risks (e.g., gene flow to native plants and alteration of soil arthropod communities) posed by the use of genetically engineered crops. This framework
indicates that the ecological model (selected from the toolkit) has become
an integrated component in the decision model and contributes information concerning possible risks in conventional croplands and croplands containing genetically modified organisms in relation to decisions concerning
the use of fertilizers, the application of pesticides, and the percentage of
crops to be planted with genetically modified plants. The ecological model
also provides input to a risk-based monitoring program as an integral part
of the overall crop management model. As illustrated by this example, it
may prove efficient and effective to work backwards from the set of
resource management challenges faced by the agencies and decisionmaking processes in designing, building, and filling the toolkit.
11.5 Data Management
A toolkit that will effectively support the use of models in environmental
decision making will necessarily include capabilities for data handling.
Models often require large amounts of data. Data are used in the processes
of (1) model development, (2) model implementation, (3) definition of
initial conditions, (4) estimation of model parameter values, and (5) model
verification and evaluation. The design and construction of the toolkit will
have to successfully address these data issues in relation to environmental
and ecological modeling. The toolkit will need the capability to operationally link models to complex, often spatially explicit data sets. As data
requirements of increasingly complex environmental models expand, the
toolkit may need the ability to perform sophisticated interpolations to fill
in missing values in constructing model-input data files. Additionally, the
toolkit should facilitate the ability to check for data errors (e.g., likely erroneous values) and either correct errors or propagate the error estimates
through the model calculations (e.g., a Monte Carlo simulation).
Models can also produce large amounts of results that will require similar
data-handling capabilities to extract information that will be meaningful
in resource management and decision making (mason and Gurney 1993).
The toolkit will necessarily address data analysis and postprocessing
issues that include (1) higher-dimensional data handling and (2) data
visualization. For example, current GIS methods are generally constrained
to the analysis and presentation of two-dimensional data or model results.
However, many environmental resource challenges involve additional
dimensions, including the vertical and time dimensions. Data visualization methods will be necessary for displaying the four-dimensional
output (i.e., three-dimensional results that vary through time). Ecological animation that uses the virtual reality markup language might
provide one of the necessary capabilities in advanced data visualiza218
Steven M. Bartell
indicates that the ecological model (selected from the toolkit) has become
an integrated component in the decision model and contributes information concerning possible risks in conventional croplands and croplands containing genetically modified organisms in relation to decisions concerning
the use of fertilizers, the application of pesticides, and the percentage of
crops to be planted with genetically modified plants. The ecological model
also provides input to a risk-based monitoring program as an integral part
of the overall crop management model. As illustrated by this example, it
may prove efficient and effective to work backwards from the set of
resource management challenges faced by the agencies and decisionmaking processes in designing, building, and filling the toolkit.
11.5 Data Management
A toolkit that will effectively support the use of models in environmental
decision making will necessarily include capabilities for data handling.
Models often require large amounts of data. Data are used in the processes
of (1) model development, (2) model implementation, (3) definition of
initial conditions, (4) estimation of model parameter values, and (5) model
verification and evaluation. The design and construction of the toolkit will
have to successfully address these data issues in relation to environmental
and ecological modeling. The toolkit will need the capability to operationally link models to complex, often spatially explicit data sets. As data
requirements of increasingly complex environmental models expand, the
toolkit may need the ability to perform sophisticated interpolations to fill
in missing values in constructing model-input data files. Additionally, the
toolkit should facilitate the ability to check for data errors (e.g., likely erroneous values) and either correct errors or propagate the error estimates
through the model calculations (e.g., a Monte Carlo simulation).
Models can also produce large amounts of results that will require similar
data-handling capabilities to extract information that will be meaningful
in resource management and decision making (mason and Gurney 1993).
The toolkit will necessarily address data analysis and postprocessing
issues that include (1) higher-dimensional data handling and (2) data
visualization. For example, current GIS methods are generally constrained
to the analysis and presentation of two-dimensional data or model results.
However, many environmental resource challenges involve additional
dimensions, including the vertical and time dimensions. Data visualization methods will be necessary for displaying the four-dimensional
output (i.e., three-dimensional results that vary through time). Ecological animation that uses the virtual reality markup language might
provide one of the necessary capabilities in advanced data visualiza218
Steven M. Bartell