2.2. BSH Drift Model 
The oil spill drift model was developed and is being 
operated by the German Federal Maritime and 
Hydrographic Agency (Bundesamt fur Seescliifffahrt 
und Hydrograpliie, BSH) [11]. It is based on water level 
and current predictions of an operational 
hydrodynamical circulation model at BSH and on wind 
forecasts of meteorological models of the German 
Weather Service (Deutscher Wetterdienst, DWD) [12]. 
The drift model delivers forecasts of the drift and 
spreading of oil spills, which serve as decision support 
for contingency planning. Besides, a backtracking mode 
allows for tracking possible origins of pollutions. 
A possible oil spill is represented by a particle cloud of 
1000 particles (Lagrangian model) that is drifting under 
the influence of current and wind and spreading as a 
function of oil type [13]. Besides, the drift model 
considers the most important oil weathering processes 
as e.g. dispersion, emulsification, evaporation. In 
addition to the spatial distribution of the drifting oil, 
stranding and deposition on the sea bottom are modelled 
[12]. If the oil type of a pollution is known, the best fit 
of physical and chemical properties can be selected out 
of eight different oil types. 
3. DRIFT MODEL INPUT 
For starting the drift model, following data files are 
requested (Fig. 4): 
The topography file delivers information about 
the topography of North Sea and Baltic Sea with 
its islands, coastlines and depths. 
Current data files are generated by the 
circulation models and stored in daily files. A list 
of paths to all required files is needed. 
The basis data file contains information about 
the pollution itself, some model parameters and 
additional information. 
basis data 
topo 
graphy 
current 
data 
/ V 
particle doud: \ 
start distribution 
r 
drift DeMarine:^) 
Environment 
model 
I remote sensing 
A particle cloud . 
for each 
time step 
Figure 4. Drift model input files 
Through the basis data file, date, time and central 
position of an oil spill are determined as essential input 
parameters. If other parameters as e.g. oil type and oil 
amount are unknown, they have to be assumed based on 
expert knowledge of BSH model operators. For model 
parameters like calculation mode (forecast or 
backtracking) or simulation length, default values are 
preset and stored in the basis data file. 
3.1. Level-I Processing 
Starting the drift model run by specifying only the spill 
centre coordinate and no particle distribution is called 
Level-1 processing. Before DeMarine, this information 
was acquired during an aircraft mission and sent to BSH 
via phone or fax. 
For Level-1 processing, the drift model itself sets up the 
particle cloud (Fig. 5) after the first time step (15 
minutes) based on the centre coordinate and considering 
the wind field. 
Level I 
^ center coordinate {EMSA) 
. start distribution 
, drift path 
Figure 5. Example of drift model run: Level-1 
processing. Center coordinate, start distribution 
generated by dri ft model (zoom) and dri ft path 
Each particle is considered independently, i.e. it drifts 
with the current and underlies the physical and chemical 
processes. Additionally, the centre of mass of the 
distribution is calculated for each time step (drift path in 
Fig. 5; the time step is 15 minutes). All information is 
stored in an output file that is used for visualisation. 
3.2. Level-II Processing 
By specifying only the centre coordinate, information 
about the shape and dimension of an oil spill cannot be 
taken into account. It was the main goal of the 
subproject to use remote sensing data for setting up the 
initial particle cloud (Fig. 4, right) and for starting the 
drift model based on the observed distribution. This 
kind of model run is called Level-II processing. Oil age 
lias to be specified as an additional parameter having a 
strong influence on spreading.