Note that we, like Evans and Parslow, provided some diffusion through the density
barrier at the bottom of the mixed layer. Thus, there is always some nutrient input
from depth. Nutrient is also recycled; as implemented here half of the phytoplankton
mortality (0.07 d−1) is returned to the mixed layer as nutrient, and 15% of grazing
reduction of P is removed from the mixed layer. Those can be taken as the “export
production”.
(^) A “standard run” of a model like this (Box 4.4; Fig. 4.6a) applies parameters and
starting values that produce output pleasing to the modeler in some sense (it fits
his/her data; it fits his/her intuition of how the cycles should look; ... ; it doesn’t
crash). The next step is varying parameters and input values to examine the sensitivity
of the model to them. For example, in this model the small fall bloom starts on model
day 250, exactly when mixing depth starts to descend. The model is very sensitive to
shifts in mixing depth. As a test, the model can be run with some random variation
(only ± 1 m) in mixed-layer depth. The necessary changes are shown as “comment”