Environmental Engineering Reference
In-Depth Information
Spatial resolution and spatial dynamics . Only spatial
models or models that can be spatialized are inte-
grated. Models are applied to the entire Wadden Sea
and operate at an appropriate spatial resolution to
reflect realistically the processes represented, the spatial
variability across the region, and its constituent geo-
graphical entities, subject todecision andpolicy-making
requirements. With a view to simplifying or aggregat-
ing the model, the effect of increasing or decreasing the
spatial resolution on the performance of the model is a
criterion for selection.
The key tradeoffs in the selection process were very
much between accuracy (of outputs and of process repre-
sentations) and simplicity (of models and of input data).
The resulting model needed to have sufficient spatial
and temporal detail and sufficient model complexity to
represent the processes accurately but needed to achieve
this over large areas in a fast and responsive manner with
a minimum of data.
21.4.3 Technical integration
Technical integration deals mostly with the hardware and
software aspects of model integration: how can pieces of
executable code be efficiently linked so that they together
perform the operations specified in the integrated model
at the right time, and so that data are exchanged in a
way that is consistent with the temporal and spatial logic
of the model? Is it possible to do this in a manner that
enables reconfiguration of the model in a straightfor-
ward manner? Can the material developed be reused for
DSS systems implemented elsewhere to address similar
aspects? Two aspects are decisive to answer these ques-
tions: the architecture chosen for representing the model
base and the integrated model and secondly, the software
technology used to implement the DSS.
With a view to developing a fast, responsive system,
operating stand alone on the PC of its end-users, an
architecture was chosen for WadBOS featuring the inte-
grated model as the core element of the DSS. Thus, the
modelbase is equipped with an integrated model fully
tailored and purposely developed to suit the precise needs
of the end-users. Rather than a suite of loosely coupled
submodels, it is a complex model by design. Each of
its submodels is adapted to that effect and (re)coded
as a software component according to a strict template.
Next, the components are coupled to one another as
required in order to establish the many linkages of the
integrated model. This architecture will generally result
in a user-friendly system. More than in other solutions
(e.g. Hahn and Engelen, 2000), it will represent most rel-
evant processes at the same level of abstraction and detail
and will perform well because of the limited overhead in
accessing and running the sub-models. It is a medium-
to high-cost solution because of the effort spent in the
design and (re)implementation. Unless object-orientedor
component-based technology (see for example, D'Souza
and Cameron-Wills, 1999) is used to implement the solu-
tion the maintenance costs can be very high, certainly if
the end-user requirements change and/or if the model
representation needs major repair and upgrading.
Scientifically proven . The process descriptions within
the models should be well understood and scientifically
proven. The model results should be as robust, reliable
and accurate as possible.
The following list of key end-user criteria was taken as
a guideline:
All processes . The WadBOS model should adequately
represent all the important processes necessary to pro-
vide the required policy outputs.
Spatial scale, temporal scale, time horizon . The WadBOS
model should provide information at a sufficient level
of spatial and temporal resolution to reflect the spatial
and temporal scale of variation in the most important
ecological and socio-economic variables. A spatial and
temporal resolution also at which problems occur that
need to be addressed as part of coastal zone policies.
Finally a time horizon that is relevant for policy design,
implementation and assessment.
Routine data . TheWadBOSmodel should be sufficiently
simple to run from routinely measured and available
data. In principle, no dedicated data are collected to
run WadBOS.
Output centred . The WadBOS model should be judged
mostly upon the quality of its output and less upon the
scientific or technical innovative character of itsmodels.
Policy centred . The WadBOS model should provide
appropriate results using indicators or variables
that directly interface with the policy-implementation
process and focus on environmental changes, anthropic
impacts, and management options. More abstract,
scientific or technical variables are less suitable for this
purpose.
Interactive . The WadBOS model should be fast, respon-
sive and interactive and should cater for a very short
attention span. A response time of less than 15 min-
utes per simulation run covering a period of 10 years
should be aimed for. Clever models, fast algorithms,
and efficient software code are required to achieve this
responsiveness.
Search WWH ::




Custom Search