S. Dick, E Kleine 
22 
(x,y,s) system. Layer thickness varies in space and 
time, controlled by the transformation from (x.y.s) to 
(x.y.z). The vertical discretization is cast in terms of 
finite s (vertical integer index). Layering (number of 
stacked cells) in the water columns may be defined 
separately for each individual water column. 
The finite counterpart of continuity equation (4) 
provides the vehicle which controls cell volume 
dynamics. In the currently running 
implementation, the widely variable and flexible 
setting has been specified to remain close to 
Mesinger’s “step-mountain” coordinate, which 
amounts to y low, p □ relatively high, and v Das large 
as required. The quantity p Dmay be viewed as a 
diffusion coefficient. With the mixing-length concept 
in mind, it assigns both length and velocity scales to 
each water parcel (mass element) in the (.r, y, .v) 
system. In each water column, the velocity scale is 
introduced as a normalising constant while the length 
scale is allowed to vary (as a function of s ). When the 
flow diminishes, the vertical coupling term in (4) 
gives the model a tendency to relax to an equilibrium 
configuration. We will refer to this distribution of 
thickness as the reference configuration. This 
configuration may differ in different parts of the 
model domain but should reflect the prevailing 
hydrodynamic conditions as stratification or current 
shear. In its present version, the BSH model uses a 
maximum of 30 layers. In areas with more intensive 
tidal mixing, bigger reference layer thicknesses have 
been defined than in areas with low tidal influence. In 
the Baltic, the reference layer thickness of the first 10 
layers is 2 m each, increasing gradually toward the 
bottom. 
Another equilibrating term used is horizontal 
smoothing. The exchange coefficient v is composed 
of another scaling velocity and grid spacing length as 
the “mixing length”, in particular (v^ \’,j = (v* R costp 
AL, v* R Atp) where v* denotes the scaling velocity 
referred to and R stands for the earth’s radius. 
The new version of BSHcmod has been tested in 
several sensitivity studies. Different simulations have 
been carried out - and are still going on - in order to 
adjust the quantities (Uy, ft, v). Finding suitable 
combinations of the quantities is a lengthy procedure 
which will have be continued in the near future. At 
the moment, we can only present preliminary results 
of a special version using the parameter setting 
described above. In this configuration, variations of 
layer thicknesses are due mostly to water level 
changes. This test version has been running in pre- 
operational mode since February 2006. 
Figs. 3 and 4 show first computation results for 
salinities and currents in the western part of the Baltic 
Sea, predicted for 30 March, 2006, 00:00 CET and 13 
April, 2006, 00:00 CET, respectively. Although the 
model layers in this version have a considerably 
reduced ability to follow the deformation of water 
masses, baroclinic structures in the form of 
meandering frontal structures, stratification, and eddy 
formation are reproduced much better than with the 
preceding (z-level) version. Flowever, more 
sensitivity studies will have to be carried out during 
the next few months in order to find suitable values 
for □)», fi and v. In our view, the suggested method 
has considerable potential which should be fully 
utilized 
12° 00° 
11° 30' 
12°00r 
13° 30° 
14° oo- 
Salinity (PSU) 
Fig. 3. Computed surface salinities in the western Baltic Sea on 30.03.2006, 00:00 CET