Subject: Vertical level specification
From: Bill Campbell (bill@metosrv2.umd.edu)
Date: Mon Mar 30 1998 - 15:55:09 MST
My research involves a technique for selecting optimal vertical levels for
numerical weather prediction. I wish to test those levels with the CCM3, but
I've run into a problem. The levels my theory predicts are full levels, but
the CCM3 "expects" half levels, constraining the full levels to be the
arithmetic means of the pairs of half levels (in order to preserve the
energy-conserving pro[perties of the vertical finite difference scheme, I
suppose). Given the top half levels, and a list of full levels, it is
trivial to compute the half levels such that each full level lies halfway
between two half levels. Unfortunately, those half levels are not
necessarily adjacent! If two full levels are close to one another, and far
from their other neighbors, the pair of half levels that brackets one full
level will bracket both. So it is not possible to arbitrarily specify a set
of full levels for the CCM3.
I then modified the program that produced the full levels to produce half
levels instead. However, some of the half levels come out very close to
others. I am therefore concerned that the model may not run very well if one
or more of the layers is very thin.So my practical question is, "What is the
minimum separation of half level necessary so that the model run of the CCM3
is not excessively noisy?". Thanks for your help,
Bill Campbell
U Maryland Dept. of Meteorology
This archive was generated by hypermail 2b27 : Thu Jun 01 2000 - 09:06:10 MDT