10 Applications of the Inverse Problem of Pollution Propagation
359
10.7.6 Sensitivity to the Resolution of the Ocean Model
The sensitivity of the technology in this respect is analysed in Andrejev et al. (2011)
based on on-line simulations of 10-day long Lagrangian trajectories calculated for
the period of 1 May 1987–31 December 1991 for the Gulf of Finland. The relevant
maps for the probability of coastal hits and for the particle age are evaluated using
the OAAS model at three different horizontal resolutions of 2, 1 and 0.5 nm and with
otherwise identical vertical resolution, initial, boundary and forcing conditions, and
trajectory calculation scheme. This range of resolutions characterizes a transition
from a poor representation of mesoscale effects by an eddy-permitting model to one
which is expected to adequately resolve statistical features of the field of mesoscale
eddies.
The dynamics of water masses in the Gulf of Finland is extremely complicated
and the resolution of even the 0.5 nm model does not perfectly resolve all the
small-scale features of water motion. The impact of subgrid-scale turbulence on
the spreading of trajectories was parameterized by the addition of a random disturbance containing a strong rotational component as described above. The resulting
set of 2D maps of cell-wise probability of hitting the coast and particle age is presented in Fig. 10.7. The general appearance of the distributions for the probabilities
and particle age are fairly similar in different resolutions. Particularly close are the
fields for 1 nm and 0.5 nm resolutions. The correlation between pointwise values
of probability and particle age is 0.98, the bias for p is 0.024 and for age as low
as 0.012. The mean absolute deviation is 0.037 and 0.21, respectively. The rmsd
between the two sets of fields is even smaller, 0.0021 and 0.081, respectively.
Also several integral measures almost coincide for the three resolutions. The
spatially averaged limiting values of the probability ¯
P (N max ) and the particle age
¯
A(N max ) also display hardly any dependence on the resolution. They both reach
almost identical stationary levels (0.66–0.69 and 5.25–5.33 days, respectively; the
standard deviations vary from 0.16 to 0.19 and 1.06 to 1.30) after five years of simulations (Andrejev et al. 2011).
The resulting solutions for fairways may be much more strongly affected by the
particular horizontal resolution of the ocean model than the integral variables and
2D maps discussed above. In spite of the almost perfect match of the integral characteristics the optimum fairways only match each other in certain parts of the gulf
(Fig. 10.19). They only overlap in a short section of the narrowest part of the gulf
between Naissaar and Porkkala and converge at a certain point to the north of Lahemaa. The fairways for the 1 nm and 0.5 nm resolutions also converge in the narrow
passages between the islands to the south of Gogland (Soomere et al. 2011b, 2011c).
In other parts of the gulf the optimum fairways calculated using different resolutions show a complicated pattern of behaviour. While all the optimum lines are
concentrated into a narrow corridor to the west of Naissaar, they deviate considerably from each other even in the narrow area between Tallinn and Helsinki. The
fairways to Vyborg calculated at different resolutions visit completely different areas of the Gulf of Finland (Fig. 10.18). While the differences between the fairways
at the 1 nm and 0.5 nm resolutions are moderate, the fairway for the 2 nm model
359
10.7.6 Sensitivity to the Resolution of the Ocean Model
The sensitivity of the technology in this respect is analysed in Andrejev et al. (2011)
based on on-line simulations of 10-day long Lagrangian trajectories calculated for
the period of 1 May 1987–31 December 1991 for the Gulf of Finland. The relevant
maps for the probability of coastal hits and for the particle age are evaluated using
the OAAS model at three different horizontal resolutions of 2, 1 and 0.5 nm and with
otherwise identical vertical resolution, initial, boundary and forcing conditions, and
trajectory calculation scheme. This range of resolutions characterizes a transition
from a poor representation of mesoscale effects by an eddy-permitting model to one
which is expected to adequately resolve statistical features of the field of mesoscale
eddies.
The dynamics of water masses in the Gulf of Finland is extremely complicated
and the resolution of even the 0.5 nm model does not perfectly resolve all the
small-scale features of water motion. The impact of subgrid-scale turbulence on
the spreading of trajectories was parameterized by the addition of a random disturbance containing a strong rotational component as described above. The resulting
set of 2D maps of cell-wise probability of hitting the coast and particle age is presented in Fig. 10.7. The general appearance of the distributions for the probabilities
and particle age are fairly similar in different resolutions. Particularly close are the
fields for 1 nm and 0.5 nm resolutions. The correlation between pointwise values
of probability and particle age is 0.98, the bias for p is 0.024 and for age as low
as 0.012. The mean absolute deviation is 0.037 and 0.21, respectively. The rmsd
between the two sets of fields is even smaller, 0.0021 and 0.081, respectively.
Also several integral measures almost coincide for the three resolutions. The
spatially averaged limiting values of the probability ¯
P (N max ) and the particle age
¯
A(N max ) also display hardly any dependence on the resolution. They both reach
almost identical stationary levels (0.66–0.69 and 5.25–5.33 days, respectively; the
standard deviations vary from 0.16 to 0.19 and 1.06 to 1.30) after five years of simulations (Andrejev et al. 2011).
The resulting solutions for fairways may be much more strongly affected by the
particular horizontal resolution of the ocean model than the integral variables and
2D maps discussed above. In spite of the almost perfect match of the integral characteristics the optimum fairways only match each other in certain parts of the gulf
(Fig. 10.19). They only overlap in a short section of the narrowest part of the gulf
between Naissaar and Porkkala and converge at a certain point to the north of Lahemaa. The fairways for the 1 nm and 0.5 nm resolutions also converge in the narrow
passages between the islands to the south of Gogland (Soomere et al. 2011b, 2011c).
In other parts of the gulf the optimum fairways calculated using different resolutions show a complicated pattern of behaviour. While all the optimum lines are
concentrated into a narrow corridor to the west of Naissaar, they deviate considerably from each other even in the narrow area between Tallinn and Helsinki. The
fairways to Vyborg calculated at different resolutions visit completely different areas of the Gulf of Finland (Fig. 10.18). While the differences between the fairways
at the 1 nm and 0.5 nm resolutions are moderate, the fairway for the 2 nm model