Die Kuste, 81 (2014), 255-271 
267 
simulated position is about 1.2 nm. The mean minimal distance of the whole simulated 
trajectory to die observed trajectory is about 0.6 km and the mean error is about 0.7 km. 
container drift start: 06-Jan-2012 17:15:00 UTC 
16 . 00 ' 
Figure 5: Seatrack web drift trajectory (blue) and observed trajectory (magenta) of two connected 
containers starting from a ship observed position (54° 18,2’N, 007° 11,6’E) at 06.01.2012 at 
17:15 UTC ending at 06.01.2012 at 23:30 UTC (plot is made with Matlab). The star marks the 
starting point of the simulation. Seatrack Web uses a wind drag coefficient of 2.3 %. 
To track possible error sources, we compare wind and wave measurements with the wind 
forcing used by the drift model. In Fig. 6 wind speed and direction at the simulated con 
tainer position are plotted over time (the wind model data refers to 10 m height). The 
dots in magenta show the observed values at Finol station (measurements are in 33 m 
height). The observed wind speed is about 12 m/s increasing to 15 m/s and die wind 
model data is about 2-4 m/s lower than the observed one (increasing from 10 m/s to 
13 m/s). Measured wind speed at Helgoland is generally a bit lower than die model wind, 
while TW Ems had a bit higher wind velocities. So overall the wind speed seems to 
match quite well. 
The measured and modelled wind direction match quite as well (the measurements are 
systematically about 5° smaller than the wind direction used in die drift model) and show 
that die wind backed from WNW to WSW. The differences between observed and mod 
elled values could be due to the height differences. Wind directions measured at TW Ems 
are almost equal to die ones at Finol, while at Helgoland die wind direction is a bit more 
nortiierly and having more variations tiian die model wind. In general the wind forcing 
used in die drift simulation seems to be consistent witii measurements, so errors in wind 
forcing seem not be die source of error for the drift simulation. 
At Finol die measured significant wave height decreases from about 2.9 m to 2.6 m 
and die mean wave direction was from NNW until 22:15 UTC. At 23:15 UTC die wave 
direction reacts on die changing meteorological conditions and backs to WNW. At the 
wave rider station soutii of Helgoland the wave height was 1.7 m at die beginning and 
increased later to 1.9 m after 21:40 UTC. The wave direction was WNW backing to W at 
about 22:40 UTC. This turn in direction coincides in time with die wave direction chang 
es at Finol. These observations show spatial and temporal variability of die wind wave 
and swell.