Integrating Landscape Structure Programs with ArcInfo

Joseph K. Berry, David J. Buckley, and Kevin McGarigal

ABSTRACT

Most approaches in integrating geographic information systems (GIS) and existing forest management programs use import and export of data exchange files between distinctly separate programs requiring considerable programming knowledge of system syntax, file types and formats, custom conversion programs and operating system procedures. Intimidating and confusing mechanics of the linkage has severely limited effective use of GIS in landscape analysis. A tight integration of data and processing capabilities are vital to successful resource applications. Coding landscape structural analysis within a GIS environment provides significant advantages in data conditioning, processing speed, and extended analysis capabilities. Users interact through cascading graphical menus to easily identify geographic extent, map features, processing parameters and output specifications for their study. Computer processing takes advantage of the set of mathematical, statistical and spatial operators within GIS programming languages to expedite program development, increase performance and enhance traditional landscape indices. This integration of raster and vector techniques also provides new indices, such as considering "effective distance" in computing proximity-related metrics. This paper discusses the differences in application development between loosely and tightly integrated approaches in linking GIS and landscape analysis. Examples using the FRAGSTATS*ARC system linking the ArcInfo and FRAGSTATS programs are used to reinforce the concepts presented.

Introduction

How will forest vegetation change in response to natural succession, imposed disturbances, and proposed management actions? How will these changes affect ecosystem values? How do changes in the shape, pattern and arrangement of forest parcels modify flows of energy, materials and species? Answering these questions requires a synthesis of scientific knowledge about important ecosystem components and a means for integrating the knowledge in a format useful to resource managers.

Ecosystem management at any scale requires comprehensive data sets and robust analysis procedures. Currently the most common landscape analysis scale being used is at the watershed level. The question now is whether resource managers can take advantage of the large investments in detailed geo-referenced data at the stand level in developing advanced analytical techniques for ecosystem management at larger scales. A major obstacle is the procedural complexity of an integrated environment for the storage, processing and display of mapped data used in forest simulation and landscape analysis. Currently the linkage involves a litany of independent and specialized programs that severely limit rapid map displays and comparisons of future conditions for direct interpretation by forest managers. Communication between these systems and GIS data has been left to end-users and requires considerable programming knowledge of both systems’ syntax, file types and formats, custom conversion programs and operating procedures. The intimidating and confusing "mechanics" of the linkage has severely limited the effective use of GIS, particularly in landscape structure analysis.

This paper describes the major considerations affecting successful integration of landscape structure analysis and GIS systems. The FRAGSTATS*ARC software (IGIS 1997) is used to demonstrate the linkage of the U.S. Forest Service’s FRAGSTATS program (McGarigal and Marks 1995) to the ArcInfo GIS (Esri 1992). The tightly integrated system utilizes a graphical user interface that controls program specification, data management, data processing and output display. The structure and organization of the menus form an intuitive, "wizard-like" approach that leads a user, step by step, through program specifications needed to perform the analysis identified by the user.

Linking GIS and Landscape Analysis Capabilities

FRAGSTATS is an extensively used program that derives a comprehensive set of useful landscape metrics. The ArcInfo system (Esri 1992) is a powerful, workstation-level GIS that is widely used by natural resource agencies, organizations and companies. Many agencies and companies have spent much of the last 10 years compiling expensive and comprehensive vegetation inventories that describe their areas of responsibility. However, most often this wealth of descriptive data has only been utilized for basic mapping purposes. The benefits from analyzing this data, using GIS analysis functions or by using externally developed analysis models or programs, has yet to be realized. However, most GIS systems were justified and acquired based on these analysis expectations.

Internal Versus External Linkage Techniques

Two basic programming techniques exist for linking landscape analysis models with ArcInfo. These are:

	internalizing the model/program by programming with ArcInfo functions using the GIS' programming languages, e.g. AML, GRID, etc.; or
	maintaining the external programs outside of ArcInfo and communicating with the model/program through inter-application communication techniques.

Internalization of the landscape analysis model/program is typically desirable. Most GIS software, and in particular ArcInfo, provides rich programming languages, (e.g. AML, GRID), and a comprehensive suite of spatial analysis functions. There are several advantages to this approach including:

	data formatting and management requirements are minimized since the GIS inherently takes care of these data integration requirements;
	a comprehensive suite of spatial operators and functions is available;
	seamless support for both vector and raster data formats and processing techniques;
	typically faster processing performance; and
	hardware / operating system platform independence of the application.

