Primitive Equation Modelling of Plankton Ecosystems
417
values of their biological parameters, and (2) the ambient environment, which is
determined extemalIy by the ambient climate (C*). The biomasses bj adjusts to
change in the environment, seasonalIy or from year to year. So the weighted average of each parameter will change with the ambient c1imate. The location of the
functional group in multi-dimensional parameter space depends on the ambient c1imate.
Pj = Pj(C*)
We can determine the functional relationship between Pj and C* by performing a
numerical experiment in which the alI the species of that functional group are
described explicitly (i.e. by LE particles). The first step is to create a stable virtual
plankton ecosystem fixed at the Azores site (27W, 4lN), with alI the species ofthe
functional group co-existing in balance with the ambient c1imate. Each species will
then lie at its stable demographic attractor bj( c* Azores). Starting from that balanced
ecosystem, the second step is to allow the water column to drift barotropically from
(27W, 41N) along the track defined by the mean circulation. The ecosystem in this
drifting water column is simulated by geographicalIy-lagrangian integration
(18.11). The ambient c1imate changes as it drifts along the track. The biomasses of
the set of species in the functional group adjust in response to their changing competitive advantage (18.9.1). The resulting data set bj{(C*(t)} can then be used to
determine Pj (C*).
This procedure assumes that each functional group can be parametrized separately. It uses a model in which only one of the model 's functional groups is
expanded to its constituent species. A more comprehensive parametrization would
be based on a model in which alI the functional groups are expanded into their constituent species. That demands more computer power than is currently available,
but it should become possible within twenty years. Meanwhile, we can parametrize
each functional group separately. That will involve a substantial programme of
numerical experimentation. It cannot begin until marine biologists have created a
data base of the primitive equations and associated parameters for alI the common
species. The value of establishing a formal procedure for parametrizing biodiversity is that it provides a rational goal for research. It is based on the discovery that
LE simulations are intrinsicalIy stable.
18.16.3 Models with several functional groups
Meanwhile, the functional groups in our models are not based on formal parametrization, but on each modelIer's personal review ofthe available data. Sometimes,
Pj,k (the value of each parameter j in a functional group k) is adjusted to make the
simulation of particular features of the ecosystem reproduce observations better.
But such "tuning" is problematic (see below 18.17). Whatever the rationale for
choosing Pj,k there is great interest in increasing the number of functional groups
from the minimal set represented in a food chain modellike the WB model. In this
model k has only two values for explicit population of "diatoms" and "copepods"
and two implicit populations of camivorous zooplankton and bacteria attached to
417
values of their biological parameters, and (2) the ambient environment, which is
determined extemalIy by the ambient climate (C*). The biomasses bj adjusts to
change in the environment, seasonalIy or from year to year. So the weighted average of each parameter will change with the ambient c1imate. The location of the
functional group in multi-dimensional parameter space depends on the ambient c1imate.
Pj = Pj(C*)
We can determine the functional relationship between Pj and C* by performing a
numerical experiment in which the alI the species of that functional group are
described explicitly (i.e. by LE particles). The first step is to create a stable virtual
plankton ecosystem fixed at the Azores site (27W, 4lN), with alI the species ofthe
functional group co-existing in balance with the ambient c1imate. Each species will
then lie at its stable demographic attractor bj( c* Azores). Starting from that balanced
ecosystem, the second step is to allow the water column to drift barotropically from
(27W, 41N) along the track defined by the mean circulation. The ecosystem in this
drifting water column is simulated by geographicalIy-lagrangian integration
(18.11). The ambient c1imate changes as it drifts along the track. The biomasses of
the set of species in the functional group adjust in response to their changing competitive advantage (18.9.1). The resulting data set bj{(C*(t)} can then be used to
determine Pj (C*).
This procedure assumes that each functional group can be parametrized separately. It uses a model in which only one of the model 's functional groups is
expanded to its constituent species. A more comprehensive parametrization would
be based on a model in which alI the functional groups are expanded into their constituent species. That demands more computer power than is currently available,
but it should become possible within twenty years. Meanwhile, we can parametrize
each functional group separately. That will involve a substantial programme of
numerical experimentation. It cannot begin until marine biologists have created a
data base of the primitive equations and associated parameters for alI the common
species. The value of establishing a formal procedure for parametrizing biodiversity is that it provides a rational goal for research. It is based on the discovery that
LE simulations are intrinsicalIy stable.
18.16.3 Models with several functional groups
Meanwhile, the functional groups in our models are not based on formal parametrization, but on each modelIer's personal review ofthe available data. Sometimes,
Pj,k (the value of each parameter j in a functional group k) is adjusted to make the
simulation of particular features of the ecosystem reproduce observations better.
But such "tuning" is problematic (see below 18.17). Whatever the rationale for
choosing Pj,k there is great interest in increasing the number of functional groups
from the minimal set represented in a food chain modellike the WB model. In this
model k has only two values for explicit population of "diatoms" and "copepods"
and two implicit populations of camivorous zooplankton and bacteria attached to