Environmental Engineering Reference
In-Depth Information
MAGIC was originally developed in the United States
(Cosby et al ., 1985), but it has found more use in
Europe after the discontinuation of America's acid-
rain research, and now the focus is on surface water.
The core processes are alkalinity mass balances, sul-
phate adsorption and cation exchange. The minimum
time-resolution of the model is one year, equal to the
numerical integration step in the model. It is a one-
layer soil model. The model is calibrated by adjusting
the parameters for weathering, initial base saturation,
selectivity coefficients, gibbsite coefficients and current
base saturation. Current stream chemistry is then used
as the target for optimization (Cosby et al ., 1985). On
surface waters, the model is acceptably well confined.
to use but provides more elaborate results. Which is best
dependsonthepurposeoftheuser.
There are several model proposals for nitrogen assess-
ments but these should be classified as tentative at best.
Existing models operate with excessive calibration; the
models are actually calibrated in such a way that inputs
become output directly. Once the demand is made for no
excessive calibration, there is virtually no model available.
For calculation of critical loads for nitrogen, SMB-N is
used. It is a simple mass balance, where each term is
estimated by largely empirical methods (uptake, immo-
bilization, permitted runoff). SOIL-N, MERLIN and
MAGIC-WAND are available and operable, but inputs
leaves so much to the user to predefine that we can safely
say that a lot more work is required before we can trust
our decisions to them.
The most important property of the models that survive
is that all have observable parameters and input data that
are strongly simplified, and can be simplified further.
These models also have in common the fact that they
were developed by strong but small groups of researchers
who had long-term financing (over more than five years)
and a strong drive to apply the models in practical life.
The models that have not survived in practical use
also have certain properties in common. Several models
could never be applied at a regional scale. This limita-
tion was often caused by the lack of simplification. If a
model is allowed to contain too many 'pet processes',
it will be hampered by many unnecessary parts, which
still require input data and computation time. Inclusion
of many processes does not necessary improve perfor-
mance. Beyond a certain point (see Figure 17.5) the
model performance quickly deteriorates as more pro-
cesses and parts are added. Too often political prestige or
private, short-sighted ambition prevented the necessary
simplification of these models. Some models have process
descriptions of such a nature that too many of the param-
eters of the model have no physical significance that can
be determined by measurement, and the parameters are
not observable. This problem effectively precludes gener-
alization and transfer to regional use. A long list of such
very impressive, but rather useless models can be made.
Figure 17.10 shows the 'critical load model'. The spe-
cific computer code is chosen with all the information
and interpretations that actually take place before the
computer code can come into play. The numbers for
deposition are not objective data; they are measured con-
centrations in a plastic cup of rainwater that somebody
collected under a tree or maybe beside it. The deposition
value is the result of one of those undefined models we
SMART is a more recent model (de Vries et al ., 1989)
and its focus is on average major soil chemistry. The
core processes are alkalinity mass balance, sulphate
adsorption and cation exchange. The minimum time
resolution of the model is one year, equal to the numeri-
cal integration step in the model. The model is calibrated
by adjusting two parameters: the weathering rate and
the initial base saturation (de Vries et al ., 1989); present
base saturation and soil chemistry are used as optimiz-
ing targets. It is a one-layer soil model. The model is
reasonably well confined on soils and surface waters.
SAFE was developed for forest soils in Sweden and
focuses on chemical weathering, simple nutrient
cycling and development of cation exchange with time
(Warfvinge et al ., 1998). It is the dynamic version of
PROFILE. It calculates soil chemistry layer by layer - it
is a multilayer model. It differs from the two other
models by calculating the chemical weathering from
physical and geological soil properties and not by using
the weathering rate as a calibration parameter. The core
processes are weathering and cation exchange. This
approach tends to cause difficulties with application to
catchments and lakes when the average soil depth for
the whole watershed is needed. The model is calibrated
by adjusting initial base saturation with present base
saturation as the optimizing target. The model is
uniquely defined for soils and acceptably well defined
for surface waters.
In summary, the MAGIC and SMART models are
easier to apply than SAFE because of the smaller amount
of input data and the greater freedom in calibration. They
are cheaper to use, but lead to simpler results. SAFE is
significantly better constrained by its stricter calibration
on fewer parameters, its higher demand on input data and
its higher stratigraphic resolution. It is more expensive
Search WWH ::




Custom Search