Geo-Mar Lett (2017) 37:163-170 
167 
Ô Springer 
Results 
2D versus 3D modelling 
Two example scenarios consider hypothetical oil releases 
at two different locations to contrast the simplified 2D 
model setup with detailed 3D simulations. In both cases, 
accidents are assumed to have taken place on 15 
May 2008. Within the 5 subsequent days, very different 
meteorological conditions occurred with wind speeds 
ranging from about 5 m/s on the first day to about 17 
m/s on the third day, when the wind turned from southeast 
to northwest. The first scenario assumes an accident in 
shallow water (depth about 8 m) at 8.27°E, 54.53°N about 
10 km to the southwest of the island of Amrum, close to 
where the PALLAS ran aground in October 1998 
(Reineking 1999). The second scenario is at 7.83°E, 
53.90°N (water depth about 20 m), 12 km to the north 
of the island of Wangerooge (roadstead “Neue Weser 
Nord”) where ships anchor before they enter the ports of 
Wilhelmshaven or Bremerhaven. 
Detailed 3D PADM scenario simulations are contrasted 
with simplified PELETS-2D simulations, the latter based on 
vertically averaged values of the same currents also fed into 
PADM. In PADM, simulations refer to heavy fuel oil. If oil is 
assumed to be 100% chemically dispersed, then the selected 
oil type matters neither in PADM nor in PELETS-2D. For 
untreated oil, however, the oil type impacts drift behaviour 
in PADM via changing strengths of physical vertical 
dispersion. 
For both scenarios. Fig. 1 shows the final situation 
after 5 days. Intermediate states after 1, 2 and 3 days are 
shown only for the second scenario. In both scenarios the 
untreated oil reaches the Wadden Sea, in scenario 1 after 
having travelled about 80 km to the south. Regardless of 
the model used, application of a dispersant considerably 
slows down particle drifts so that the oil/dispersant mix 
ture remains outside the tidal basins. After 5 days, dis 
tances between the centres of gravity of particle clouds 
simulated with PADM and PELETS range between 3 
and 4 km respectively. An exception is undispersed oil 
in scenario 2 because, in this case, the majority of the 
particles in PADM already beach at the northern shoreline 
of a barrier island. Beaching is neglected in PELETS, so 
that tracer particles may “slide” along the barrier island’s 
coast, making it more likely that they enter the Wadden 
Sea. 
Ensemble simulations 
The key result of this study (Fig. 2) is a probability map based 
on 2,190 hypothetical oil spill events assumed to have oc 
curred within each of 636 different cells of a regular grid 
covering the inner part of the German Bight. The map shows 
the spatial distribution of the chances that application of a 
100% effective chemical dispersant immediately after an oil 
spill occurred would successfully prevent pollution from en 
tering sensitive areas (shaded in blue). 
Fligh success rates can be found in regions that are at least 
5-10 km away from sensitive areas. The maximum distance 
that makes effects of dispersion dispensable depends on the 
orientation of the coast relative to the main wind directions. 
Along the south-north oriented part of the German coast, the 
regions of high probabilities that applications of dispersants 
would be successful in the sense of this study (successfril=oil 
does not enter sensitive areas) overlap with regions with water 
depths between 10 m and 20 m, where the application of 
dispersants is officially restricted by the outcome of a case- 
by-case assessment. 
In any oil-combating measures, time plays an important 
role. If oil slicks reach sensitive areas within, say, a few hours, 
then there may not be enough time for organizing the appli 
cation of dispersants. On the other hand, if dispersed oil does 
not reach any sensitive areas within 72 h, then it will usually 
be sufficiently diluted to avoid substantial harm to the ecosys 
tem. Figure3 shows 10th percentiles of travel time for untreat 
ed oil on the surface and dispersed oil in the water column. 
Generally, travel times increase with the distance from sensi 
tive areas. The maximum drift time taken into account equals 
the length of drift trajectories (7 days) calculated. Percentiles 
refer to the total amount of oil that enters any sensitive area 
within 7 days. Percentiles remain undefined for grid cells from 
which no pollution arises, a situation that occurs only for 
dispersed oil sheltered from wind drift. 
Discussion 
The example described above shows that, although simplified 
PELETS-2D simulations differ from frill 3D simulations with 
PADM, the changes of drift behaviour brought about by 
(100% effective) chemical dispersion are clearly not masked 
by, and do not crucially depend upon, the simplifying assump 
tions made. In this context, it should also be kept in mind that 
empirical data for model validation are lacking. 
Detailed numerical drift and fate simulations are nowadays 
state-of-the-art tools for predicting the evolution of an oil spill 
incident (see, for example, Brostrôm et al. 2013). Most of the 
uncertainty involved resides in the intensities of oil 
weathering processes under specific atmospheric/marine con 
ditions. Relatively simple parameterizations are employed to 
approximate spatially averaged effects of complex processes 
on scales unresolved by the model. But even when the oil 
weathering model can be considered quasi-perfect, informa 
tion about how any specific oil would behave in contact with a 
specific dispersant is often very sketchy. In this study, the