Knowledge-based Information Management in Decision Support for Ecosystem Management in the Pacific Northwest U.S.

ABSTRACT.

The Pacific Northwest Research Station (USDA Forest Service) is developing a knowledge-based information management system to provide decision support for watershed analysis in the Pacific Northwest region of the U.S. The decision support system includes:

(1) a GIS interface that allows users to graphically navigate to specific provinces and watersheds and display a variety of themes (vegetation, streams, roads, topography, etc.) and other area-specific information (relevant regulations, existence and location of analyses, plans, etc.),
(2) an analysis component that helps identify major concerns and the hierarchies of associated ecosystem processes requiring analysis, and assists the user in selecting an appropriate subset of analyses and in identifying and prioritizing data requirements and their sources,
(3) a report manager that displays the history, status, and details of analyses, and that documents the analysis process,
(4) a project manager that assists with planning and monitoring of data acquisition, and
(5) a hypermedia system that provides powerful navigation tools for accessing information in various policy and procedure documents.

The core of the system is the analysis component which contains dependency networks that link problem-solving knowledge about concerns, ecosystem processes, and data to specific landscape units. The goal dependency approach provides a scientifically sound method for determining data requirements, as well as a basis for prioritizing, acquiring, and evaluating information for watershed analyses. An initial set of concerns that comes from the public participation process expands into a more complete set of concerns that reflects underlying ecological associations. This network approach thus explicitly accounts for both public values and ecological processes.

INTRODUCTION

This is the first in a series of papers, documenting our object-oriented 

approach to analysis and design of a knowledge-based information 

management system that provides decision support for forest ecosystem 

management. Because this is the seminal paper in the series, it is useful to 

first provide some historical perspective.



Historical Perspective

The current trend among resource management agencies toward 

application of ecosystem management concepts has gradually evolved 

through the process of public debate on resource use over the past 35 

years, beginning with the rise of the environmental movement in the early 

1960s and the concomitant increase in a sense of public responsibility for 

the environment (Caldwell et al. 1994). In the late 1960s and 1970s, a 

flurry of environmental legislation in the United States (National 

Environmental Policy Act, Endangered Species Act, National Forest 

Management Act, and others) ensued. Increasing demands on managers to 

address complex, often contradictory laws and regulations in planning 

documents taxed management capabilities. At the same time, growing 

public distrust and dissatisfaction with management of public lands  was 

evidenced by increasingly frequent legal challenges to resource 

management by Federal agencies. In the Pacific Northwest, events 

culminated in a series of legal challenges to Forest Service management 

policy with respect to the northern spotted owl (the so-called Dwyer 

decisions, Caldwell et al. 1994) and a series of reports sponsored by the 

agency (Johnson et al. 1991, Thomas et al . 1990, Thomas et al. 1993). 

The net effect was to bring forest management on Federal land in the 

Pacific Northwest to a virtual standstill by 1992.



President Clinton convened the Forest Conference in Portland in April 

1993 in an attempt to bring all concerned parties together and try to break 

the gridlock on forest resource management in the region. The Forest 

Ecosystem Management Assessment Team (FEMAT) was commissioned 

shortly thereafter.



    In its instructions to the FEMAT, the Forest Conference Executive 

    Committee stated that the team's objectives were to identify management 

    alternatives that "attain the greatest economic and social contributions 

    from the forests" consistent with meeting "the requirements of the 

    applicable laws and regulations..." (Thomas 1994).



The FEMAT Report (FEMAT 1993) was issued in July 1993. Despite 

many detractors, the FEMAT process has produced the most concrete and 

thorough statement to date of what is required of ecosystem management 

(USDA Forest Service and USDI Bureau of Land Management 1994). The 

Standards and Guidelines, published as an attachment to the latter Record 

of Decision, define a variety of land allocation types, prescribe detailed and 

specific management procedures for them, and establish an aquatic 

conservation strategy with particular emphasis on management of riparian 

areas. Soon after publication of the FEMAT Report, the Regional 

Interagency Executive Committee (RIEC, since superseded by the 

Regional Ecosystem Office) was commissioned to oversee implementation 

of what has come to be known as the President's Plan. 



Nineteen working groups were established under the direction of RIEC to 

handle various aspects of implementing the President's Plan, including, for 

example, working groups on research and monitoring, information 

management, watershed analysis, adaptive management areas, and the 

adaptive management process per se. A group conspicuously missing from 

the RIEC working groups, however, was a system design group with 

responsibility to fit all the pieces together into a coherent system for 

analysis, planning, and management. 



