Environmental Engineering Reference
In-Depth Information
(Gallego and Heath 2003). To test the effect of a density- and habitat-dependent
settlement scenario, a simulation was carried out where juvenile cod could settle
within an age window (90-100 days) but only if there was sufficient free settlement
“carrying capacity” (
0.5 individuals m
2
) within a suitable depth range
(20-325 m). Pelagic mortality (where implemented) increased exponentially within
that window.
Finally, computing hardware constraints prevented me from carrying out a
combined simulation with Gaussian egg production, horizontal diffusion with an
adequate number of particles per start location (
50), ontogenetic vertical migra-
tion behaviour, mortality and the more complex settlement scenario. Rather, two
simulations with a partial combination of those characteristics were carried out (see
Table
20.1
), although their results will not be presented here in detail.
20.3 Model Results
The sensitivity analysis presented here will concentrate on the effect of varying model
complexity on the
patterns
of transport and settled juvenile distribution, rather than
the absolute values. The comparisons between the baseline and more complex runs
will be largely qualitative. Although I will present a metric of average model
performance, the mean absolute error (MAE; Willmott and Matsuura 2005), I will
illustrate differences in transport patterns with particle trajectory plots (Fig.
20.2
). I
will also illustrate differences in juvenile distribution patterns with gridded (1.0
longitude by 0.5
latitude) maps of the percentage of the total (summed over the
model domain) settled juvenile population present in each grid cell (Fig.
20.3
).
A more detailed quantitative analysis has been carried out. However, it is less
relevant in the context of the application of the biophysical model in the overall
modelling system of Heath et al. (2008) and it cannot be adequately summarized in
the space available to this Case Study. However, in general, although a number of
“biological” techniques can be employed to validate biophysical model results (see
Paris et al. 2009), the appropriate ones for this type of sensitivity analysis are best
taken from the physical fields such as oceanographic modelling literature (see
“model error quantification techniques” referred to by Lacroix et al. 2009) and
atmospheric/climate research (e.g. Taylor 2001).
20.3.1 Mortality
There should not be noticeable differences in the particle trajectories between the
baseline run (Fig.
20.2a
) and its equivalent with mortality (not shown), as its only
effect would be to “cut short” certain tracks where the weighting of superindivi-
duals became
0. Overall, the effect of mortality on juvenile distribution patterns,
as implemented here and at the spatial resolution of this analysis was quite subtle
