12.4 Decision Support Systems Defined
pacity of an ecosystem is fixed, yet key stakeholders all want to extract a product from that ecosystem at a higher level, a compromise midway between the levels will be unsustainable.
The next major subsystem, spatial and nonspatial data management, organizes the available descriptions of the ecological and management components of ecosystem management. Data must be
available to support choices among alternative
management scenarios and to forecast the consequences of management activities on the landscape.
There is a tension between the increasing number
of goals that decision makers and stakeholders
value and the high cost of obtaining data and
understanding relationships that support these
choices. Monitoring both natural and anthropogenic disturbance activities and disturbance-free
dynamics of managed forest ecosystems are also
extremely important if an EM-DSS is to accurately
portray the decision choices and their consequences. Barring blind luck, the quality of the decision cannot be better than the quality of the
knowledge behind it. Poor data can lead to poor decisions. It is difficult to conceive of prudent ecosystem management without an adequate biophysical
description of the property in question.
The next four subsystems, knowledge-based,
simulation model, help-hypertext, and data visualization management, deal with effectively managing knowledge in the many diverse forms in which
it is stored, represented, or coded (see Rauscher et
aI., 1993, for more detail). Knowledge that is not
language based is either privately held in people's
minds or publicly represented as photographs,
video, or graphic art. Language-based knowledge
is found in natural language texts of various kinds,
in mathematical simulation models, and in expert
or knowledge-based systems. Data visualization
software has been developed that can manipulate
photographic, video, and graphic art representations of current and future ecosystem conditions.
Data visualization software is beginning to be incorporated into EM-DSS on a routine basis to help
decision makers see for themselves the likely impact of their decisions on the landscape.
In the last 20 years, an impressive amount of
mathematical simulation software has been developed for all aspects of natural resource management. Schuster et al. (1993) conducted a comprehensive inventory of simulation models available
to support forest planning and ecosystem management. They identified and briefly described 250
software tools. Jorgensen et al. (1996) produced another compendium of ecological models that in171
corporate an impressive amount of ecosystem theory and data. The simulation model management
subsystem of the EM-DSS is designed to provide
a consistent framework into which models of many
different origins and styles can be placed so that
decision makers can use them to analyze, forecast,
and understand elements of the decision process.
Despite our most strenuous efforts to quantify
important ecological processes to support a theory
in simulation model form, by far the larger body of
what we know can only be expressed qualitatively,
comparatively, and inexactly. Most often this qualitative knowledge has been organized over long
years of professional practice by human experts.
Theoretical and practical advances in the field of
artificial intelligence applications in the last 20
years now allow us to capture some of this qualitative, experience-based expertise into computer
programs called expert or knowledge-based systems (Schmoldt and Rauscher, 1996). It is still not
possible to capture the full range and flexibility of
knowledge and reasoning ability of human experts
in knowledge-based software. We have learned,
however, how to capture and use that portion of expertise that the human expert considers routine. The
knowledge management subsystem of the EM-DSS
is designed to organize all available knowledgebased models in a uniform framework to support
the decision-making process.
Finally, a large amount of text material exists
that increases the decision maker's level of understanding about the operation of the decision support system itself, the meaning of results from the
various modeling tools, and the scientific basis for
the theories used. This text material is best organized in hypertext software systems. Hypermedia
methodology supports a high degree of knowledge
synthesis and integration with essentially unlimited
expandability. The oak regeneration hypertext
(Rauscher et aI., 1997b) and the hypermedia reference system to the FEMAT report (Reynolds et aI.,
1995) are recent examples of the use of hypertext
to synthesize and organize scientific subject matter. Examples of the use of hypertext to teach and
explain software usage can be found in the help
system of any modem commercial computer program.
The software subsystems of an EM-DSS described so far help decision makers to organize the
decision problem, formulate alternatives, and analyze their future consequences. The decision methods management subsystem (Figure 12.4) provides
tools and guidance for choosing among the alternatives, for performing sensitivity analysis to iden-
pacity of an ecosystem is fixed, yet key stakeholders all want to extract a product from that ecosystem at a higher level, a compromise midway between the levels will be unsustainable.