Early Team Activity

The ecosystem management decision support (EMDS) design team of the 

Pacific Northwest Research Station (U.S. Department of Agriculture, 

Forest Service) was organized in November 1993 to begin developing a 

prototype decision support system (DSS) for ecosystem management. We 

first met as a team in February 1994. Team members had previously 

reviewed the FEMAT Report and numerous working group reports to 

RIEC as background material. Our initial task was to determine what 

aspects of decision support for ecosystem management could benefit most 

from the application of decision support technologies, given the basic 

expectations of ecosystem management (USDA Forest Service 1992), the 

current state of conceptual development of an ecosystem management 

process (FEMAT 1993), and current technological capabilities for 

implementing such a process. 



Focus on watershed analysis

Watershed analysis emerged as a particularly promising area for 

development. The FEMAT Report clearly identified watershed analysis as a 

key component of the ecosystem management process, and a first draft 

guide to watershed analysis (1) had just been completed, which,  on review, 

described a reasonably well defined process suitable for decision support 

system development. Moreover, a watershed analysis DSS can probably be 

scaled easily up to province-scale analysis.



By the end of the team's first meeting, we had roughed out a conceptual 

model for a watershed analysis decision support system, which was 

reviewed with key PNW Station and Region 6 staff. Based on favorable 

reviews of the team's concepts, at our next meeting, we began a detailed 

analysis of the watershed analysis process described in the first draft guide. 

Briefly, the watershed analysis process described by the draft guide can be 

summarized as:



1. Identify issues and concerns, setting priorities among them, setting 

   goals and objectives for management, and formulating key 

   questions;

2. Identify ecosystem processes that require analysis;

3. Set priorities among ecosystem processes of concern;

4. Implement analysis modules for data acquisition and analyses;

5. Describe ecosystem states and processes in the current landscape;

6. Predict trends for ecosystem states and processes; and

7. Summarize analytical results, and organize information for 

   reporting on design of riparian reserves, restoration activities, 

   transportation planning, monitoring, cumulative effects, and other 

   activities.



Relevant to step 2, the guide contained five tables, each representing a top-

level category of concern (e.g., anadromous fish). Each table contained a 

list of concerns, ecosystem state requirements for satisfying each concern, 

and an associated reference to more specific concerns. For example, a 

potential concern for anadromous fish was acceptable spawning sites, 

which was to be analyzed in terms of sufficient spawning gravel, 

appropriate substrate size (both of which are affected by mechanisms under 

substrate), and sufficient flow (which is affected by mechanisms under 

baseflow). Each mechanism, in turn, had an associated cause whose state 

was to be assessed, and data requirements. Although the first draft guide 

used various terms such as concerns, mechanisms, and causes, they can all 

be thought of as various levels of concern, so the present example suggests 

a hierarchy of concerns. Accordingly, the team decided that knowledge of 

relations among concerns, ecosystem processes and states, and data 

requirements for those processes and states could be best represented by 

dependency networks (Stone et al. 1986). 



PROJECT GOALS AND OBJECTIVES

The basic goal of decision support for ecosystem management is to 

maintain and improve forest ecosystem health and productivity by 

providing managers with guidance consistent with laws, regulations, and 

scientific principles related to ecosystem management and public values. 

Specific long-term objectives of the DSS project in support of this goal are 

to provide managers of Federal forest land with a system that:



1. Increases efficiency of decision processes in ecosystem analysis, 

   planning, and management;

2. Improves managers' ability to explain the reasoning behind 

   decisions to the public;

3. Ensures compliance with laws and regulations;

4. Supports scientifically sound principles of ecosystem analysis, 

   planning, and management;

5. Improves managers' understanding of laws, regulations, ecosystem 

   management principles, and public values and how they apply to 

   analysis, planning, and management;

6. Improves consistency in analysis, planning, and management within 

   and across spatial scales;

7. Integrates adaptive management into the DSS; and

8. Links individual site installations of the DSS into a regional 

   network to ensure consistency of decisions in time and space where 

   appropriate.



The less ambitious, but more immediate, objective of the project is to 

provide information management and decision support for watershed 

analysis in particular. We emphasize the larger and more long-term 

objectives, because these define a larger context within which the first 

prototypes need to be developed if they are to evolve beyond decision 

support tools for isolated watershed analyses.