The primary disadvantage of internalizing the model/program is that frequent re-programming of an existing model, e.g. coded in FORTRAN, C, etc., is cost prohibitive. In many cases the model has been developed over many years and can be fairly large. Significant effort would be required to duplicate the model in the GIS. This effort is typically viewed negatively - why re-code the model when it is already done ? As well, an extensive knowledge of ArcInfo is required before any internalization can be undertaken. In many cases this level of expertise is not readily available to the researcher or scientist.

If a substantial amount of development has already been undertaken on an external model/program it is usually desirable to maintain the model outside of ArcInfo. However, this approach requires developing a mechanism for ArcInfo to interact with the model. Interaction requirements usually involve data input and definition, data conversion (to a format the model can read), passing of parameters for the model run, execution of the model/program, and conversion of the model output to an ArcInfo compatible format for display purposes. In most instances a graphical user interface (GUI) is developed with ArcInfo, historically ARC Macro Language (AML) based. The easiest approach is to utilize the ArcInfo application GUI as the front end for the model execution. The GUI does not directly interact with the model during execution , but rather is used to define parameters for the model run. The passing of parameters to the external program can occur in a variety of ways. The number of parameters will help determine the most appropriate method for passing arguments and parameters to the external program. In cases where numerous parameters are required the use of ASCII parameter files is best. In many cases ASCII parameter files are already used by the external model. In this regards, the ArcInfo GUI is simply used to write parameters files and for display of output.

The public domain version of vector FRAGSTATS developed by McGarigal and Marks utilized the external program approach. Much of the data input and processing was done with ArcInfo AML, but the indices calculation was done with C programs, i.e. edge contrast metrics. The ArcInfo AML code utilized functions to prepare the spatial data for analysis, converted the data into ASCII files, and then called the C programs for calculating metric output. In this context no direct interaction occurs, other than ArcInfo executing the C program.

More recently inter-application communication (IAC) capabilities have been added to the ArcInfo AML affording more direct interaction between external programs and ArcInfo. These techniques allow spawning of multiple independent tasks, i.e. model runs, simultaneously from the ArcInfo GUI. This is ideal for batch like processing. In this regard the client GUI is released after initiating a model run. The external program acts like a server process echoing messages back to the GUI depending on status. The GUI is free to interrogate previous model run output. The GUI can communicate with the model server at any time by sending requests to the server process. This is ideal for changing parameters during iterations of a model.

The IAC approach is ideal for integrating complex simulation models having many parameters with ArcInfo. Multiple simulation runs can be started at once, and parameters can be changed interactively through the GUI to adjust the model based on interpreted output. This approach has been successfully implemented with the CAPARS system that integrates sophisticated atmospheric dispersion models with ArcInfo for emergency response to chemical disasters (Buckley et all 1997).

A primary requirement for managing model/program integration with a GIS is a meta-data system. Whether the implementation approach utilizes internal or external linkages, there is a requirement to maintain meta-data information about each model/program run. A meta-data approach ensures that information about each run is captured and maintained by the application. This meta-data affords users the ability to re-visit runs for display, interpretation and/or evaluation. The meta-data tells the user about the specific model/program run by storing input parameters, characteristics and output data. A suite of model run management tools is required to aid the user in managing model run output.

The FRAGSTATS*ARC Integration Approach

Previously, communication between FRAGSTATS and ArcInfo has been left to end-users and required considerable programming knowledge of both systems’ syntax, file types/formats, custom conversion programs and operating system procedures. Unfortunately, these intimidating and confusing "mechanics" of the linkage has severely limited the effective use of landscape analysis, in particular FRAGSTATS, in natural resource planning and management.

A fully integrated link between ArcInfo and FRAGSTATS has been developed with a single user interface that controls program specifications and operation, data integration, and output display. The "seamless link" provides a user-friendly, and intuitive, graphical menu system consisting of scroll lists, toggles and buttons to facilitate program specifications and automatically generate appropriate tables, charts and maps for enhanced interpretation of the results. The FRAGSTATS*ARC interface develops such an environment by controlling:

The FRAGSTATS*ARC interface focuses primarily on internalization of the FRAGSTATS program. To take advantage of the inherent data management and processing capabilities found within ArcInfo all indices calculations have been internalized and programmed in AML. All C code has been eliminated. This results in better data and file management, and also in significant performance enhancements. Internalizing the program calculations has been designed to take advantage of the inherent database management (DBMS) table capabilities inherent in ArcInfo. With this approach specific indices are only calculated once for the entire data set. DBMS table calculations inherently work on all features (polygons/records) with a single command. Previously with the public domain version of FRAGSTATS each indices was executed separately for every feature. Accordingly, execution times are significantly reduced with average program runs taking minutes instead of hours. This provides an operational framework for adding many new indices to FRAGSTATS, and embedding FRAGSTATS within other analysis applications, such as timber supply models, or harvest planning applications.

