From: loki@cres1.anu.edu.au (Jay Larson)
Message-Id: <9409200023.AA21908@cres1.anu.edu.au>
Subject: Re: CCM1 land/sea mask wanted
To: ccm-users@ncar.ucar.edu
Date: Tue, 20 Sep 94 10:23:48 EST
Hi Jozef,
In response to your question:
> I am currently working on CCM1 data processing from CCM1 2xCO2 experiment
> and I need to calculate land and sea statistics seperately. I need land
> sea mask for CCM1 in form 0-ocean, 1-land, or any other form giving
> information about land and ocean distribution in CCM1 R15. Thanks in
> advance for your help!
I guess I'll respond with a query of my own, first...are you doing your analy-
sis online at NCAR (i.e. are you using the model history tapes on their Mass
Storage System), or do you have the files on your own machine? If you're on-
line at NCAR, you can call up land/ocean/seaice masks for R15 sets that are
available for January and July, but are prescribed. The way you do this in
the processor is to specify the mask file (only need to do this in case A if
you're doing multi-case analyses) using this ICP:
SFCTTAP = '/CCMPROC/SR15JA'
or
SFCTTAP = '/CCMPROC/SR15JL'
the first of these being January, the second July. The way you would specify
what surface type over which to do the processing would be specified using the
following ICP (which is case-dependent):
MASKSA = 'LAND','LAND'
MASKSA = 'OCEAN','OCEAN'
MASKSA = 'SICE','SICE'
the first of these is "land only," the second is "ocean only," and the third is
"sea ice only." You can also combine maskings if you wish to exclude things,
for example, if you wish to exclude sea ice grid points, you can do this:
MASKSA = 'LAND','OCEAN'
and this will allow processing of all grid points except those which have sea
ice.
Using the supplied masks in the processor is foolproof if you just want to com-
pile "over-land" statistics. If you are interested in "over-sea ice" or over
ocean statistics, and you REALLY want to be sure, you should use the ORO field
output from the model with the same time sampling as the data you are studying.
This is an important point, given the fact that the placement of sea ice in GCM
runs that allow the sea ice to evolve (e.g. if the version of the model itself
has a thermodynamic or dynamic sea ice component) creates a time-dependent sea
ice/ocean boundary. Thus, the creation of a set of "over ocean" or "over sea
ice" statistics should be done with some awareness of this changing boundary.
If you are doing this processing off-line (i.e. not from history tapes using
the CCM processor), you should try to get your hands on the ORO field from the
model run, with the same time sampling as the data you have. If you have
access to the full history tapes, I'd suggest writing out your own ORO files
using the CCM processor, and then write a program that uses these ORO files
as the basis of a time-dependent ORO mask to compile your statistics. In the
interest of doing this in a completely rigorous, bomb-proof fashion, I'd advise
going with this strategy...