In the larger context, most views of the world cross multiple ownerships. 

Coordinating analysis, planning, and management activities within and 

between public agencies and adjacent private landowners will be essential 

to ensure that assumptions about effects of management scenarios are 

valid. Therefore, users of the DSS at any particular administrative unit will 

need access to information managed by other units. DSSs operating at 

individual locations will need to be designed to operate in a network. In the 

DSS network, an administrative unit will have a local version of the DSS 

which can obtain any necessary information from other systems by 

telecommunication over the network. Network communication between 

administrative units will allow creation of explicit methods for achieving a 

negotiated resolution of multiple-scale decisions over time and space, a 

capability required to realize objectives 6 and 8. We noted earlier that part 

of the attraction of developing a prototype for the watershed scale is that 

we anticipate that it can be easily scaled up to operate at the level of 

ecosystem provinces (sensu FEMAT). This consideration is relevant to 

objective 6.



The first prototype DSS now under development is intended to provide a 

foundation that can be easily built upon to eventually realize the project's 

long-term objectives. We view the development of a competent 

information management system as a requisite condition for many of the 

long-term objectives.



KNOWLEDGE BASES

Dependency Networks as Metadata

Metadata is data about data. For example, a data base that stores 

information about the location, reliability, data collection protocols, and 

originators of inventory data (stored in a separate data base) is information 

about data, and hence can be considered metadata. In a broad sense, this 

would be any data involved in the interpretation of other data. Metadata 

can also exist in the form of rules or procedures used to arrive at the 

interpretation of data. In the narrow sense, the analysis DSS's metadata are 

dependency networks (Stone et al. 1986) that make up its knowledge base. 

These networks formalize current, available knowledge about 

watershed-level phenomena. The dependency networks can be used by 

analysis teams to identify data needs, and assist in the interpretation of 

field-derived data (Nash et al. 1992).



Use of Dependency Networks

Dependency networks are used in the Analysis subsystem described 

subsequently to formalize current scientific understanding of the 

(hierarchical) relations among concerns, ecosystem processes, and data 

requirements. Dependency networks are composed of objects (goals,  

subgoals, and data links). Goals and subgoals can be weighted and their 

truth value determined via fuzzy logic if necessary. Uses of dependency 

networks in our context are to:



1. identify data requirements for an analysis, 

2. rank missing data in order of relative importance to the analysis, 

   and

3. report the truth value of conclusions about ecosystem states and 

   processes given existing data.



In particular, we are using the NetWeaver application, developed at Penn 

State University  (Saunders et al. 1989, 1992), for design of dependency 

networks (2).



Dependency Networks for Watershed Analysis

In addition to the general system analysis and design activity of the EMDS 

design team, the team is also organized into three subteams that are 

constructing dependency networks for six broad topic areas:



1. terrestrial vegetation,

2. terrestrial fauna,

3. anadromous fish,

4. riparian systems,

5. surface water supply, and

6. roads and structures.



For each topic area, a subteam of 2 knowledge engineers is working with 

2-3 scientists and resource specialists to develop a prototype network. On 

completion, prototype networks will be reviewed by a larger panel of 8-10 

scientists and specialists. 



ANALYSIS AND DESIGN METHODS

Object-oriented methods for analysis and design are being used for system 

development. In more traditional methods of top-down structured design, 

analysis and design are based on functional decomposition of problems, and 

the methods evolved to support programming languages such as 

FORTRAN and COBOL. In contrast, object-oriented analysis and design 

have evolved to support languages such as Smalltalk, Object Pascal, and 

C++, in which problem decomposition is based on the concept of objects 

(Booch 1994). Object-oriented problem decomposition offers a number of 

advantages over the functional approach:



     Object-oriented decomposition yields smaller systems through the reuse

     of common mechanisms, thus providing an important economy of expression. 

     Object-oriented systems are also more resilient to change and thus 

     better able to evolve over time, because their design is based upon 

     stable intermediate forms. Indeed, object-oriented decomposition greatly

     reduces the risk of building complex software systems, because they are

     designed to evolve incrementally from smaller systems in which we already

     have confidence (Booch 1994). 



Rapid prototyping, with short cycles for analysis, design, and 

implementation, is fundamental to object-oriented methods, and also fits 

well with knowledge engineering methods being used by the team for 

development of the dependency networks. In the following section, we use 

Booch diagrams (Booch 1994) to graphically illustrate the physical and 

logical structure of the system. System diagrams are not intended to be 

comprehensive. For class diagrams in particular, we show only the most 

important high-level classes and relations. 



