Experiences
on Running POM w/WAD:
–
email yours to lyo@princeton.edu.
---------------------------------------
Jul/30/2007:
>>>>>>Contributed
by l.oey; lyo@princeton.edu:<<<<<<
Subject:
pom98_wad.f
w/Smolarkiewicz scheme for
water depth, nsmolar=1:
In
general, the code is robust (and faster) with nsmolar
= 0 so that water depths are not
solved with Smolarkiewicz scheme
(i.e. code does not call advt2d). However,
for high-resolution experiments (dx ~
2 km and
preferably less), the Smolarkiewicz
option with nsmolar = 1 can give
smoother results, say with
less isolated thin-fluid cells. The
modified code (download
here)
is for the same island/seamount problem as the posted version pom98_wad.f
except that we now have nsmolar = 1
and also finer dx =
2 km (previously dx =
4 km) over the island.
In
general, the Smolarkiewicz
option (i.e. nsmolar = 1) requires
finer grids and maybe even a
larger value of hc – see next.
>>>>>>Contributed
by l.oey; lyo@princeton.edu:<<<<<<
Subject:
Value of hc:
The
default hc = 0.05 m is used to
separate dry (if water
depth <= hc) and wet cells, and is
adequate for cases tested with the Smolarkiewicz
option turned off (i.e. nsmolar = 0).
However, when nsmolar = 1, the
hc may need to be larger, say hc
= 0.1 m, in order to keep the model stable. I
found the following report
for
the ADCIRC model; it catalogs
sensitivity tests for various hc
(report’s Hmin) which may
be useful in our case also.
---------------------------------------