Introduction - Environmental Modeling - page 6

y -axis. Chwif and Barretto ( 2000 ) envisage a similar picture, putting 'level of
detail' on the x -axis and 'model confidence' on the y -axis. All these terms can be
taken as different terms for the complexity of a model on one side and its perfor-
mance on the other side.
The method, how to construct complex models, is another topic which is left out
in the topic. The major drawback of complex models is the increased number of
parameters, sometimes to a drastical extend. The situation may be worsened by the
fact that many new parameters are usually difficult to obtain or have to be deter-
mined by parameter estimation runs with the model. Another drawback may
appear, if the model becomes very sensitive to one or more parameters, i.e. that
relatively small changes of a parameter induce a tremendous effect on the output
results. A complex model which depends sensitively on numerous unknown
parameters can surely not be used as a predictive tool.
However, complex models have their justification. Whether they can be success-
ful also depends on the architecture, design and construction itself, especially on the
analytical and/or numerical techniques.
A complex model concerning sediment phosphorus and nitrogen processes is
presented by Harper ( 2000 ): the SNAPP model is constructed in MATLAB

and
contains even a graphical user interface. As another example Luff et al. ( 2001 )
present a MATLAB library to calculate pH distributions in marine systems.
Kumblad et al. ( 2003 ) construct an ecosystem model of the environmental transport
and fate of carbon-14 in a bay of the Baltic Sea, just to give another example.
A complex MATLAB

®

surface fluid flow model for rivers, streams and estuaries is
presented by Martin and Gorelick ( 2005 ).
It is not the aim of modeling to set up complex models. The opposite of that
statement is a more suitable goal: the aim of modeling is to find simple models that
explain some aspects of a real system. Unfortunately that aim turns out to be
a tricky one, because every real system appears to be complex, as long as there is
ample knowledge about the driving mechanisms. Moreover, if a system is complex,
a simple model can explain a few aspects at the most and that may not be enough to
solve a real problem.

®

1.2 Introduction to MATLAB ®

MATLAB is a mathematical software, originated and mainly developed by

mathematicians (Moler 2004 ). The name envisages a laboratory for matrix
calculations, where the mathematical term of a matrix refers to an array of numbers
such as

®

12
34

A ¼ (1.1)