatedsuccessionaltime-step
ImplementPlantingCOA
-Decidewhatto plant
-Decidewhereto plant
-Distributeplantedacres
-OutputupdatedIandcover
-Outputupdatedsuccessionaltime-step
Figure 4. Land Management Plan Object State Variables and Behaviors
State Variables:
Land management objectives -This state variable is acollection of all land management
objectives the land user cares about and aranking of their importance. Currently the only
objective available in this prototype is managing the Henslow’s Sparrow. As new models are
added to the suite, it may include management of other species or management for erosion control.
Activating or deactivating these objectives can cause different COAS to be invoked.
Behavio~
1. “Implement Burning COA” -This behavior simply initiates the Burning COA. The Burning
COA acts as aproxy for the land manager’s burning plan. The Burning COA will decide
where to burn (based on abum rotation map and hectares to be burned) and will bum it, and it
will update the landcover map and successional time-step map (held as state variables of the
Landcover object).
2. “Implement Planting COA” –This behavior initiates the Planting COA. This COA acts as a
proxy for the land manager’s planting plan. It is slightly more complex than the Burning
COA because the Henslow’s Sparrow is amanagement objective (if selected). Once the COA
has established what to plant (erosion control or native species), it then decides whereto plant
it. If native species are to be planted, the Planting COA will “ask” the Henslow’s Sprmow
object what the current state of the patches and aaiiitional hectares needed state variable is.
This will invoke aprocess of the Henslow’s Sparrow Habitat Model that creates patches of
excellent habitat. To be considered excellent Henslow’s Sparrow habitat, patches need to be
at least 65 hectares in size. Planting native species next to apatch and therefore fi.dfilling the
patch size requirement could greatly benefit the species. The planting COA will perform an
analysis that will check for adjacent damaged grassland areas that can be planted to fulfill the
65-hectare requirement. If native species can be planted in such away as to fulfill the
requirement, the COA will plant to fulfill the 65-hectare patch size requirement. Viewing the
planted areas map after the simulation can be very helpful to aland manager interested in
knowing where to plant native species in order to manage the Henslow’s Sparrow. In
addition to the behavior discussed here, similar types of COAS can be created to perform
)
—-. —-..—

multiple species management or for balancing wildlife objectives with other objectives such
as erosion control or training realism.
The Land Use Plan Entity object holds two state variables and one behavior (Figure 5).
t! Land Use Plan Object
State variables
Numberof MIMs
Trainingdistributionmap
Behavior
ImplementTrainingCOA
-Decidewhereto train
-Distributetrainingimpact
-OutputupdatedIandcover
-Outputupdatedsuccessionaltime-step
Figure 5. Land Use Plan Object State Variables and Behaviors
State Variable~ Except for the Training COA, no other models cumently implemented in the
00-IDLAMS prototype need the variables held by the Land Use Plan object. However, because
existing models and applications that could provide these variables have been identified, they have
been included as state variables.
1.
2.
Number of Manuever Impact Miles (MZMS) –MINIs are calculated as part of the Army
Training and Testing Area Carrying Capacity (ATTACC) program (previously known as the
Evaluation of Land Value Study (ELVS) (Concepts Analysis Agency, 1996). One MIM has
the equivalent impact to the land base as an M1A2 tank driving one mile in an Armor
battalion field training exercise (FIX). The MIM value for each mission activity is derived
from the number and types of vehicles used, the miles that the vehicles travel, and other
characteristics of the activity that affect the land base, such as digging or demolition.
Training distribution map - This state variable is aSpatial Data Set obiect that represents the
spatial ~istribution of land use activities. Land use”activities are no; distributed uniformly
across the installation. The training distribution map should reflect the effects of topography,
vegetation, and other environmental influences on the distribution of land uses such as
training and testing. It should also reflect the doctrinal requirements of training and historic
land use patterns.
Behavior:
1. “Implement Training COA” -The Land Use Plan object can implement the Training COA,
which uses the state variables number of MIMs and training distribution map as input and
outputs an updated Iandcover map and successional time-step map (state variables held by the
Landcover object).
-~,, ,, , ,,,:.% -77,~ -..
.. . . . ...’ ,&. ->. ,!J .. . . .. .. .. . . . . ... ... .~.. .,,,, .. . . .. . . . ,. .,....> ,, .+-.
,——,... —