FRAGSTATS*ARC also utilizes the inter-application communication (IAC) approach for program integration. Under a comprehensive meta-data framework multiple FRAGSTATS program runs can be initiated using IAC methods. The IAC method has been implemented to afford batch processing of program runs. Individual FRAGSTATS runs are queued up as separate FRAGSTATS servers freeing up the interface for use.

The FRAGSTATS*ARC Application Desktop

The FRAGSTATS*ARC desktop interface provides a common processing environment and graphical interface for running FRAGSTATS within ArcInfo. User specifications for input data, FRAGSTATS analysis parameters and ArcInfo output options are made by toggling options in a series of procedural, step-by-step, menus. The general FRAGSTATS run process flow is illustrated in Figure 2.

The FRAGSTATS*ARC desktop enables scientists and managers to interact with both FRAGSTATS and ArcInfo without extensive procedural knowledge of either system. It tightly integrates both systems in a friendly environment allowing users to directly concentrate on ecological planning and management implications, without the distracting "mechanics" of program control and output generation.

FRAGSTATS*ARC has been designed with a fixed format menu interface that mimics the conventional Windows interface by placing main menus at fixed positions on the screen. This approach is often referred to as a desktop. The diagram shown below illustrates the basic desktop with relative menu areas and positions. The basic premise of the desktop interface is the definition between graphics windows space and popup menu space. The desktop utilizes a percentage of the screen, to the left and below the main menu, for popup menus. The remaining area is designated for the graphics window.

Basic Versus Advanced Interface

The FRAGSTATS*ARC interface has been designed to accommodate users with two different levels of ArcInfo experience. The Basic interface will accommodate most users who have a working knowledge of the ArcInfo data model. The Advanced interface is intended to supplement the Basic interface capabilities by providing a more comprehensive suite of ArcInfo data manipulation tools. The more experienced ArcInfo user, or those needing a complete range of data manipulation and pre-processing capabilities, will find the Advanced interface more appropriate. The primary capabilities for pre-processing your data, running FRAGSTATS, and reviewing FRAGSTATS output are available in both the Basic and Advanced interface. The Basic level is a simplified pulldown main menu with all the capabilities required to run FRAGSTATS.

The Advanced interface provides a more comprehensive main menu that also provides access to numerous ArcInfo data management and manipulation capabilities. These advanced tools provide access to virtually any ArcInfo command. The Advanced level is intended to satisfy the needs of the more experienced ArcInfo user. In many situations significant data pre-processing is required before it is ready for FRAGSTATS processing. A number of data integrity checks have been built into the FRAGSTATS*ARC data selection tools that notify the user of topological and spatial errors in the selected input data set. The Advanced interface provides a comprehensive suite of tools to aid the user in correcting these spatial errors prior to running FRAGSTATS.

Selecting Landscape Structure Indices

The step-by-step Wizard interface for FRAGSTATS focuses on the seamless linkage of indices selection and required parameters. Users are automatically prompted for required parameters when a specific category of indices are selected for calculation. Table 1 presents the general categories of indices currently available within FRAGSTATS*ARC. A user simply has to toggle on the category of metrics they are interested in to see a list of individual indices for the PATCH, CLASS and LANDSCAPE level. If a data parameter is required, the user will automatically be prompted through a popup menu to define the parameter(s).

The design of the GUI for metric selection reflects the inherent synchronization between metrics and required parameters. This mechanism provides an inherent flexibility for extending indices as the product develops and ensures that required parameters are automatically linked to categories of metrics for the user. The approach minimizes the up-front knowledge required from the user, and also instills an understanding of the relationships between data parameters and indices values with the user. An obvious extension is to provide the capability for users to add their own indices to the metrics list with future releases of FRAGSTATS*ARC.

