CODE

Cartesian coordinates Ocean model with three-Dimensional adaptive
mesh refinement and primitive Equations

Layout – RUN_01-decadal variability of Icelandic waters

 

A model layout was chosen which enables a highly resolving simulation of Icelandic waters embedded in the North Atlantic system. This way, also remote large scale structures like the Subtropical Gyre or the Beaufort Sea Ice Gyre and their linkage to the climate of Icelandic waters are included. The model domain is assumed to be closed along the equator and within Bering Strait. A model drift towards unrealistic temperature and salinity fields is avoided by a 365 days Newtonian relaxation towards climatologic values. This scheme allows the simulation of decadal and shorter scale variability, however slightly under-estimating the decadal variability.

 

 

1. Model domain and adaptive mesh refinement

With a blending technique between five different stereographic projections, having their projection points on the 40°W meridian between the North Pole and the Equator, the North Atlantic topography (GEBCO 2003 topography (BODC 2003)) is transformed into a Cartesian coordinates system. Afterwards, neglecting the Mediterranean, the North Atlantic/Arctic Ocean basin is subdivided into basic grid cells of 128 km horizontal and 160 m vertical spacing (fig. 1).

 

Fig. 1: The model domain in Cartesian coordinates resolved with basic grid cells of 128 km horizontal spacing.

 

Then, static adaptive mesh refinement, based on topographic structures, is applied. The horizontal refinement level 1 (= 64 km) is applied to the Nordic Seas, the North Sea, the Irminger and Iceland basin, the Canadian Archipelago and along the northern Mid-Atlantic ridge. (fig. 2).

 

Fig. 2: The computational mesh after the first step of horizontal mesh refinement.

 

The refinement level 2 (= 32 km) is applied along the Greenland-Iceland-Scotland ridge, the Greenland and North-West European shelf (fig. 3).

 

Fig. 3: The computational mesh after the second step of horizontal mesh refinement.

 

The refinement level 3 (= 16 km) is applied along the Greenland-Iceland-Scotland ridge and to the Davis Strait (fig. 4).

 

Fig. 4: The computational mesh after the third step of horizontal mesh refinement.

 

The refinement level 4 (= 8 km) is applied along the Greenland-Iceland-Scotland ridge, to the Davis Strait and along the steepest slopes of the Iceland Sea (fig. 5).

 

Fig. 5: The computational mesh after the fourth step of horizontal mesh refinement.

 

The refinement levels 5, 6 and 7 (= 4 , 2 and 1 km) are applied to the coastal waters around Iceland (fig. 6, 7 and 8).

 

Fig. 6: The computational mesh after the fifth step of horizontal mesh refinement.

 

Fig. 7: The computational mesh after the sixth step of horizontal mesh refinement.

 

Fig. 8: The computational mesh after the final seventh step of horizontal mesh refinement.

 

Vertically the adaptive mesh refinement is firstly applied to the near surface ocean of the entire model domain. Using six steps of mesh refinement the basic cell thickness =160 m is reduced to =2.5 m close to the sea surface (fig. 9).

 

Fig. 9: The computational mesh’s vertical structure.

 

Additionally, depending on the horizontal level of refinement, a minimum level of vertical mesh refinement is set. Hence, grid cells with =1 km have a maximal thickness of =10 m (see table 1). However, thereby the general structure with higher vertical resolution close to the sea surface and a reduced resolution for deeper waters is not altered. CODE does not use a vertical mesh refinement close to the sea bottom.

 

level of horizontal mesh refinement

[km]

minimum level of vertical mesh refinement

[m]

0

128

0

160

1

64

0

160

2

32

0

160

3

16

0

160

4

8

1

80

5

4

2

40

6

2

3

20

7

1

4

10

Table 1: The minimum level of vertical mesh refinement as a function of horizontal mesh refinement.

 

 

2. Forcing data

The atmospheric forcing is taken from the NCEP/NCAR re-analysis data set (Kalnay et al. 1996). It consists of six-hourly fields of the following seven variables: precipitation rate, specific humidity (2 m), sea level pressure, air temperature (2 m), total cloud cover, zonal and meridional wind speed (10 m). This data is pre-processed by interpolating it onto the model’s basic (=128 km) grid. Further interpolation, spatially onto the grid’s higher resolving parts and temporally for each time step, is done during the simulation.

 

Around Iceland the freshwater release of numerous rivers is a further important process of forming the ocean's upper layers stratification, with consequences to the coastal current field and to the ocean's primary production as well. To simulate this, the model uses estimates of the monthly mean freshwater runoff along the Icelandic coast line. This data set is based on simulations of the hydrological model of Orkustofnun, the National Energy Authority of Iceland . This way the model contains the discharge of 58 Icelandic rivers (fig. 10).

 

 

 

Fig. 10: The location and mean discharge of rivers included within the simulation.

 

 

The simulated temperature and salinity fields are restored to the climatologic fields of the PHC (Polar Science Center Hydrographic Climatology) data set (Steele et al. 2001). The restoring consists of a 365-days Newtonian, scale selective scheme towards the 12 monthly fields of the PHC. The term “scale selective” means that differences between simulation and climatology are computed only up to the horizontal resolution of 16 km (refinement level 3). If the grid cell is smaller, the correction term of the surrounding 16 km cell (in the terminology of the tree-algorithm: the 16 km “ancestor” cell) is applied. This way, smaller-scale structures which cannot be resolved by the climatology are not damped by the restoring.

 

 

3. Output data

 

During the simulation three-hourly means of the following 12 variables, describing the ocean’s physical state, are computed and stored on the hard disk: the velocity field (u,v,w), the sea surface elevation (), the temperature (T) and salinity (S) field, the horizontal and vertical exchange coefficients (AH, AV), the sea ice drift (ui, vi), the sea ice thickness (hi) and coverage (Ai). The averaging period of three hours was chosen to resolve tidal dynamics. Additionally, in order to obtain deeper insights into the near-surface thermodynamics, the three-hourly means of following variables are stored: sea surface freshwater flux (precipitation – evaporation, P-E), short-wave radiative heat flux (QSW), long-wave radiative heat flux (QLW), latent heat flux (QLAT)and sensible heat flux (QSENS).