Models in Policy Formulation and Assessment: The WadBOS Decision-Support System - Environmental Modelling: Finding Simplicity in Complexity - page 337

in policy-making, were interviewed. From these sessions
answers were obtained relative to the following questions:

of readily usable mathematical models. These missing
models needed to be developed.
Partway through the modelling phase, a third
knowledge-acquisition round began, in which end-users,
domain experts, scientists, and model developers were
confronted with the mathematical representation of the
system. These were very intense exercises in which con-
sensus was sought relative to the formal representations
chosen, the level of integration accomplished and the
simplifications introduced.
With the information thus obtained, the knowledge
acquisition and modelling phases were completed and
the technical integration of models started, and so did the
technical implementation of the system.

• What and where are the (system) boundaries of the
Wadden system?
• What knowledge should be included inWadBOS?What
knowledge is available, from whom or from where can
it be obtained?
• What are desirable levels of detail and accuracy of
WadBOS?

The expectations were diverse, but consensus grew over
a representation covering the entire Wadden Sea, not
just the land adjoining, representing the natural system
and all the major human activities that take place on the
water. The reciprocal linkages between the natural and the
human system as well as the appropriate representation of
the processes in their complexly coupled multispatial and
multitemporal context was considered very fundamental.
The information thus gathered, complemented with
material obtained fromliterature research and the analysis
of policy documents, was the input of an initial systems
analysis: the main processes were identified and their
definition in terms of measurable characteristics and state
variables was carried out. Next, an initial set of system
diagrams and conceptual models were drafted and for
each process considered experts were identified.
In a second series of knowledge-acquisition sessions,
structured interviews were organized with the selected
domain experts. Most of the work with the experts was
carried out in sessions involving two to five people only.
Often a member of the potential end-users would partici-
pate. During the interviews, visual modelling was applied:
a graphical modelling tool was used to draft, discuss,
upgrade and complete the system diagrams and concep-
tual models according to the experts' knowledge. At the
end of these sessions a rather complete description of the
Wadden system was available in the form of graphical,
qualitativemodels. There was a very reasonable consensus
relative to this representation. Ambiguity and difference
of opinion had been resolved to the degree possible in
a number of consecutive visits and discussions in which
conflicting views were clarified and resolved.
As part of the knowledge-acquisition sessions, a num-
ber of existing mathematical models had been detected
and evaluated for incorporation (see below) and the
translation of the qualitative models into a mathemati-
cal representation began. To the extent possible, existing
models were incorporated or adapted for this purpose.
However, a lot of knowledge was not available in the form

21.4.2 Modellingand integrationofmodels

With a view to meeting the general objectives of the
WadBOS system, much effort went into separating the
detail from the essence in the domain representations.
Component (sub)models were integrated in theWadBOS
model base with a view to complete the system diagram
of the integrated model and to make it operational (see
Section 21.4.1). The models, desired for integration, are
what Mulligan (1998) names policy models as opposed to
research models . The actual selection and incorporation of
models was based on both scientific and end-user criteria.
The following scientific criteria were taken into consid-
eration:

• Models fitting the integration scheme . Only models were
integrated that fulfil a specific task within the Wad-
BOS integration scheme not dealt with by any other
(sub)model. (Sub)models compute a subset of the
state-variables and exchange the necessary informa-
tion among one another at the right temporal and
spatial scales during the calculations.
• Compatibility of scientific paradigms . Only models are
integrated that from a scientific/operational point of
view can be integrated. The scientific assumptions and
constraints underlying the models are assessed. Most
of the models used in WadBOS are spatial, dynamic,
nonequilibrium or quasi-equilibrium models that are
solved by means of simulation. Models using both rule
based and algebraic solution methods are retained.
• Time scales and temporal dynamics . Only dynamic mod-
els are integrated. Models need to span a strategic time
horizon (ten years) and operate at time steps reflect-
ing the inherent characteristics of the processes and
decision-making time frame (tidal cycle, one month,
one year).