Information Technology Reference
In-Depth Information
Salmon usually returns to the site in the tributary where it was born once.
But some individuals will go into other rivers. This can be incorporated by giving
the simulated fish appropriate preferences with a certain variability.
So far only upstream migration was discussed. Downstream migration will be
very much the same,only the graph will be different,every cell has at most one
neighbour,but some cells are neighbours of two or more cells. We can connect
down- and upstream migration. Since usually the permeabilities in both direc-
tions will be different,the simplest solution is to model this with two graphs
connected by one cell - symbolising the estuary for example. We can follow fish
from spawn places to the sea and back. Here,the effect of releasing hatched
salmon fingerlings at headwaters can be observed.
During the high season of downstream migration of eel temporally variation
of the regime of the turbines at electric power plants even their stop is discussed
[3]. The date of migration varies from year to year,therefore power plant regime
depends on observations or capture of fish. This aspect can be also an important
part of the reintroduction program for atlantic salmon.
For catadromous fish the graphs will be connected via the river's source.
Finally it is possible to model different fish species which compete for example.
3 Discussion
While in 1885 from the river Rhine 250.000 salmon had been caught today it
is virtually extinct there. Beside water pollution the fragmentation of rivers by
dams has been a relevant cause. Today the water quality is much better,but
dams are still a barrier for migrating fish.
We introduce a concept for modelling migrating fish. A river fragmented
by dams is represented by a grid of cells. Individuals are selected by random,
their migration to a neighbour cell is successful with probability
p
i
. This value
is specific to each cell and describes the dam's permeability. The permeability
parameters of the existing dams can be estimated from fish counts: at many
dams upstream migrating fish are monitored when passing through fish ladders
for example. Values can also be obtained from other comparable dams or turbine
types. With appropriate resources even catch and release experiments before and
after a dam are possible.
Our model allows to test different strategies of fish migration improvement.
It can be part of an application oriented decision support system. Results are
obtained fast and last not least easy visualisation facilitates communicating them
to policymakers. With the model different permeability settings can be tested.
If we built in a part that computes these permeabilities on base of an cost-
effect function specified for each barrier we can use a stochastic hill-climbing
method for example to get an optimal strategy for distribution of resources.
Also strategies obtained from analytic results with simplifying assumptions can
be tested.
Space is represented discrete in our model. Every cell represents one im-
poundment. This implies that the time it takes for a fish travels from one end