SYSTEM FUNCTIONS AND LOGICAL STRUCTURE

Major elements of the system's logical structure include representations of 

the ArcView, Analysis, Report Manager, Project Manager, and Hypermedia 

Reference subsystems. The following notation is used in diagrams that follow: 



   + for physical structure, arrows indicate compilation dependency 

     in classic Booch notation (we use them somewhat loosely to indicate

     control flow); 



   + for logical structure, open and filled circles indicate using and

     has-a relationships, respectively; open and filled squares indicate

     containment by reference and value, respectively; and arrows indicate

     inheritance.
  






ArcView 

Functional description.
ArcView is used as a navigation and 

information-display tool, and is the primary  access to other EMDS 

subsystems through modified versions of its application menus.







ArcView is also the system's primary database manager, maintaining and 

querying a data catalog that is used to identify file names, records, and 

fields, and their physical locations given data requirements for display or 

analysis.



Interface.
ArcView provides a GIS interface that allows the user to 

visually navigate to specific provinces and watersheds and display a variety 

of GIS themes (vegetation, streams, roads, topography, etc.) and other 

area-specific information (relevant regulations, existence and location of 

analyses, plans, etc.). 
Existing data for display consists of GIS data stored 

in Oracle's data sets on workstations and other, non-GIS data stored in 

Microsoft Access, or Borland Paradox, databases on PCs. In the first 

EMDS system prototype, database access is limited to local area networks.

Later implementations will provide for data access over a wide area 

network once the required communication infrastructure is in place in at 

least some sites in Region 6. Basic GIS capabilities as part of the primary 

user interface are provided to orient watershed analysis teams to the 

analysis situation.



Because ArcView 2.0 is already a full-featured, relatively mature GIS 

application, system design involves relatively little modification to the basic 

ArcView environment. 
The only significant change being introduced is to 

simplify the user interface by hiding application functions not directly 

related to requirements for overall system functionality. However, the 

modified environment will still provide optional access to the full set of 


ArcView features.


Data catalog.
USDA Forest Service Region 6 and the Siuslaw National 

Forest are now in the process of collecting and organizing descriptions of 

all Oracle workstation and PC databases (Microsoft Access and Borland 

Paradox) considered potentially useful to the EMDS system. ArcView 

functions as the system's central data server, handling requests from other 

subsystems (discussed further in the following sections),  querying 

appropriate databases, and passing the required data back to the client 

subsystem.



Communication.
All other subsystems can be invoked from ArcView 

menus. In addition, dependency network nodes from the Analysis 

subsystem can be selected for addition to ArcView's table of contents. 

Within ArcView, evaluated goal states for a selected node are displayed as 

a new theme, with access to relevant attributes also managed by ArcView. 



Analysis

Functional description.
The Analysis subsystem uses dependency 

networks to represent the current state of ecosystem processes and 

watershed properties and impacts of watershed-level activities on these 

processes and properties. It provides a link between the goals and 

objectives of managers and specific on-the-ground questions that need to 

be answered to adequately assess those objectives. Later implementations 

will expand the scope of analysis to include the province level, and provide 

for integration between the two levels of analysis. The Analysis subsystem
 

helps identify ecosystem processes requiring analysis, 

assists in selecting appropriate subsets of analyses, identifying and 

ranking data requirements, and evaluating ecosystem states, and 

provides information to the Project Manager and Report Manager 

subsystems for tracking progress of, and documenting, an analysis, 

respectively.



The Analysis subsystem includes a browsing facility to let the user navigate 

through the networks in a very intuitive way, so users can examine the 

structure of dependencies, and select appropriate goal nodes for inclusion 

in an analysis.

The user has the option of ignoring irrelevant concerns, in which case they 

are prompted to provide documentation for the rationale. Each goal node 

also provides five types of documentation:
 

1. authorities who were the source of information on how the node's 

   relations were defined,

2. literature citations,

3. explanatory information about the role of the node in the network 

   structure,

4. comments from the watershed analysis team, and

5. assumptions about ecosystem processes that are relevant to the 

   node.



