Ocean Dynamics 
S 
20 
N 
a) 
mM o 
—\ Oo 
Do 
a] 
Canuto et al. 
(2002) 
7°E 8°E 9°E 
Canuto et al. Canuto et al. 
(2010) (2010) + realisab. 
8°E 9°E 7°E 8°E 9E 
S' ES 
X 
—— 
Sr 
A 
ka 
KR“ 
55°N 
S54°N 
—_ 
S 
2 
N 
La) 
MN 
A A 
© 
N 
S 
u 
N 
La 
mM © 
vo 
© 
(N 
- 
I = 
55°N 
54°N 
55°N 
;4°N 
0.0 0.5 
eastward velocity [m/s] 
Fig. 2 Eastward component of surface currents before, during and after the peak of storm event on 5 December 2013. The black dot at the top left 
describes position 6.55° E/55.58° N 
physically stable code that provides reliable and therefore por- 
table and reproducible results. Large £’s, on the other hand, 
indicate a stability problem in the code, which can have both 
technical and physical causes. 
Within this study, the £’s of a 12-h simulation between a 
model run compiled in optimization level O2 and a run com- 
piled in optimization level O3 were analysed. This analysıs 
showed without explicit stability and realizability checks local 
£’s of the eastward current component in the order of 0.7 m/s 
(Fig. 4, left and middle), i.e. in the order of magnitude of the 
eastward current itself. This is a clear indication of (local) 
instability. It was also found that instability no longer occurs 
with the implementation of the additional stability and 
realizability checks (Fig. 4, right). In this case, the £’s in the 
entire area were in the range 1077 
5 Results of downstream drift model 
In the previous chapters, it has been shown that missing 
realizability criteria can lead to at least temporally and 
spatially limited instabilities and thus partly unphysical cur- 
rents in certain situations. In order to demonstrate the impact 
9f the obviously incorrect currents on the whole application 
tange, we have carried out two drift calculation comparisons 
yased on fictitious cases using the BSH’s operational down- 
stream drift model SeaTrackWeb (Maßmann et al. 2014), 
whereby, on the one hand, currents generated without addi- 
tional realizability criteria in turbulence closure based on 
Canuto et al. (2002) and, on the other hand, currents generated 
with additional realizability criteria in turbulence closure 
zased on Canuto et al. (2010) were used as the forcing for 
the drift calculations. 
The first scenario describes the drift of an object (e.g. a 
container, a buoy, a floating boat or ship or even a human 
body, which has gone overboard), while in the second case, 
the drifting of oil after an oil spill has been simulated, where 
15,000 t of oil have spilled. Both fictitious cases “occurred” at 
the position 6.55° E/55.58° N on 5 December 2013 at 10:00 
UTC shortly before the storm peak (the position is shown in 
Fig. 2). The drift was calculated 72 h in each case, i.e. until 8 
December 2013, 10:00 UTC. 
2 ‚pringer