306
T. Soomere
Earlier simulations indicate that much of the water is apparently transported back
and forth at the entrance to the gulf (Andrejev et al. 2004a, 2004b) and that the net
exchange forms a relatively small fraction of the total exchange. The almost constant
number of particles leaving the gulf in a month (Fig. 9.11) suggests that the outflow
of surface water is much more regular than the processes in the entire water column.
Only in August and September (which are usually much windier than May, June or
July, Mietus 1998) it is less intense.
This counter-intuitive behaviour probably reflects the different proportion of the
wind-induced Ekman drift (see Chap. 2, Sect. 2.3.5 for details) in different seasons. For example, strong SW winds (that are common in autumn) generate surface
Ekman drift to the east or SE, that is, towards the gulf interior. The surface dynamics seems to be relatively weakly correlated with the dynamics of the underlying
water masses during windy months and sometimes may even form an anticyclonic
gyre (Soomere et al. 2011a). In calm seasons and under ice cover, however, the underlying dynamics evidently will play a much larger role in the surface dynamics
(Gästgifvars et al. 2006).
The number of particles hitting the coast exhibits substantial temporal variability (Fig. 9.11). The lesson is that applications of the presented method require a
careful choice of the governing parameters for each particular sea area and circulation and trajectory model. A reliable statistics of coastal hits requires a sensible
amount of selected particles (carrying the adverse impact) to reach a properly defined nearshore. On the one hand, the time window should provide enough time for
a reasonable number of coastal hits to occur. For example, values of t W below 10
days are probably inappropriate for the analysis of coastal hits in the Gulf of Finland
during most months. In single years and the calmest months only a small fraction of
even 15–20-day long trajectories would enter the nearshore. On the other hand, the
average percentage of coastal hits should be well below 100 % to properly resolve
spatial details of the ‘ability’ of different sea points to provide coastal pollution.
As a rule of thumb, t W should not be much longer than the typical time at which
the largest number of hits occurs (Viikmäe et al. 2010); otherwise the rest of the
computing time of trajectories will not be justified. Still, the use of relatively long
time windows generally better resolves the spatial structure of maps of the time it
takes to reach the coast for particles released in different offshore domains. The
length of the time window used in experiments for the Gulf of Finland varies from
10 days (Soomere et al. 2010, 2011a; Andrejev et al. 2011; Lu et al. 2012) to 20
days (Soomere et al. 2011c).
As noted above, the calculation scheme does not account for the impact of
subgrid-scale turbulence on the shape of the trajectory. Although it is not likely
that the relevant effects could substantially modify the statistics of trajectories, it is
still interesting to roughly estimate the magnitude of related effects under the very
unlikely assumption that they always tend to deviate the simulated path from its
actual appearance. Its impact apparently is small in terms of statistics of isotropic
flow patterns and/or more or less circular basins but may considerably affect the
probability of coastal hits in elongated basins such as the Gulf of Finland.
First of all, the typical spreading of initially closely located particles over the
time window should remain below the width of the narrowest part of the gulf. If
T. Soomere
Earlier simulations indicate that much of the water is apparently transported back
and forth at the entrance to the gulf (Andrejev et al. 2004a, 2004b) and that the net
exchange forms a relatively small fraction of the total exchange. The almost constant
number of particles leaving the gulf in a month (Fig. 9.11) suggests that the outflow
of surface water is much more regular than the processes in the entire water column.
Only in August and September (which are usually much windier than May, June or
July, Mietus 1998) it is less intense.
This counter-intuitive behaviour probably reflects the different proportion of the
wind-induced Ekman drift (see Chap. 2, Sect. 2.3.5 for details) in different seasons. For example, strong SW winds (that are common in autumn) generate surface
Ekman drift to the east or SE, that is, towards the gulf interior. The surface dynamics seems to be relatively weakly correlated with the dynamics of the underlying
water masses during windy months and sometimes may even form an anticyclonic
gyre (Soomere et al. 2011a). In calm seasons and under ice cover, however, the underlying dynamics evidently will play a much larger role in the surface dynamics
(Gästgifvars et al. 2006).
The number of particles hitting the coast exhibits substantial temporal variability (Fig. 9.11). The lesson is that applications of the presented method require a
careful choice of the governing parameters for each particular sea area and circulation and trajectory model. A reliable statistics of coastal hits requires a sensible
amount of selected particles (carrying the adverse impact) to reach a properly defined nearshore. On the one hand, the time window should provide enough time for
a reasonable number of coastal hits to occur. For example, values of t W below 10
days are probably inappropriate for the analysis of coastal hits in the Gulf of Finland
during most months. In single years and the calmest months only a small fraction of
even 15–20-day long trajectories would enter the nearshore. On the other hand, the
average percentage of coastal hits should be well below 100 % to properly resolve
spatial details of the ‘ability’ of different sea points to provide coastal pollution.
As a rule of thumb, t W should not be much longer than the typical time at which
the largest number of hits occurs (Viikmäe et al. 2010); otherwise the rest of the
computing time of trajectories will not be justified. Still, the use of relatively long
time windows generally better resolves the spatial structure of maps of the time it
takes to reach the coast for particles released in different offshore domains. The
length of the time window used in experiments for the Gulf of Finland varies from
10 days (Soomere et al. 2010, 2011a; Andrejev et al. 2011; Lu et al. 2012) to 20
days (Soomere et al. 2011c).
As noted above, the calculation scheme does not account for the impact of
subgrid-scale turbulence on the shape of the trajectory. Although it is not likely
that the relevant effects could substantially modify the statistics of trajectories, it is
still interesting to roughly estimate the magnitude of related effects under the very
unlikely assumption that they always tend to deviate the simulated path from its
actual appearance. Its impact apparently is small in terms of statistics of isotropic
flow patterns and/or more or less circular basins but may considerably affect the
probability of coastal hits in elongated basins such as the Gulf of Finland.
First of all, the typical spreading of initially closely located particles over the
time window should remain below the width of the narrowest part of the gulf. If