There may be, no doubt will be errors and omissions in that approach, and thus, they
may worry about criticism. Therefore, the rules of interaction must recognize gently
their courage. The process does promise to make young scientists wiser and older
scientists younger.
A general strategy for modeling with experts is to build a STELLA model of the
phenomena that all expect will be needed to answer the questions posed by them at
the outset. That model should cover a space and have a time step that is commensurate with the detail needed for the questions. The whole space and time needed for
the model may exceed reasonable use of the desktop computer. Eventually, any
modeling enterprise may become so large that the program STELLA is too cumbersome to use. However, translators exist for the final STELLA model, converting
its equations into other computer code, such as C
+ or FORTRAN. Translated
models can be duplicated and put into parallel modes, adjoining cells on a landscape, for example and run simultaneously on large computers. For example, in
spatial ecological modeling we use STELLA to capture the expertise of a variety of
life science professionals. We then electronically translate that generic model into
C
+ or FORTRAN and apply it to a series of connected cells, for example as many as
120,000 in a model of the Sage Grouse [7]. The next step is to electronically
initialize these now cellularized models with a specific Geographic Information
System of parameters and initial conditions maps. We then run the cellularized
combine on a large parallel-processing computer or a large network of paralleled
workstations. In this way, the knowledge-capturing features of STELLA can be
seamlessly connected to the world’s most powerful computers.
The cellular or parallel approach to building dynamic spatial models with
STELLA and running those models on ever more powerful computers is receiving
increasing attention in landscape ecology and environmental management. An
alternative, but closely related approach has been chosen by Ruth and Pieper [8]
in their model of the spatial dynamics of sea level rise. The model consists of a
relatively small set of interconnected cells, describing the physical processes of
erosion and sediment transport. Each cell of the model is initialized with sitespecific data. These cells are then moved across the landscape to create a mosaic of
the entire area to be covered by the model. In its use of an iconographic programming language, its visual elements for data representation and its representation of
system dynamics the model are closely related to pictorial simulation models [9]
and cellular automata models [10]. The approach is flexible, computationally
efficient, and typically does not require parallel-processing capabilities. Though
slightly awkward, it is also possible to use STELLA to carry out object-oriented
models.
It is the intention of this book to teach you how to model, not just how to use
models. We have chosen STELLA toward this end because it is a very powerful
tool for building dynamic models. The software also comes with an “Authoring”
version that enables you to develop models for use by others who may be
uninterested in the underlying structure of the model. However, since model
development and understanding are the purpose of this book, we do focus here on
the modeling process itself.
1.9 Extending the Modeling Approach
27
may worry about criticism. Therefore, the rules of interaction must recognize gently
their courage. The process does promise to make young scientists wiser and older
scientists younger.
A general strategy for modeling with experts is to build a STELLA model of the
phenomena that all expect will be needed to answer the questions posed by them at
the outset. That model should cover a space and have a time step that is commensurate with the detail needed for the questions. The whole space and time needed for
the model may exceed reasonable use of the desktop computer. Eventually, any
modeling enterprise may become so large that the program STELLA is too cumbersome to use. However, translators exist for the final STELLA model, converting
its equations into other computer code, such as C
+ or FORTRAN. Translated
models can be duplicated and put into parallel modes, adjoining cells on a landscape, for example and run simultaneously on large computers. For example, in
spatial ecological modeling we use STELLA to capture the expertise of a variety of
life science professionals. We then electronically translate that generic model into
C
+ or FORTRAN and apply it to a series of connected cells, for example as many as
120,000 in a model of the Sage Grouse [7]. The next step is to electronically
initialize these now cellularized models with a specific Geographic Information
System of parameters and initial conditions maps. We then run the cellularized
combine on a large parallel-processing computer or a large network of paralleled
workstations. In this way, the knowledge-capturing features of STELLA can be
seamlessly connected to the world’s most powerful computers.
The cellular or parallel approach to building dynamic spatial models with
STELLA and running those models on ever more powerful computers is receiving
increasing attention in landscape ecology and environmental management. An
alternative, but closely related approach has been chosen by Ruth and Pieper [8]
in their model of the spatial dynamics of sea level rise. The model consists of a
relatively small set of interconnected cells, describing the physical processes of
erosion and sediment transport. Each cell of the model is initialized with sitespecific data. These cells are then moved across the landscape to create a mosaic of
the entire area to be covered by the model. In its use of an iconographic programming language, its visual elements for data representation and its representation of
system dynamics the model are closely related to pictorial simulation models [9]
and cellular automata models [10]. The approach is flexible, computationally
efficient, and typically does not require parallel-processing capabilities. Though
slightly awkward, it is also possible to use STELLA to carry out object-oriented
models.
It is the intention of this book to teach you how to model, not just how to use
models. We have chosen STELLA toward this end because it is a very powerful
tool for building dynamic models. The software also comes with an “Authoring”
version that enables you to develop models for use by others who may be
uninterested in the underlying structure of the model. However, since model
development and understanding are the purpose of this book, we do focus here on
the modeling process itself.
1.9 Extending the Modeling Approach
27