168 
Geo-Mar Lett (2017) 37:163-170 
*£) Springer 
Days until 
sensitive 
areas get 
affected 
0.5 
1 
2 
3 
4 
5 
6 
Fig. 3 Spatial distribution of the 10th percentiles of travel times of 
untreated {left) and dispersed (right) oil entering sensitive areas (blue). 
The 10 m and 20 m depth contours are marked in grey and dark blue 
respectively. Grid cells from which no dispersed oil enters any sensitive 
area are faded out 
focus on pure drift behaviour avoided the inclusion of too 
many details that would have complicated the broad picture. 
Long-term model-based reconstructions of environmental 
conditions (e.g. Weisse et al. 2009) are nowadays available 
and provide a realistic picture of natural variability. In the 
context of risk assessments, such datasets are of great value 
in that they implicitly represent the correct statistics of weather 
conditions (e.g. Chrastansky and Callies 2009). In this study 
several years of archived output from operational BSH simu 
lations were used, relieving the analyst from the need to pre 
specify and properly weight a number of fixed reference 
weather situations. Wind and current data were supplemented 
by wave data generated for climate study applications. 
ft is important to realize that numerical values of the prob 
abilities that application of a dispersant at given locations 
would improve the situation after a hypothetical accident oc 
curred (Fig. 2) may have different interpretations. Generally, 
application of dispersants was labelled unsuccessful if signif 
icant wave heights were not in the required range of 0.5-3.0 
m. Wave energy being either too low or too high occurred for 
somewhat less than 25% of all simulated oil spill events, so 
that a success rate of about 75% could not be surpassed. This 
puts the upper limit (65%) of the probability scale attached to 
Fig. 2 into perspective. 
In addition, the use of dispersants was labelled unsuccess 
ful in all cases when it did not improve the situation. This 
applies in particular to those accidents far from the coast that 
would not endanger sensitive tidal flats anyway (within 1 
week’s time). Applications of dispersants were labelled 
unsuccessful in the sense that they were needless. This also 
applies to inshore regions if barrier islands shelter tidal basins 
from being polluted. 
Whether or not the use of dispersants at nearshore locations 
diminishes the amount of oil entering the intertidal zone de 
pends on whether the wind drift suppressed by chemical dis 
persion would act in favour (offshore breezes) or to the disad 
vantage (onshore breezes) of coastal protection. Even for on 
shore breezes, however, chemical dispersion may be useless if 
the location of pollutant release is close enough so that the 
pollutant can enter sensitive areas following tidal transport. 
This depends on the tidal phase at which an accident occurs 
as well as on the orientation and size of the local tidal ellipse. 
Tidal ellipses in the German Bight were described by Carbajal 
and Pohlmann (2004), Port et al. (2011) and Stanev et al. 
(2015), for instance. When dispersed oil can enter the 
Wadden Sea via tidal currents, any positive effects of disper 
sion through changed drift paths disappear. This explains the 
broader belt of low success rates in the Jade-Weser region 
(east of 8° east. Fig. 2). 
One simplifying assumption in this study is that dispersants 
were applied immediately after an accident took place. In 
practice, dispersants would be applied as soon as possible to 
start the intended effect in a timely manner. Otherwise oil 
viscosities increasing with time could hamper the formation 
of small oil droplets. Flowever, toxicity of dispersed oil may 
also be affected by weathering of oil prior to chemical disper 
sion. According to French-McCay and Payne (2001), espe 
cially evaporation of the most toxic oil components from a