Figure 5. The indices selection menu enables users to select categories and levels of indices to be calculated. The required parameters are automatically identified and prompted for.

Figure 5 shows the main metric selection menu. By toggling on a category(s) for calculation, the list of individual indices, and required data parameters, are refreshed in the lower scroll lists. The Categories, representing logical orderings of individual indices, and data parameters are linked at this level. It was decided that the provision of required data parameters menus for individual indices would be information overload for the average user, and would result in a less than intuitive interface.

Intregrated Query and Display Capabilities

Another inherent advantage of internalizing the FRAGSTATS program into ArcInfo is the availability of the GIS' display and query functions. A major impediment of the public domain FRAGSTATS is the lack of integration with the GIS display tools, and the substantial ArcInfo knowledge required to convert FRAGSTATS output to a compatible format needed to generate meaningful displays.

With FRAGSTATS*ARC the output metric tables are seamlessly linked with the input polygon coverage providing an array of query, listing, reporting, and display capabilities. A full range of table listing and reporting tools are provided that transparently operate on the FRAGSTATS output tables without the user requiring knowledge of file names or command syntax. Users simply select a FRAGSTATS run (through the meta-data interface) and the query, display or reporting tools automatically use the output for the active run. In addition to the table reporting tools, a robust thematic mapping interface is provided allowing users to generate multiple displays of output indices. Also, a tool is provided to aid the user in defining an appropriate legend (definition of classes intervals) based on the frequency distribution of the individual metric. Figure 6 presents an example of the primary thematic mapping tool.

In addition to the thematic mapping tools, a comprehensive query environment is included. The query tools seamlessly link the output metrics with the input ArcInfo coverage and provide a robust interface for querying FRAGSTATS results. A full range of query capabilities are provided including attribute, spatial and topological query functions allowing the user to address virtually any query requirement. This includes the overlap of other data sets, ( roads, riparian buffers, etc.) with the source polygon data coverage. The query environment is designed around a query ladder that automatically tracks the user's query logic in a step-by-step manner. Users can walk up and down the query ladder revising their logic and query specifications as desired based on the results of query steps. This approach provides maximum flexibility for defining queries, and modifying them based on the characteristics of a specific data set, or the results of the FRAGSTATS analysis. The thematic mapping and formatted reporting tools are seamlessly integrated with the query environment allowing users to generate maps and reports based on query subsets. Query logic can also be saved across sessions and restored for future query sessions.

Figure 7 illustrates an example of a query for small, irregular aspen stands. This example is composed of three simple attribute (by logical expression) steps where the output PATCH indices where queried for:

1. Dominant species = Aspen
2. Area less than 15 ha
3. Shape index greater than 2.0 (fairly irregular).

These areas are displayed as RED in the figure. Using the topological functions this query could easily be extended to address concerns of proximity, adjacency to other features, e.g. roads, or where queried features overlap with other features, such as high suitability areas.

The query environment provides tools to help the user in flexibly defining their own query logic, and modifying the logic as they dictate based on their interpretation of the query results. This approach is intended to guide the user without imposing fixed processing steps, or a prerequisite knowledge of ArcInfo analysis commands. It is expected that more comparative functions will be added with future releases to further aid the user in interpreting FRAGSTATS output.

Extending Landscape Analysis Capabilities

Landscape ecology provides a strong conceptual and theoretical foundation for understanding landscape structure, function and change. These factors can be defined as: "structure" is the spatial relationships among the distinctive ecosystems, or spatial elements, present; "function" identifies the interactions among the spatial elements; and "change" is the alteration in the structure and function of the ecological mosaic over time (Forman and Gordon 1986). The ability to quantify landscape structure is prerequisite to the study of landscape function and change. For this reason, considerable emphasis has been placed on the development of landscape metrics.

Many of the relationships can be expressed by indices derived through analysis of the shape, pattern and arrangement of three landscape element "levels": 1) individual patches (e.g., vegetation polygon), 2) classes of related patches (e.g., polygons of the same type/condition), and 3) entire landscape sets (e.g., all polygons). In addition to characterizing landscape indices by the level of spatial elements, they can be grouped by broad "categories" identifying the nature of the processing. These categories include area, patch density/size/variability, edge, shape, core area, nearest neighbor, diversity and contagion/interspersion metrics (see Table 1).

Table 1. Metrics computed in FRAGSTATS, grouped by subject area. See McGarigal and Marks. (1995) for mathematical definition of each metric. (mu) represents map units (meters or feet).

