Predicting Off-Road Army Training Impacts with the Aid of Arc/Info's Open Development Environment (ODE)

Abstract:

To assist in the prediction of vegetation and soils impacts during Army training exercises a model was developed at the U.S. Army Engineer Research and Development Center's Construction Engineering Research Laboratory. The Training Use Distribution Model (TUDM) predicts the spatial distribution and intensity of off-road maneuver miles. The implementation of the model required a user interface and processing architecture built in Visual Basic to interact with a Geographic Information System (GIS) capable of grid analysis. Environmental System Research Institute's Arc/Info Open Development Environment (ODE) GIS provided the perfect match for this need.

by William D. Meyer and Patrick J. Guertin

Background:

Since the early 80's, researchers and land managers from the Engineering Research and Development Center - Construction Engineering Research Laboratory (ERDC-CERL), have been developing means to quantify and predict sustainable vehicle use rates associated with maneuver training exercises for Army training lands, referred to as Land Based Carrying Capacity (LBCC). Early efforts have included Balbach and Coin's (1984) proposed conceptual model for quantifying land use demands based on vehicle category. Warren and others (1989) integrated the Universal Soil Loss Equation (USLE) with a geographic information system to assist in planning military activities to reduce environmental impacts. Diersing and others (1989) built upon this approach and created the Tracked Vehicle Day (TVD). The TVD approach characterized maneuver training areas (MTAs) by the level of vehicle use that could be sustained before erosion rates exceeded acceptable levels. All these efforts were primarily focused on determining the effects of current training events on the landscape, and were not developed as tools to predict long-term conditions based on future use.

Since these early efforts, two new projects have been undertaken. The first is the Army's current standard for determining carrying capacity, this being the Army Training and Testing Areas Carrying Capacity model (ATTACC) (Anderson et. al. 1996). Similar to early efforts, the ATTACC model bases carrying capacity on erosion levels. In addition, it adds the abilities to estimate costs associated with land rehabilitation as a result of use, thus allowing the model to be used in the standard business processes used by the Army to calculate funding requirements for training. The second model, the Ecological Dynamics Simulation model (EDYS), is based on successional dynamics of plant communities in MTAs (McLendon et. al. 1996, Childress et. al. 1996). EDYS surpasses early efforts as it goes beyond soil erosion to measure sustainable use levels by incorporating plant community variables and biodiversity. In addition, EDYS is designed to predict future training land conditions, allowing land managers to assess future land conditions based and experiment with alternate management strategies.

The TVD and other early methods treated the impacts of training activities on the landscape as uniform occurrences. This assumption of uniformity does not reflect the effects of topographic, vegetation, and other environmental influences on the distribution of land use (Dubois 1994, Krzysik 1994). The ATTACC model uses a variety of methods based on slope, vegetation and similar factors to represent the distribution of training related impacts as training events occur (Guertin...). To incorporate the effects of maneuver impacts as ecological stressors into the EDYS model, ERDC-CERL researchers developed the distribution pattern ideas used within ATTACC and incorporated them into a GIS environment to allow for long-term distribution and intensity estimations. The product resulting from this effort is the Training Use Distribution Model (TUDM).

The TUDM Model:

The Training Use Distribution Model was developed to provide long-term predictions of off-road army maneuver training impact distributions and intensities for carrying capacity and other ecological simulation models, especially EDYS. In short, represent the patterns of vehicle traffic on the MTA landscape as well as the number of vehicle passes over a specific area of terrain. Model development was based on the following criterion; the use of simple models and existing installation field and GIS data to facilitate quick, low cost implementation to a variety of Army installations. Additionally, model output needed to be in an easily accessible GIS format to capture spatial patterns as well as interface with ecological models which functioned in a GIS environment, preferably utilizing commercial off the shelf software (COTS).

Based on these criterion, TUDM was developed to represent the accumulative impacts of maneuver training over extended periods of time (1 year +), with output calculated at both monthly and yearly intervals. These intervals coincide with vegetation growth functions in the EDYS model (McClendonet et. Al. 1996). Output takes the form of GIS layers representing the following levels of vehicle impact at a 50 meter X 50 meter grid scale: 1) percentage of cell not impacted by vehicle traffic, 2) percentage of cell impacted by 1 to 5 vehicle passes, 3) percentage of cell impacted 5 or more times by vehicle traffic, and 4) average tracking per cell. These levels of impacts were defined based on the soil impacts of vehicle traffic reported in Thurow, et. Al.(1995).

