circumstances? In dealing with this issue, it is helpful to separate the actual
extrapolation from the predictive outputs of the model. For example, some
hydrodynamic models can be transferred and used in different locations.
However, the predictive utility will be a function of the site-specific data
collected to set boundary conditions and to parameterize the model.
Similarly, for conceptual models, the structure and processes, both physical
and ecological, may be generalized, but ultimately the utility will be a function of how well site-specific parameters are represented, particularly the
exposure pathway and process components. For example, an ecological
effects submodel dealing with nutrient effects on seagrasses may be highly
generalizeable and applicable to several sites where seagrasses exist. Thus,
the successful extrapolation of quantitative models relates to the ability of
the model adequately to represent the problem setting and the steps used
in model calibration and validation to the specific issue at hand. In addition, extrapolation may be dependent on how successful the model has been
when used in other cases. If the model has been widely used for a class of
problems (e.g., hydrodynamics), confidence is enhanced that the extrapolation issues have been addressed. If the model has not used for the types
of problems of concern, then an increased burden is placed on demonstrating the applicability of the model.
5.2.4 Accuracy and Precision
How well does the model meet the decision-making needs with respect
to accuracy and precision? Different decisions demand differing degrees
of accuracy and precision. For example, selecting among different management options may simply involve ranking the risks of one option over
another, and that relative assessment is sufficient without a high level of
precision. In other cases, a comparison among options may require quantitative assessments with high precision and accuracy. For example, a
water quality model used to predict the discharge concentrations of toxic
pollutants to comply with water quality criteria requires a high degree of
accuracy and precision. The issue is (1) to determine, a priori, the required
level of accuracy and precision for the particular decisions to be made and
(2) to assure that the models selected will meet the criteria. The key here
is that each decision has its own needs relative to model accuracy and
precision and models must be tailored to meet those goals and needs.
5.2.5 Goals, Endpoints, and Benchmarks
How well do conceptual and simulation models relate to the specific
ecological goals, associated ecological endpoints, and target benchmarks?
As discussed above, the identification of ecological goals is necessary
if appropriate decision making is to result. That is, ecological goals are
the articulation of societal interests in the environment, and ecological
5. Overcoming Barriers to the Use of Models
97
extrapolation from the predictive outputs of the model. For example, some
hydrodynamic models can be transferred and used in different locations.
However, the predictive utility will be a function of the site-specific data
collected to set boundary conditions and to parameterize the model.
Similarly, for conceptual models, the structure and processes, both physical
and ecological, may be generalized, but ultimately the utility will be a function of how well site-specific parameters are represented, particularly the
exposure pathway and process components. For example, an ecological
effects submodel dealing with nutrient effects on seagrasses may be highly
generalizeable and applicable to several sites where seagrasses exist. Thus,
the successful extrapolation of quantitative models relates to the ability of
the model adequately to represent the problem setting and the steps used
in model calibration and validation to the specific issue at hand. In addition, extrapolation may be dependent on how successful the model has been
when used in other cases. If the model has been widely used for a class of
problems (e.g., hydrodynamics), confidence is enhanced that the extrapolation issues have been addressed. If the model has not used for the types
of problems of concern, then an increased burden is placed on demonstrating the applicability of the model.
5.2.4 Accuracy and Precision
How well does the model meet the decision-making needs with respect
to accuracy and precision? Different decisions demand differing degrees
of accuracy and precision. For example, selecting among different management options may simply involve ranking the risks of one option over
another, and that relative assessment is sufficient without a high level of
precision. In other cases, a comparison among options may require quantitative assessments with high precision and accuracy. For example, a
water quality model used to predict the discharge concentrations of toxic
pollutants to comply with water quality criteria requires a high degree of
accuracy and precision. The issue is (1) to determine, a priori, the required
level of accuracy and precision for the particular decisions to be made and
(2) to assure that the models selected will meet the criteria. The key here
is that each decision has its own needs relative to model accuracy and
precision and models must be tailored to meet those goals and needs.
5.2.5 Goals, Endpoints, and Benchmarks
How well do conceptual and simulation models relate to the specific
ecological goals, associated ecological endpoints, and target benchmarks?
As discussed above, the identification of ecological goals is necessary
if appropriate decision making is to result. That is, ecological goals are
the articulation of societal interests in the environment, and ecological
5. Overcoming Barriers to the Use of Models
97