The Henslow’s Sparrow Entity object holds two state variables and two behaviors (Figure 6).
Henslow’s Sparrow Object
II
PII
4%’ State variables
Habitatsuitabilitymap
,.
?Patchesmapandadditionalacresneeded
Behavior
Updatehabitatsuitabilitymap
IG;t patchesof excellent_vegetation
characteristics
Figure 6. Henslow’s Sparrow Object State Variables and Behaviors
State Variables:
1. Habitat suitability map –This state variable is the Spatial Data Set object that results from
implementing the Henslow’s Sparrow Habitat Model registered to 00-IDLAMS.
2. Patches map and additional acres needed- This state variable combines the Spatial Data Set
object calculated within the Henslow’s Sparrow Habitat model and the number of hectares
that are needed to fulfill the 65-hectare patch size requirement.
1. “Update habitat suitability map” –This behavior implements the Henslow’s Sparrow Habitat
model, which uses the current states of the landcover map and burned areas map (held by the
Landcover object) as input.
2. “Get patches of excellent vegetation characteristics” –This behavior is implemented when
the Planting COA asks for an updated state of the patches map and additional hectares
needed state variable.
SUMMARY
Effective ecosystem management relies upon assessing the components of the system simultaneously.
Modeling and simulation in support of adaptive ecosystem management can be better accomplished
through adynamic, integrated, and flexible approach that incorporates scientific and technological
components into acomprehensive ecosystem-modeling framework. This framework should be flexible,
allowing new data, concepts, and technologies to be incorporated without significant reworking of the
system. In addition, this approach should provide run-time feedback among models, thereby making it a
more dynamic tool for exploring and providing insight into the interactions among ecosystem processes.
Finally, the approach should be cost effective by encouraging the reuse of existing technology and by
allowing applications to be executed in their native languages. An object-oriented approach provides the
flexibility, dynamics, and reusability needed for the next-generation ecosystem-modeling framework.
This research will result in an object-oriented prototype (00-IDLAMS) to test and evaluate anext-
generation software architecture. This architecture will allow for acomprehensive landscape or ecosystem
approach to integrated natural resources planning and ecosystem management in support of the long-term
military mission. 00-IDLAMS offers enhanced capabilities to:
.Allow for the integration of existing legacy–type models without extensive reworking,
thus capitalizing upon previous investments in already available models and
applications;
.Encourage the development of object libraries that contain alarge number of reusable
objects to represent awide variety of natural and artificial elements of the environment
and therefore reduce the long-term cost of redeveloping objects and technologies;
._... ,, ... ,.,,->77’.=-.7 ... .<.”. %. .,! :,. ,: : . . ...
“<
,.:, .
-
~. -.- ..>+-, .,..’ %: e. . ., ,.. cr,----
—....—.
:.-... .

.Provide an integrated architecture that allows for the ability to reflect the dynamics of
living ecosystems, land uses, and land management practices;
●Support softwye applications that can operate at multiple spatial and temporal scales;
and
.Incorporate new data, concepts, and technologies that will bring the best available
knowledge, science, and technology to bear on decision-making processes important to
the military mission.
REFERENCES
Bennett D.A., M.P. Armstrong, and F. Weirich. (1996). An Object-Oriented Model Base Management
System for Environmental Simulation. In: GLS and Environmental Modeling: Progress and Research
Issues. Goodchild et al., eds. GIS World Books, Fort Collins, CO, pp. 439=M4.
Berry, M.W., R.O. Flamm, B.C. Hazen, and R.L. MacIntyre. (1996). ASystem for Modeling Land-Use
Change. IEEE Computational Science &Engineering 3(l): pp. 24-35.
Concepts Analysis Agency. (1996). Evaluation of Land Value Study (ELVS), Study Report CAA-SR-96-
5, U.S. Army Concepts Analysis Agency, Bethesda, MD.
Frysinger, S.P., D.A. Copperman, and J. P. Levantino. (1996). Environmental Decision Support Systems
(EDSS): An Open Architecture Integrating Modeling and GIS. In: GZS and Environmental Modeling:
Progress and Research hues. Goodchild et al. Eds. GIS World Books, Fort Collins, CO, pp. 357-362.
Fedra, K. (1996). Distributed Models and Embedded GIS: Integration Strategies and Case Studies. In: GIS
and Environmental Modeling: Progress and Research Issues. Goodchild et al., eds. GIS World Books, Fort
Collins, CO, pp. 413-418.
Li, Z., P.J. Sydelko, M.C. Vogt, C.M. Klaus, K.A. Majerus, and R.C. Sundell. (1998). Integrated Dynamic
Landscape Analysis and Modeling System (IDLAMS): Programmer’s Manual. ANL/ESD~ -146,
Argonne National Laboratory, Argonne, IL, and USACERL-98/127, U.S. Construction Engineering
Research Laboratories, Champaign, IL.
Maxwell, T., and R. Costanza. (1995). Distributed Modular Spatial Ecosystem Modelling. International
Journal of Computer Simulation, Special Issue on Advanced Simulation Methodologies, 5(3): 247–262.
—,, .... . . ../7- T-?——