TUDM was developed based on the following framework:

1. User Input Mechanism
2. Maneuver Distribution Impacts Map (MIDM)
3. Event Schedule Data Base
4. Event Placement Database
5. Intensity Calculation model
6. Output Mechanism

This framework was incorporated into a GIS environment where processing is accomplished through a combination of Environmental Systems Research Institute's (Esri) Arc/Info 7.2.1, ArcGrid raster analysis environment and a Java sub routine.

The User Input Mechanism currently exists as a graphical user interface (GUI), which allows model users to construct training mission scenarios for model interpretation. Data input into the front-end includes the types and quantity of maneuver exercises to be simulated, as well as dates and locations. Examples of the GUI, along with programming details are discussed in later sections of this paper. Future versions of TUDM may include an interface that allows EDYS, and other models to interact without user input.

The primary elements of the TUDM framework are the (II) MIDM and the (III) Event Schedule Database. TUDM predictions revolve around the principal that military maneuver training events can be described in terms of miles driven, and average vehicle track widths per event. Distance multiplied by track width results in area tracked, which is then transferred to the MIDM. The MIDM map provides a probability surface, which allows vehicle traffic to be allocated to the MTA landscape based on slope, vegetation structure, and other factors which influence where disturbance occurs. MIDM was developed using regression techniques that have been effectively used in range and wildlife management for placing animal activity on a landscape based on factors which influence behavior (Senft, Rittenhouse, and Woodmansee 1983; Van Manen and Pelton 1997). The event database is populated with event data including miles trafficked, vehicles types, and average vehicle track widths.

The event placement database (IV), Intensity Calculation model (V), and Output Mechanism round out the internal workings of TUDM. The event placement database contains probability information for placing exercises in historically accurate MTAs when locations are not specified by the user. The Intensity Calculation model is based on a binomial distribution (impacted/non-impacted) and calculates the percent areas within each grid cell that are impacted at different intensities (none vs. single vs. multiple vehicle passes). The output mechanism currently exists as options to produce GIS map layers in two formats; Arc/Info's Arc Grid raster map, or Geographic Resources Analysis Support System (GRASS) ASCII raster map, and an accompanying map viewer.

Model Architecture

Object-Oriented Method:

TUDM was developed in the Visual Basic 6.0 and Java 1.2 programming languages. Object-oriented principles were applied in TUDM through the implementation of Visual Basic form-level classes and the creation of three custom classes. Classes form the basis for objects by defining an object's properties and methods. Properties refer to data members needed by an object and methods are processes that the object performs. An example of an object used in the TUDM program is the Exercise object, where exercises are constructed to be grouped into scenarios for model analysis. When an object needs to perform an activity it is instantiated (created) from its base class. The properties and methods defined for the class form the interface between the object and the class. The properties for the class are set through the object, interface and code contained in the class is executed through interfacing with an object's methods. Through the use of this approach programs are more easily maintained. In the event of an error (bug), the error only needs to be corrected in the base class for the fix to be propagated throughout the entire program. Objects can also broadcast messages to each other through the use of an event. This is useful when you need to let one object know that another object has completed a processing activity.

Class Architecture

TUDM is composed of five main components: a front-end Graphical User Interface (GUI), a back-end processing architecture, a Microsoft Access database of Army doctrine training maneuver inputs, utility modules, and an HTML help system. An illustration of the five main components and their sub components are presented in the diagram below.

(Figure 1) TUDM Object Diagram

(Figure 2) Graphical User Interface Object Diagram

The graphical user interface (GUI) was developed to provide the means for manual interaction with TUDM. Working from the main form the user would step through a series of input screens to parameterize a model run. This interaction is expressed in the object diagram listed below.

(Figure 3) TUDM Object Interaction

(Figure 4) Main Form

The screen presented above is the main model input form. Here, the user begins by selecting from the Maps section the input maps necessary for model processing. Clicking on the Select button for the Impact Map executes the TUDM MapBrowse object interaction yielding the Map Browser selection screen.

(Figure 5) MapBrowse Object Interaction

(Figure 6) Map Browser Form

Highlighting and clicking on the Select button will select the impact map for model processing. This same process is repeated to select the Training Areas map.