Processing approach and data handling procedures ultimately determine the efficiency and capabilities of an integrated system. As a stand-alone system, the metrics in the original FRAGSTATS program were calculated both in AML macros and for more complex indices, in external C programs. Data preprocessing was performed in the GIS and specially formatted import files were used to transfer data to C programs. Internal processing involved a series of programming loops that operated on individual spatial items (polygon features). Results were written as a formatted report, with no provisions for viewing the results in map form. Some capabilities exist to unload the output reports into ASCII delimited files that could be loaded into a DBMS of choice. However, this process is manual and in-direct. As noted in the previous sections of this paper, the primary operational limitations of such a loosely integrated system were its procedural complexity and ineffective presentation of results. The major technical limitations included inefficiencies in program development, maintenance, performance and extension.

Vector Versus Raster Processing Techniques

Most of the metrics identified in Table 1 can be evaluated in both vector and raster data structures. The original public domain version of FRAGSTATS was available as two distinct programs to support vector or raster data formats. The vector version supported ArcInfo polygon coverage format through a loose integration between AML and C programs. The raster version is a stand-alone C program that supports a small array of raster data formats. No direct support for ArcInfo GRID format is provided. Accordingly, considerable effort was involved in developing, modifying and maintaining the two programs. Additionally, specific indices were only available in the raster version, and not the vector version.

In contrast, the FRAGSTATS*ARC system is programmed entirely within the GIS macro language (AML). This approach replaces the AML/C integration of the vector version (and the C code of the raster version) with a single development environment using direct GIS function calls. This approach significantly streamlines program development and maintenance and affords seamless integration of vector and raster data processing techniques. This provides a programming framework for extending the list of metrics fairly easily in the future. The next version of FRAGSTATS*ARC will provide a suite of new indices, and will provide capabilities for the user to create their own indices. In this regard, the list of metrics is user extendable. This flexibility is made possible the internalization into the GIS programming language and taking advantage of the inherent GIS functional capabilities.

With this design the user does not need to be aware of the processing technique (vector versus raster) that will be utilized to calculate a particular metric. The software automatically converts data on-the-fly to/from vector-raster input sources depending in the requirements of the category of indices selected for calculation. The current version supports vector polygon data input and converts to raster GRID's as required. This accommodates the growing list of metrics that are inherently raster based. The next version of FRAGSTATS*ARC will also support direct raster input data, and provide vector polygon conversion where required.

In addition, the GIS programmed version realizes significant increases in performance. The performance of the core FRAGSTATS metric calculations has been significantly enhanced. Calculations are now done directly on linked data tables reducing run times to minutes from hours ! In addition, run times for large data sets (coverages) do not increase linearly. Run times for large coverages only increase a minor amount due to the use of the DBMS table calculation techniques inherently available in the GIS. For example, Table 2 reflects the results of tests undertaken on a small and medium size data set with the new FRAGSTATS*ARC software versus the public domain version.

Table 2. Comparison of processing times for FRAGSTATS programs.

Future Efforts

The extended spatial analysis operations in most GIS packages provide new opportunities for extending data handling and analytic capabilities. For example, vegetation maps at different years can be compared, the transition zones isolated as a new set of spatial elements, and the structural character of the derived data set analyzed. The "effective proximity" operator in most grid-based GIS systems can be used to extend many of the indices. For example, the "nearest neighbor" metrics identified in Table 1 currently uses simple proximity based on straight-line connectivity to establish distance among spatial elements. Under this limiting assumption, two elements will be identified as close (as the crow flies), although they are divided by open water that restricts seed, animal and other movements between them. By simply introducing a direct call to the extended set of spatial operators, such as effective proximity, new and more realistic indices are easily developed.

Similarly, spatial statistics provides new "tools" for comparing and summarizing the results of structural analysis. For example, the maximum, minimum, range and average of indices can be easily derived through "region-wide (zonal)" operators for subsets of a project area, such as all locations with steep, northern slopes. Comparative and predictive spatial operators, as well as descriptive statistics, can provide new and useful "views" of landscape structure relationships. Most GIS systems provide investigative tools, such as clustering and regression, as direct function calls. These tools readily extend landscape structure analysis to other spatially dependent variables, such as the distribution of animal populations.

