11 Applications of an Oil Drift and Fate Model for Fairway Design
385
in Chap. 10 and in Andrejev et al. 2011; Soomere et al. 2011c, among others). It is a
measure for each individual oil particle, equivalently, each particle has its own residence time. This measure is highly variable (Viikmäe et al. 2010). As we consider
not just ensembles of particles, but multi-ensembles of 1000 particles per release,
it is necessary to estimate a characteristic residence time that adequately represents
all the simulations. 3 We use an appropriate statistical measure derived from the histogram of the residence time—the distribution of the amount of landed oil particles
as a function of time (cf. Chap. 9 and Viikmäe et al. 2010).
The oil drift model tracks the position of each particle until it leaves the Gulf of
Finland or settles on the shore, or on the seabed. As mentioned above, oil parcels
ending up on the seabed are ignored and only all those particles that land at a depth
shallower than 0.5 m are counted. This excludes sea bed fixation and ensures that
only near coastal contamination is studied. The analysis of the residence time at sea
focuses on the identification of cluster points, i.e., times at which large numbers of
particles land. The relevant histograms have a clearly defined peak (Fig. 11.3). The
location of this peak is a natural choice for the characteristic residence time. For
each release position, the first major peak in the histogram was chosen to represent
the typical residence time of oil at sea.
11.4.2 Seasonal Drift Patterns
The Gulf of Finland provides a good example for drift model based investigations of
characteristic oil drift patterns. Situated in the north-eastern extension of the Baltic
Sea (Fig. 11.2), it is a semi-enclosed sub-basin with a strong salinity gradient that
is supported by relative saline water inflow from the deeper layers of the Baltic Sea
and by river discharge from the largest regional fresh water source, the River Neva
(see Chap. 6). Its general circulation is strongly wind driven, but density driven currents play a role as well (Andrejev et al. 2004). Wind driven inflow of more saline
water at the Estonian (southern) side and outflow of somewhat less saline water at
the Finnish (northern) side of the gulf produce an extremely complicated circulation pattern that still contains several persistent features that might be used for the
design of optimum fairways (Soomere and Quak 2007). Seasonal variations of river
inflow, solar radiation and wind forcing make the stratification and current patterns
also highly variable in space and time. To derive climatologically valid fairway designs that serve all seasons, large ensembles of realistic oil drift simulations have to
be analysed. Counting 8 drift simulations per month, for a period of 3 years and for
528 release positions, we use a total of 152,064 drift simulations. This pool of simulations allows also for the evaluation of the relative importance of the direct wind
impact and current-induced drift patterns, as the oil drift model include parametrizations for both.
3 Ensemble averages are not appropriate, because some of the particles travel for a long time, especially while submerged and solely current driven. Ensemble minima are problematic as well. They
often represent nothing more than the drift geometry and the shortest distances to the coast.
385
in Chap. 10 and in Andrejev et al. 2011; Soomere et al. 2011c, among others). It is a
measure for each individual oil particle, equivalently, each particle has its own residence time. This measure is highly variable (Viikmäe et al. 2010). As we consider
not just ensembles of particles, but multi-ensembles of 1000 particles per release,
it is necessary to estimate a characteristic residence time that adequately represents
all the simulations. 3 We use an appropriate statistical measure derived from the histogram of the residence time—the distribution of the amount of landed oil particles
as a function of time (cf. Chap. 9 and Viikmäe et al. 2010).
The oil drift model tracks the position of each particle until it leaves the Gulf of
Finland or settles on the shore, or on the seabed. As mentioned above, oil parcels
ending up on the seabed are ignored and only all those particles that land at a depth
shallower than 0.5 m are counted. This excludes sea bed fixation and ensures that
only near coastal contamination is studied. The analysis of the residence time at sea
focuses on the identification of cluster points, i.e., times at which large numbers of
particles land. The relevant histograms have a clearly defined peak (Fig. 11.3). The
location of this peak is a natural choice for the characteristic residence time. For
each release position, the first major peak in the histogram was chosen to represent
the typical residence time of oil at sea.
11.4.2 Seasonal Drift Patterns
The Gulf of Finland provides a good example for drift model based investigations of
characteristic oil drift patterns. Situated in the north-eastern extension of the Baltic
Sea (Fig. 11.2), it is a semi-enclosed sub-basin with a strong salinity gradient that
is supported by relative saline water inflow from the deeper layers of the Baltic Sea
and by river discharge from the largest regional fresh water source, the River Neva
(see Chap. 6). Its general circulation is strongly wind driven, but density driven currents play a role as well (Andrejev et al. 2004). Wind driven inflow of more saline
water at the Estonian (southern) side and outflow of somewhat less saline water at
the Finnish (northern) side of the gulf produce an extremely complicated circulation pattern that still contains several persistent features that might be used for the
design of optimum fairways (Soomere and Quak 2007). Seasonal variations of river
inflow, solar radiation and wind forcing make the stratification and current patterns
also highly variable in space and time. To derive climatologically valid fairway designs that serve all seasons, large ensembles of realistic oil drift simulations have to
be analysed. Counting 8 drift simulations per month, for a period of 3 years and for
528 release positions, we use a total of 152,064 drift simulations. This pool of simulations allows also for the evaluation of the relative importance of the direct wind
impact and current-induced drift patterns, as the oil drift model include parametrizations for both.
3 Ensemble averages are not appropriate, because some of the particles travel for a long time, especially while submerged and solely current driven. Ensemble minima are problematic as well. They
often represent nothing more than the drift geometry and the shortest distances to the coast.