The user then moves to the Scenario section of the main form and either selects a predefined scenario or decides to create a custom scenario. If a custom scenario is chosen by selecting the custom radio button and clicking on the create button, the scenario section of the object diagram executes.

(Figure 7) Scenario Object Interaction

(Figure 8) Scenario Builder Form

Through the scenario builder the user names the scenario to be built, sets the beginning and end dates for the scenario, and begins to assemble exercises to be included in the scenario. Clicking on the Add button will execute the following section of the object diagram and bring up the Exercise Selection form.

(Figure 9) Exercise Object Interaction

(Figure 10) Exercise Selection Form

The exercise selection form is directly linked to the Access database of Army Doctrine exercise descriptions. When the user selects a battalion type (BN Type) and a description from the drop down list boxes, a query is executed that returns the miles, track width, and percentage tracked from the database for use as model inputs. If a user does not know where an exercise takes place they may check the "Auto Select Training Areas" check box and training areas will be selected automatically based on a probability of past exercise patterns. Once the user is satisfied with what they have selected a click of the "Ok" button will add the exercise to the scenario. They may continue to add exercises to the scenario or return to the main screen to run the model.

The last element of the GUI is the ability to view model outputs through the MapViewer object. Selecting the Map Viewer button will execute the MapViewer section of the object diagram, which brings up the Map Viewer form.

(Figure 11) Map Viewer Object Interaction

(Figure 12) Map Viewer Form

(Figure 13) Map Legend

When the user selects a map to view, a legend describing the map's contents also displays. Through the map viewer the user can view outputs from the model, zoom in, zoom out, or query cell locations for model output values.

Processing Architecture (Back-End)

(Figure 14) Processing Architecture Object Diagram

The processing architecture was designed to fully leverage TUDM's ability to integrate on a machine level with another program or with a different front-end GUI without the need for code rewriting, thus keeping with the intent of following an object-oriented design paradigm. The design allows any program or GUI to parameterize the model, provided it correctly sets the input accessors.

Once the input parameters are set through the GUI, the model can run. During the run, the TUDM base class module maintains model control.

(Figure 15) Processing Control Object Diagram

The TUDM module cycles through a series of loops and subroutines while it executes the analysis code. The model begins with the process of sorting the interval collection into yearly or monthly intervals. It then sorts through the scenario, grouping the exercise inputs and training areas into a processing list. At this point the model works with one 50 x 50-meter grid cell at a time processing the inputs against the Maneuver Impact Distribution Map values through the Esri Grid Control object and the Java binomial probability sub-routine. This interaction is accomplished utilizing a combination of AML procedures, GRID processing, and System commands as depicted in the following section of code.

Public Sub makeProbabilityMaps(SelGrid As String, width As Double)

grid.Command "verify off"

grid.Command "mape " & SelGrid

grid.Command "setwindow " & SelGrid

'A cell is 50 x 50 meters

grid.Command "tracks = int( " & SelGrid & " / " & width & " / 50 )"

grid.PushString "ARC"

grid.PushString "SELECT TRACKS.VAT"

grid.PushString "OUTPUT ../VALUES.DAT INIT"

grid.PushString "PRINT VALUE"

grid.PushString "QUIT STOP"

grid.Command "arc info"

grid.Command "&describe tracks"

nClasses = grid.GetVariable("grd$nclass")

'Call Java program that calculates probability

Dim binprob As String

Dim fact As Double

Dim cutoff As Double

fact = 1000000

cutoff = 200

'Class 0

End Sub

Importance of Java Sub-Routine:

Utility Modules

The utility modules provide frequently used functions and sub routines that support the processing architecture. Contained with in these modules are the Esri constants, AML capabilities, strings collection, and file input and output functions. Several of the Esri AML capabilities provide the means by which the model extracts information from Arc/Info Grid files that are essential for model inputs. The Esri strings collection also provides the means to receive prompts from the Arc Grid object pertaining to the progress of Grid Analysis processing.

Model inputs are supplied through an Access database of information on training exercises based on Army Doctrine. The database contains the battalion name, description of an exercise, and its relevant information concerning miles tracked, track widths, and percentage of tracking. To determine what percentage of an area is tracked by wheeled vehicles, the percentage tracked is subtracted from 100. The database also contains a table of probabilities for where a training exercise occurs based on past training exercise patterns. If a user does not know where an exercise occurs they may select an option that uses this probability table to distribute training miles based on historical record. The data connection object is a Visual Basic object that facilitates the link between the program and the Access database. In the TUDM model the database is connected to the model through the GUI, but it could just as easily be connected directly to the processing architecture if a program-to-program interaction was desired.