Interface.
The interface for the analysis subsystem consists of a set of 

windows and other screen objects that allow the user to perform specific 

operations on the network hierarchy:



1. open network and navigate to (browse) any node in the hierarchy 

   by one of two or three methods,

2. load, run, and save an analysis profile,

3. edit (i.e. activate or de-activate links in the analysis profile, and/or 

   enter values for analysis profile objects intermediate between data 

   and the ultimate metadata node) items within the analysis profile,

4. view associated short descriptive or explanatory hypertext or 

   related passages in Hypermedia Reference subsystem documents,

5. check availability and timestamp of data for data links in the 

   knowledge net, 

6. tag anything associated with a data link or subgoal (files, hypertext, 

   etc.) for inclusion in an analysis report, and

7. fetch existing data.



Browsing and editing dependency network nodes, and running the 

NetWeaver engine to fetch data assists the user in identifying the 

availability of data, and prioritizing acquisition of missing data. The process 

of browsing and editing generates a project-specific analysis profile 

(knowledge base) via pruning:



1. User selects goals and subgoals that are related to concerns about 

   ecosystem processes.

2. The NetWeaver engine identifies all connected (related) higher and 

   lower level concerns.

3. User browses and prunes the dependency nets to ignore irrelevant 

   or uninteresting concerns, resulting in a final pruned dependency 

   net.
  



Communication.
The Analysis subsystem is called from ArcView by 

selecting the Analysis menu item. Most communication between the two 

subsystems involves requests from the Analysis subsystem to ArcView:



1. requested data to satisfy data requirements of knowledge base data 

   links, and

2. requests to display dependency network node states as new GIS 

   themes.



All dependency network nodes have a set of 5 text objects as attributes of 

the node. The text objects can contain hypertext references to material in 

the Hypermedia Reference subsystem.



The Analysis subsystem does not communicate directly with either the 

Project Manager or Report Manager subsystems. However, analysis 

profiles are available to both of these other subsystems as NetWeaver 

scripts.



Network navigation and editing.
Navigation through the networks, and 

editing links and node states, can be done in either text mode or graphic 

mode. Both modes display two child windows within the display area of 

the main Analysis application window:



1. The hierarchy window displays network topology as either an 

   indented hierarchical list of nodes in text mode or a graphical map 

   of nodes in graphic mode. 

2. The information window displays information about the currently 

   selected node in the hierarchy window, and contains screen objects 

   that provide additional information and control over some aspects 

   of the hierarchy display.
 

Running an analysis.
The run command in Analysis subsystem calls the 

NetWeaver DLL with instructions to evaluate the network. The full 

evaluation process involves the following steps:



1. Run command in Analysis initiates request to DLL to evaluate 

   network,

2. DLL traces active links among goal dependencies,

3. DLL identifies unsatisfied data links and passes data requirements 

   to Analysis, 

4. if necessary, Analysis issues data request to ArcView,

5. ArcView interrogates data catalog for source of data, and uses 

   SQL calls to appropriate databases,

6. ArcView returns data values to Analysis,

7. Analysis returns data values to NetWeaver DLL,

8. DLL updates data links,

9. DLL re-evaluates network and returns updated states of nodes to 

   Analysis.



Checking availability and timestamp of data. On completing a run, the 

user can view the status of an analysis. A status window displays a table 

with a record for each goal node and data link. A record includes:



1. node name

2. percent of data requirements met

3. priority (relative contribution of this node’s missing data to total 

   missing data)

4. total number of antecedents

5. number of immediate antecedents

6. total number of dependents

7. number of immediate dependents

8. date of most recent data source

9. date of oldest data source.



Some of the functions available for the status window are:



1. select record,

2. view the list of values associated with selected record (note that, in 

   general, a data link processes a list of values, so each goal also has 

   a list of states or outcomes),

3. sort records by priority of missing data, and 

4. view only records with missing data.



Project Manager

Functional description.
Through its Project Manager subsystem, the 

system provides a project summary of particular interest to line officers. 

The core functions of the Project Manager subsystem are provided by a 

higher-level dependency network that polls the set of dependency networks 

generated by the Analysis subsystem. Its function is to assist with planning 

and monitoring data acquisition, and tracking and documenting the 

progress of an analysis by using analysis profiles generated by the 

dependency networks.



The first prototype implementation will be limited to summary reporting 

features. However, these should be valuable to both line officers and 

watershed analysis teams. In the second or later implementation, we will 

add tools to assist with scheduling of expenditures of money and personnel 

time, so that data collection projects can be timed and coordinated to make 

the best use of operational resources. This type of operational tracking is 

essential to creating informative and credible reports, but this full 

implementation is not feasible for the expected delivery schedule of the first 

prototype.



