[MITgcm-devel] wetting and drying in ocean model

[Daniel Goldberg]

Hi all

I am hoping to begin a discussion about wetting and drying in the ocean
model -- that is, advancement of the horizontal boundary into dry columns.

I've had brief discussions with a few people about this problem, and one
thing proposed (if i understand correctly) is to change the part of the
domain over which the shallow water or primitive equations are solved, and
to allow other, simpler physics to govern where there is a thin layer.
Perhaps this is the simplest way to implement this -- but it seems that the
implicit free surface (solve_for_pressure) would then need to be solved
with boundary conditions that are externally imposed.

There seems to be another approach though, of using "artificial porosity":
http://onlinelibrary.wiley.com/doi/10.1002/fld.959/abstract
The essence is that a new variable, \psi, determines barotropic pressure,
while \eta still determines free surface. \psi and \eta agree unless \eta
is very small, in which case \psi can be negative. I don't know if a
similar approach has been attempted or implemented -- if so please let me
know.

I'm not at all sure about this -- but I think that the method could be
implemented in a way that changes are confined to SOLVE_FOR_PRESSURE and
MOMENTUM_CORRECTION_STEP. Moreover the advance of water into "dry" regions
would be governed by the physics of the flow itself, not externally. (Plus,
I think this approach might be easier to take the adjoint.)

There would be some drawbacks:

(1) What the authors call "trickle flow" -- i.e. water flowing over dry
topographic boundaries in a way that it is not supposed to. This effect can
be controlled though.

(2) For mass conservation, the equivalent of CG2D would need to solve a
nonlinear equation. The nonlinearity, though, would be pretty benign
(arising from the dependence of \eta on \psi when \eta is small) and I
think some efficient iterative solution could be found. In runs where
solve_for_pressure takes less than 10% of the run time, making it take
longer might not be so damaging.

Very interested in hearing feedback on this.

Dan

-- 

Daniel Goldberg, PhD
Lecturer in Glaciology
School of Geosciences, University of Edinburgh
Geography Building, Drummond Street, Edinburgh EH8 9XP


em: D <dgoldber at mit.edu>an.Goldberg at ed.ac.uk
web: http://ocean.mit.edu/~dgoldberg
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mitgcm.org/pipermail/mitgcm-devel/attachments/20150702/b03bed1b/attachment.htm>