The next major subsystem, spatial and nonspatial data management, organizes the available descriptions of the ecological and management components of ecosystem management. Data must be
available to support choices among alternative
management scenarios and to forecast the consequences of management activities on the landscape.
There is a tension between the increasing number
of goals that decision makers and stakeholders
value and the high cost of obtaining data and
understanding relationships that support these
choices. Monitoring both natural and anthropogenic disturbance activities and disturbance-free
dynamics of managed forest ecosystems are also
extremely important if an EM-DSS is to accurately
portray the decision choices and their consequences. Barring blind luck, the quality of the decision cannot be better than the quality of the
knowledge behind it. Poor data can lead to poor decisions. It is difficult to conceive of prudent ecosystem management without an adequate biophysical
description of the property in question.
The next four subsystems, knowledge-based,
simulation model, help-hypertext, and data visualization management, deal with effectively managing knowledge in the many diverse forms in which
it is stored, represented, or coded (see Rauscher et
aI., 1993, for more detail). Knowledge that is not
language based is either privately held in people's
minds or publicly represented as photographs,
video, or graphic art. Language-based knowledge
is found in natural language texts of various kinds,
in mathematical simulation models, and in expert
or knowledge-based systems. Data visualization
software has been developed that can manipulate
photographic, video, and graphic art representations of current and future ecosystem conditions.
Data visualization software is beginning to be incorporated into EM-DSS on a routine basis to help
decision makers see for themselves the likely impact of their decisions on the landscape.
In the last 20 years, an impressive amount of
mathematical simulation software has been developed for all aspects of natural resource management. Schuster et al. (1993) conducted a comprehensive inventory of simulation models available
to support forest planning and ecosystem management. They identified and briefly described 250
software tools. Jorgensen et al. (1996) produced another compendium of ecological models that in171
corporate an impressive amount of ecosystem theory and data. The simulation model management
subsystem of the EM-DSS is designed to provide
a consistent framework into which models of many
different origins and styles can be placed so that
decision makers can use them to analyze, forecast,
and understand elements of the decision process.
Despite our most strenuous efforts to quantify
important ecological processes to support a theory
in simulation model form, by far the larger body of
what we know can only be expressed qualitatively,
comparatively, and inexactly. Most often this qualitative knowledge has been organized over long
years of professional practice by human experts.
Theoretical and practical advances in the field of
artificial intelligence applications in the last 20
years now allow us to capture some of this qualitative, experience-based expertise into computer
programs called expert or knowledge-based systems (Schmoldt and Rauscher, 1996). It is still not
possible to capture the full range and flexibility of
knowledge and reasoning ability of human experts
in knowledge-based software. We have learned,
however, how to capture and use that portion of expertise that the human expert considers routine. The
knowledge management subsystem of the EM-DSS
is designed to organize all available knowledgebased models in a uniform framework to support
the decision-making process.
Finally, a large amount of text material exists
that increases the decision maker's level of understanding about the operation of the decision support system itself, the meaning of results from the
various modeling tools, and the scientific basis for
the theories used. This text material is best organized in hypertext software systems. Hypermedia
methodology supports a high degree of knowledge
synthesis and integration with essentially unlimited
expandability. The oak regeneration hypertext
(Rauscher et aI., 1997b) and the hypermedia reference system to the FEMAT report (Reynolds et aI.,
1995) are recent examples of the use of hypertext
to synthesize and organize scientific subject matter. Examples of the use of hypertext to teach and
explain software usage can be found in the help
system of any modem commercial computer program.
The software subsystems of an EM-DSS described so far help decision makers to organize the
decision problem, formulate alternatives, and analyze their future consequences. The decision methods management subsystem (Figure 12.4) provides
tools and guidance for choosing among the alternatives, for performing sensitivity analysis to iden-