Communication.
Access to the Project Manager subsystem is provided 

through a menu item in ArcView. Optionally, while viewing the analysis 

summary, the user can invoke the Analysis subsystem in view mode for a 

more detailed examination of data requirements.



The Hypermedia Reference subsystem can be called from a menu item on 

the Project Manager menu bar. 



The Project Manager does not communicate directly with the Report 

Manager, but its output files are available to the Report Manager via the 

tag list manager, described later.



The summary reports:

1. trueness value associated with each top-level goal that has been 

   selected for analysis, and

2. a synthesis of data requirements from all of the separate nets.



Hypermedia Reference

Functional description.
The Hypermedia Reference subsystem provides 

powerful navigation tools for accessing information in the FEMAT Report, 

the Record of Decision, and Standards and Guides. The initial content of 

the subsystem could easily be expanded to include the Watershed Analysis 

Handbook now under development by the Region 6 Watershed Analysis 

Coordination Team. Legal references such as text of NEPA and NFMA 

can also be added in the second implementation, if that is desirable. 



Communication.
Direct access to this subsystem is provided through:



1. menus of other subsystems, and

2. embedded hypertext in hypermedia objects managed by other 

   subsystems.



In the first prototype, direct access to the Hypermedia Reference 

subsystem is provided through menus in all other subsystems. In the second 

prototype, we plan to provide constrained, intelligent search capabilities 

within the Analysis subsystem to access more specific information in the 

FEMAT report, Record of Decision, Standards and Guides, based on a 

user's current location in the dependency networks.



Report Manager

Functional description.
The Report Manager for the first prototype will 

most likely be implemented in KnowledgePro, using a Microsoft Word 

DLL currently in production. In later prototypes, we also plan to introduce 

additional knowledge-based functionality that provides a template for the 

7-step watershed analysis process. 



The Report Manager subsystem assists with assembling numerous pieces of 

information into a coherent report. 

The Report Manager handles display of project history and status, and 

other details of an analysis, and provides the user with tools to assemble 

reports of an analysis, which may include maps, information from the 

project log, status and project management files, as well as hypertext 

excerpts from items in the Hypermedia Reference subsystem.



The system includes a globally accessible tag utility that can optionally be 

used to add the contents of any system display window to the tag list. A 

key tool in the Report Manager that complements the tag utility is a tag list 

manager that is used to browse among, and select items from, a list of 

items for inclusion in a report.


Communication.
The ArcView subsystem provides direct access to the 

Report Manager through a menu item. Documents from the Hypermedia 

Reference subsystem can be easily incorporated into the report to provide 

expanded explanation for the analysis process, or accessed as an on-line 

reference by a watershed analysis team as an aid to conducting an analysis. 

Other subsystems do not communicate directly with the Report Manager, 

but many of their display window contents and all output files are available 

to the Report Manager via the tag list manager discussed below.



Contents of a watershed analysis report.
Watershed analysis reports include 

both text and graphics. Graphics include:



1. maps generated by ArcView, 

2. graphic images from the Hypermedia Reference subsystem, and 

3. possibly other graphic images.

4. Text sources for reports include: 

5. hypertext from the Hypermedia Reference subsystem,

6. text from the annotation function group in the Analysis subsystem, 

7. individual analysis profiles,

8. analysis status reports from the Analysis subsystem, and

9. project status from the Project Manager subsystem.



Tag utility.
Tagging information is performed by a utility that is accessible 

from all subsystems. The function of the tag utility is to mark any 

information in a display window of any subsystem for possible inclusion in 

an analysis report (Figure 4). The most general approach for implementing 

the tag function requires two files:



1. a tag data file that stores both data and pointers to data, and 

2. a tag list file that contains pointers to tag data file records and other 

   descriptive information about a tagged item.



A tag list file record contains:



1. a pointer to a tag data file record,

2. brief description for identifying content,

3. the time stamp of the source data, and

4. a time stamp indicating when the information was added to the tag 

   file.
 

Tag list manager utility.
The tag list manager utility is used to browse, 

review, arrange, and preview information that has been assembled by the 

tag utility. Functions of the tag list manager are:



1. display contents of the tag list file,

2. check time stamp of tag list items against data source,

3. view selected tag list item,

4. arrange order of tag list items,

5. insert selected item(s) at current location in report document, and

6. insert complete contents of the tag file in report document 

   according to tag list order.