HTML Help System

The help system, written in the Hypertext Markup Language (HTML), was developed in Microsoft HTML Help Workshop. The help system consists of a tree structure of topics and content pages that are written in HTML. Each of the pages is viewed using the Microsoft Internet Explorer web browser. A complete set of content pages and topics are compiled into a help project file which can then be linked to TUDM through the program's properties. Help can be invoked by hitting the F1 key or by clicking on a "Help" button or menu item. Clicking of either option elicits a Windows Application Programming Interface (API) call, which activates the help system and Internet Explorer. An example of the Help System interface is shown below.

System Requirements:

To run the prototype you must have the following:

6 Gigabyte or greater hard drive

Windows NT 4.0 Workstation w/service pack 3.0 or higher

Microsoft Internet Explorer 4.0 or higher

Work Station Arc/Info 7.2.1

Java Runtime Environment 1.2

256 MB RAM

6 Gigabyte or greater hard drive

Windows NT 4.0 Workstation w/service pack 3.0 or higher

Microsoft Internet Explorer 4.0 or higher

Work Station Arc/Info 7.2.1

Java Runtime Environment 1.2

During testing of the model it was found that exercise scenarios that contained more than 5 exercises with grater than 5 training areas required a processing configuration consistent with the preferred system requirements. However, scenarios of a lesser size may be successfully run with the minimum system requirement configuration.

References:

Anderson, A.B., R. Hunt, L. Chenkin, D. McFerren, P. Sydelko. 1996. Evaluation of Land Value Study (ELVS)/ Army Training and Testing Area Carrying Capacity (ATTACC). In: Gary Larsen (ed.) Proceedings of the 5^th Annual Land Rehabilitation and Maintenance Workshop, 27-29 August 1996, LA Crosse, Wisconsin.

Balbach, H.E. and S. Coin. 1984. Modeling Military Land Use Demands by Vehicle Categorization. Agronomy Abstracts, 1984 Annual Meetings of the American Agronomy Society. December 1984.

Diersing, V.E., R.B. Shaw, S.D. Warren, and E.W. Novak. 1989. A User's Guide for Estimating Allowable Use of Tracked Vehicles on Non-wooded Military Training Lands. J. Soil Conserv. 43(2):191-195.

Dubois, P. 1994. Impacts on Soils and Vegetation at Yakima Training Center, Washington. IN: The Proposed Stationing of Mechanized or Armored Combat Forces. Unpublished USACERL Report.

Guertin, P. J. 2000. Evaluation of the Maneuver Impacts Distribution Map and It's Use in ATTACC. Technical Report-00/01 (ERDC/CERL March 2000).

Krzysik, A.J. 1994. Biodiversity and the Threatened/Endangered/Sensitive Species of Ft. Irwin, CA. USACERL Tech. Rep. EN-94/07, Champaign, IL.

McLendon, T., W.M. Childress and D.L. Price. 1997. A Decision Support System for Allocation of Training Activities on U.S. Army Installations. IN: Jefferey M. Klopatek and Robert H. Gardner (eds.) Landscape Ecological Analysis: Issues, Challenges, and Ideas. Ecological Studies Series. Springer-Verlag. New York.

Senft, R.L., L.R. Rittenhouse, and R.G. Woodmansee. 1983. The use of regression models to predict spatial patterns of cattle behavior. Journal of Range Management. 36:553-557.

Thurow, T.L., S.D. Warren, and D.H. Carlson. Tracked Vehicle Effects on the Hydrologic Characteristics of Central Texas Rangeland, Technical Manuscript EN-95/02 (USACERL February 1995).

Van Manen, F.T. and M.R. Pelton. 1997. A GIS Model to Predict Black Bear Habitat Use. Journal of Forestry, 95:8, pp 7-12.

Warren, S.D., V.E. Diersing, P.J. Thompson and W.D. Goran. 1989. An erosion-Based Land Classification System for Military Installations. Environmental Management 13:2251-257.

Patrick J. Guertin

ERDC/CERL

PO BOX 9005

Champaign, IL 61826-9005

Phone: 1-800-USA-CERL (1-800-872-2375)

Fax: 217-373-7251

E-Mail: p-guertin@cecer.army.mil