SOME OCEAN MODEL FUNDAMENTALS
25
Potential density or isopycnal coordinates: This coordinate is commonly used for idealized adiabatic simulations, with increasing use
for operational and global climate simulations, especially when
combined with pressure coordinates for the upper ocean in a hybrid
context.
Generalized hybrid vertical coordinates: Models formulated for
general vertical coordinates allow for different vertical coordinates
depending on the model application and fluid regime. Models with
this facility provide an area of focus for the next generation of
ocean models.
What about the horizontal grid? Although horizontal grids do not
greatly determine the manner that many physical processes are represented or parameterized, they greatly influence on the representation of
the solid-earth boundary, and affect details of how numerical schemes
are implemented.
Should we cast the model variables on one of the traditional A
through E grids of Arakawa and Lamb, 1977? Which one? The
B and C grids are the most common in ocean and atmospheric
modelling. Why? Section 3.2 of Griffies et al., 2000a provides
some discussion of this question along with references.
What about spectral methods commonly used in atmospheric models? Can they be used accurately and effectively within the complex geometry of an ocean basin? Haidvogel and Beckmann, 1999
present a summary of these methods with application to the ocean.
Typically, spectral methods have not been useful in the horizontal
with realistically complex land-sea boundaries, nor in the vertical
with realistically sharp pycnoclines. The reason is that a spectral
representation of such strong gradients in the ocean can lead to unacceptable Gibbs ripples and unphysically large levels of spurious
convective mixing.
Should the horizontal grid cells be arranged according to spherical
coordinates, even when doing so introduces a pesky coordinate
singularity at the North Pole? What about generalized orthogonal
coordinates such as a bipolar Arctic coupled to a spherical region
south of the Arctic (Figure 1)? Such grids are very common today
in global modelling, and their use is straightforward in practice
since they retain the regular rectangular logic assumed by spherical
coordinate models. Or what about strongly curved grid lines that
contour the coast, yet remain locally orthogonal? Haidvogel and
25
Potential density or isopycnal coordinates: This coordinate is commonly used for idealized adiabatic simulations, with increasing use
for operational and global climate simulations, especially when
combined with pressure coordinates for the upper ocean in a hybrid
context.
Generalized hybrid vertical coordinates: Models formulated for
general vertical coordinates allow for different vertical coordinates
depending on the model application and fluid regime. Models with
this facility provide an area of focus for the next generation of
ocean models.
What about the horizontal grid? Although horizontal grids do not
greatly determine the manner that many physical processes are represented or parameterized, they greatly influence on the representation of
the solid-earth boundary, and affect details of how numerical schemes
are implemented.
Should we cast the model variables on one of the traditional A
through E grids of Arakawa and Lamb, 1977? Which one? The
B and C grids are the most common in ocean and atmospheric
modelling. Why? Section 3.2 of Griffies et al., 2000a provides
some discussion of this question along with references.
What about spectral methods commonly used in atmospheric models? Can they be used accurately and effectively within the complex geometry of an ocean basin? Haidvogel and Beckmann, 1999
present a summary of these methods with application to the ocean.
Typically, spectral methods have not been useful in the horizontal
with realistically complex land-sea boundaries, nor in the vertical
with realistically sharp pycnoclines. The reason is that a spectral
representation of such strong gradients in the ocean can lead to unacceptable Gibbs ripples and unphysically large levels of spurious
convective mixing.
Should the horizontal grid cells be arranged according to spherical
coordinates, even when doing so introduces a pesky coordinate
singularity at the North Pole? What about generalized orthogonal
coordinates such as a bipolar Arctic coupled to a spherical region
south of the Arctic (Figure 1)? Such grids are very common today
in global modelling, and their use is straightforward in practice
since they retain the regular rectangular logic assumed by spherical
coordinate models. Or what about strongly curved grid lines that
contour the coast, yet remain locally orthogonal? Haidvogel and