The check-time-stamp function:



1. can be invoked by the user to perform a comprehensive check of all 

   items in the tag file (reports all items that may need updating), and 

2. also executes automatically whenever the insert functions are 

   executed unless the user has turned off autochecking.



PRESENT STATUS AND FUTURE POSSIBILITIES
 

Concepts of ecosystem integrity and sustainability are closely intertwined 

with human values (Kay 1993). A basic premise of ecosystem management 

is that combined application of knowledge and technology can be used to 

promote desirable ecosystem conditions that benefit both the environment 

and social and economic systems (Salwasser 1994). Bormann et al. (1993), 

describing a framework for management of sustainable ecosystems, have 

defined ecological sustainability as the intersection of societal values and 

ecological capacity. Inherent in each of these views is the notion that 

ecosystem sustainability cannot be adequately addressed apart from societal 

values.



The DSS for watershed analysis that has been described clearly 

encompasses only a few aspects of a full ecosystem management process 

(Bormann et al. 1994); it provides a knowledge-based information 

management system that assists with data acquisition and evaluation of 

ecological integrity.  However, the scope and complexity of an ecosystem 

approach to natural resource management (Franklin 1994) highlight the 

basic need to acquire, access, operate on, and manage very large quantities 

of very diverse information. In this respect, the DSS provides an essential 

foundation upon which to build a more comprehensive decision support 

system for ecosystem management. Also, insofar as the hypermedia 

capabilities of the system enhance the ability of managers to effectively 

communicate the rationale behind an analysis process and its outcomes, the 

DSS can make a useful contribution to the dialog that is essential to the 

adaptive management process of reconciling societal wants with ecological 

reality. 



With object-oriented system development, we envision short development 

cycles in which incremental enhancements to the original prototype are 

introduced at about 6-month intervals. A likely near-term enhancement is 

the integration of groupware technologies that further enhance such dialog 

(Fox, TERRA Lab, Fort Collins, personal communication). In discussing 

the Report Manager subsystem, we mentioned incorporating 

knowledge-based support for the full watershed analysis process described 

in the Watershed Analysis Guidebook, now being developed by Forest 

Service Region 6. The Guidebook itself will likely be added to the 

Hypermedia Reference subsystem. 



In the longer term, we envision building upon the basic information 

management foundation needed to support watershed analysis to eventually 

provide knowledge-based decision support for the full adaptive 

management process. A logical first step in this longer term development, 

which also completes development of the watershed analysis component, is 

integration of a system for process models used to consider alternative 

future management scenarios (Leavesley, USGS, Boulder, personal 

communication). Numerous other on-going development efforts may 

eventually be integrated into the foundation system. In such an 

evolutionary process, it is useful to keep in mind Booch's (1994) advice: 

the analysis and design of complex systems require a clear architectural 

vision to sustain long-term development.



ACKNOWLEDGMENTS

The authors thank Mark Twery (Northeastern Forest Experiment Station, 

Morgantown, WV), Doug Fox (Rocky Mountain Forest and Range 

Experiment Station, Fort Collins, CO), and Mike Rauscher (Southern 

Forest Experiment Station, Asheville, NC) for their review of the 

manuscript. We also thank the following individuals, who have been 

instrumental in construction of the dependency networks: Kelly Burnett, 

Marty Raphael, and Bernard Bormann (Pacific Northwest Research 

Station, Portland, OR); Jon Martin, Ginnie Grilley, Cindy McCain, and Bob 

Metzger (Siuslaw National Forest, Corvallis, OR); Bruce McCammon 

(USDA Forest Service, Region 6, Portland, OR); Todd Parker (Mount 

Hood National Forest, Gresham, OR); and Dick Holthausen (USDA Forest 

Service, Washington Office).



ENDNOTES

1. A federal agency guide for pilot watershed analysis, Version 1.0. 

   February 1993.



2. The use of trade or firm names in this publication is for reader 

   information and does not imply endorsement by the U.S. Department 

   of Agriculture of any product or service.



REFERENCES

Booch, G. 1994. Object-oriented analysis and design with applications. 

Benjamin/Cummings, New York. 589 p.



Bormann, B.T., M.H. Brookes, E.D Ford, A.R. Kiester, C.D. Oliver, J.F. 

Weigand. 1993. A framework for sustainable-ecosystem management. Gen. 