Extending Landscape Metrics

An extension project for FRAGSTATS*ARC is planned to provide additional landscape structure indices and new capabilities for generating custom queries and rapid comparison of landscape indices. It will enable a user to launch several landscape structure analyses, then generate comparative statistics and output these results as a variety of tables, charts and map displays. For example, a user might compare three harvesting scenarios for an area to determine which indices significantly change and output the results a map of affected landscape units.

In addition, development of a Windows 95/NT version (a GIS platform independent version based on Esri MapObjects software) is under consideration. This version will link to standard desktop mapping systems, supporting formats such as ArcView and MapInfo, as well as ArcInfo. This version will have the additional advantage of not requiring an ArcInfo license to operate. This will provide greater flexibility of deploying FRAGSTATS to the end user's desktop. Most importantly, several new indices and modifications, such as the lacunarity index and various other texture metrics, will be incorporated within the FRAGSTATS*ARC extensions environment.

Conclusions

The quantification of landscape structure provides a new set of tools for understanding complex spatial patterns and interactions among landscape features. They promise new approaches for visualization, formulation and solving of real-world problems. However, the tools must be tightly integrated into a system designed for use by resource managers, scientists and other end-users with minimal procedural knowledge of the programs involved. Several advantages are realized by programming within the GIS environment:

1) spatially dependent processing is expedited and extended,
2) program development, maintenance and modification are simplified,
3) data handling is in a common environment, and
4) user interaction with the system is streamlined.

The FRAGSTATS*ARC system tightly integrates FRAGSTATS and ArcInfo capabilities into a common graphical user environment for storage, processing and display of mapped data and results germane to landscape structure analysis.

Acknowledgments

Three groups are collaborating on the FRAGSTATS*ARC system. The original author of FRAGSTATS, Dr. Kevin McGarigal, provide ecological expertise, algorithm design and computational code testing. Innovative GIS Solutions, Inc. (IGIS) are the primary software developers for this project under the direction of David Buckley. Berry and Associates // Spatial Information Systems (BASIS) provides project coordination and expertise in GIS modeling and spatial statistics.

The FRAGSTATS*ARC system extends the original FRAGSTATS effort supported by the Coastal Oregon Productivity Enhancement Program of which the USDA Forest Service Pacific Northwest Research Station (PNW) is a major partner.

References

Esri, 1992. Understanding GIS: the ArcInfo method. Redlands, California: Environmental Systems Research Institute.

Forman, R.T.T. and M. Gordon. 1986. Landscape Ecology. New York: John Wiley & Sons. 619 p.

Li, , H. 1990. Spatio-temporal pattern analysis of managed forest landscapes: a simulation approach. PhD. Thesis, Oregon State University, Corvallis, Oregon. 166 p.

McGarigal, Kevin and Barbara Marks. 1995. FRAGSTATS: spatial pattern analysis program for quantifying landscape structure. Gen. Tech Rep. PNW-GTR-351. Portland, Oregeon: USDA Forest Service, Pacific Northwest Research Station. 122 p.

O’Neill, R.V., J.R. Krummel, R.H. Gardner, G. Sugihara, B. Jackson, D.L. DeAngelis, B.T. Milne, M.G. Turner, B.Zygmunt, S.W. Christensen, V. H. Dale, and R.L. Graham. 1988. Indices of landscape pattern. Landscape Ecology 1:153-162.

Turner, M.G. 1990. Spatial and temporal analysis of landscape patterns. Landscape Ecology 4:21-30.

Turner, M.G. and R.H. Gardener (eds.). 1991. Quantitative Methods in Landscape Ecology. New York: Springer.

Wykoff, W.R., N.L. Crookston, and A.R. Stage. 1982. User’s guide to the stand prognosis model. General Technical Report INT-133. Ogden, UT: USDA Forest Service, Intermountain Forest and Range Experiment Station.

Joseph K. Berry, President, Berry & Associates, Inc. ( jberry@innovativegis.com )
David J. Buckley, Principal Consultant, Innovative GIS Solutions, Inc. ( buck@innovativegis.com )
Kevin McGarigal, Assistant Professor( McgarigcalK@forwild.umass.edu )

Suite 300, 2000 South College Ave. Fort Collins, CO 80525 E-mail: jberry@innovativegis.com WWW: http://www.innovativegis.com/basis	Telephone: (970) 490 - 5900 FAX : (970) 490 - 2300 E-mail: igis@innovativegis.com WWW: http://www.innovativegis.com