11.3.1 Model Makers
Model makers (and users) desire models that accurately and reliably
predict the occurrence or nonoccurrence of a particular environmental
impact [e.g., Burns (1986)]. Model makers strive to avoid the dissemination
of models that might fail to forecast an actual impact (i.e., model maker’s
ruin). One problem model makers face is that of providing the management
community with models that cannot be validated for all conceivable applications. It remains neither possible nor desirable to completely specify an
ecological system in the form of a model. Therefore, making simplifying
assumptions is an unavoidable aspect of the modeling process. Because of
such simplifications, all ecological and environmental models are invalid
descriptions at some level of structural and functional detail. Model validity has also been addressed from the perspective of comparing model predictions with available data. Because the results of all possible future
model–data comparisons cannot be evaluated, ecological and environmental models can never be “validated” from this perspective. Under these
kinds of constraints, the model maker must determine the conditions (e.g.,
initial conditions, regions of model parameter space, and environmental
forcing functions) where the model has been demonstrated to perform with
sufficient accuracy and precision to support correct and reliable decision
making. These conditions define a domain of applicability for the model
(Figure 11.3). The model maker works to increase the domain of applicability through continued data collection, model testing, and model refinement. From the viewpoint of the model maker, all models are by definition
invalid; however, some are useful (Mankin et al. 1975).
11. Effective Use of Ecological Modeling: The Toolkit Concept
213
True
False
True
False
Model accurately predicts
that an environmental effect
will occur
Model accurately predicts
that no effect will occur
Model fails to predict an
environmental effect –
Model maker's ruin
Model predicts an environmental effect when none
actually occurs –
Model user's ruin
Adverse impact predicted
Adverse impact observed
Figure 11.2. Perspectives concerning the development and use of modeling tools
in environmental decision making.
Model makers (and users) desire models that accurately and reliably
predict the occurrence or nonoccurrence of a particular environmental
impact [e.g., Burns (1986)]. Model makers strive to avoid the dissemination
of models that might fail to forecast an actual impact (i.e., model maker’s
ruin). One problem model makers face is that of providing the management
community with models that cannot be validated for all conceivable applications. It remains neither possible nor desirable to completely specify an
ecological system in the form of a model. Therefore, making simplifying
assumptions is an unavoidable aspect of the modeling process. Because of
such simplifications, all ecological and environmental models are invalid
descriptions at some level of structural and functional detail. Model validity has also been addressed from the perspective of comparing model predictions with available data. Because the results of all possible future
model–data comparisons cannot be evaluated, ecological and environmental models can never be “validated” from this perspective. Under these
kinds of constraints, the model maker must determine the conditions (e.g.,
initial conditions, regions of model parameter space, and environmental
forcing functions) where the model has been demonstrated to perform with
sufficient accuracy and precision to support correct and reliable decision
making. These conditions define a domain of applicability for the model
(Figure 11.3). The model maker works to increase the domain of applicability through continued data collection, model testing, and model refinement. From the viewpoint of the model maker, all models are by definition
invalid; however, some are useful (Mankin et al. 1975).
11. Effective Use of Ecological Modeling: The Toolkit Concept
213
True
False
True
False
Model accurately predicts
that an environmental effect
will occur
Model accurately predicts
that no effect will occur
Model fails to predict an
environmental effect –
Model maker's ruin
Model predicts an environmental effect when none
actually occurs –
Model user's ruin
Adverse impact predicted
Adverse impact observed
Figure 11.2. Perspectives concerning the development and use of modeling tools
in environmental decision making.