Tech. Rep. PNW-GTR-331. Portland, OR, U.S. Department of 

Agriculture, Forest Service, Pacific Northwest Research Station. 61 p.



Bormann, B.T., P.G. Cunningham, M.H. Brookes, V.W. Manning, and 

M.W. Collopy. 1994. Adaptive ecosystem management in the Pacific 

Northwest Gen. Tech. Rep. PNW-GTR-341. Portland, OR, U.S. 

Department of Agriculture, Forest Service, Pacific Northwest Research 

Station. 22 p. 



Caldwell, L.K., C.F. Wilkinson, and M.A. Shannon. 1994. Making 

ecosystem policy: three decades of change. J. For. 92:7-10.



Forest Ecosystem Management Assessment Team. 1993. Forest ecosystem 

management: an ecological, economic, and social assessment. Report of the 

Forest Ecosystem Management Assessment Team. U.S. Gov. Print. Off., 

Washington, DC.



Franklin, J.F. 1994. Ecological science, a conceptual basis for FEMAT. J. 

For. 92:21-23.



Johnson, K.N., J.F. Franklin, J.W. Thomas, and J. Gordon. 1991. 

Alternatives for management of late-successional forests of the Pacific 

Northwest. A report to the Agriculture Committee and the Merchant 

Marine Committee of the US House of Representatives. Coll. For., Oreg. 

State Univ., Corvallis. 59 p.



Kay, J.J. 1993. On the nature of ecological integrity. Pages 201-212 in: 

Ecological integrity and the management of ecosystems (S. Woodley, J. 

Kay, and G. Francis, ed.). St. Lucie Press. 220 p. 



Nash, B.L., M.C. Saunders, B.J. Miller, C.A. Bloom, D.D. Davis, and J.M. 

Skelly, 1992.  Forest health, an expert advisory system for assessing foliar 

and crown health of selected northern hardwoods.  Can. J. For. Res.22: 

1770-1775. 



Rumbaugh, J., M. Blaha, W. Premerlani, F. Eddy, and W. Lorensen. 1991. 

Object-oriented modeling and design. Prentice Hall, Englewood Cliffs, 

New Jersey.



Salwasser, H. 1994. Ecosystem management: can it sustain diversity and 

productivity. J. For. 92:6-11. 



Saunders, M.C., R.N. Coulson, J. Folse. 1990. Applications of artificial 

intelligence in agriculture and natural resource management.  pp. 1-14 in A. 

Kent and J. Willeams (eds) Encyclopedia of Computer Science and 

Technology. Volume 25, supplement 10. Marcel Dekker Inc., New York. 

405pp.



Saunders, M. C., E. G. Rajotte, and J. W. Travis. 1989. The use and 

significance of expert systems in small fruit IPM. In N.J. Bostanian, L.T. 

Wilson, and T. J. Dennehy [eds], Monitoring and Integrated Management 

of Arthropod Pests of Small Fruit Crops.  Intercept, London.  301pp.



Stone, N.D., R.N. Coulson, R.E. Frisbie, and D.K. Loh. 1986. Expert 

systems in entomology: three approaches to problem solving.  Bull. 

Entomol. Soc. Amer.  32:161-66.



Thomas, J.W. 1994. Forest Ecosystem Management Assessment Team: 

objectives, process and options. J. For. 92:12-19.



Thomas, J.W., E.D. Forsman, J.B. Lint, E.C. Meslow, B.R. Noon, and J. 

Verner. 1990. A conservation strategy for the northern spotted owl: a 

report of the Interagency Scientific Committee to address the conservation 

of the northern spotted owl. USDA For. Serv., US Bur. Land Manage., 

USDI Fish and Wildl. Serv., Natl. Park Serv., Portland, OR. 427 p.



Thomas, J.W., et al. 1993. Viability assessments and management 

considerations for species associated with late-successional and old-growth 

forests of the Pacific Northwest. USDA For. Serv., Washington, DC. 530 p. 



USDA Forest Service 1992. Taking an ecological approach to 

management. Proceedings of  a National Workshop. Salt Lake City, April 

27-30, 1992. For. Serv. WO-WSA-3. 241 p.



USDA Forest Service and USDI Bureau of Land Management. 1994. 

Record of decision for amendments for Forest Service and Bureau of Land 

Management planning documents within the range of the northern spotted 

owl. USDA For. Serv. and USDI Bur. Land Manage. 74 p.