Subject: Re: CCM formatted SST dataset as b.c.
From: Erik Kluzek (erik@cgd.ucar.edu)
Date: Mon Nov 30 1998 - 15:03:18 MST
On Mon, 30 Nov 1998, Jan Dutton wrote:
>
> I am writing to ask if a CCM history formated SST dataset
> can be used as the lower boundary condition
> (when properly reformated as a netCDF dataset) for
> a ccm.3.6 run.
>
Yes, this is fine, you just have to run it through ccm2nc, as you show
below. Your "real" problem is the "Y2K" problem associated with older CCM
data-sets. I'll explain this more below...
> I am trying to use an SST dataset obtained from the MSS in
> the BNDTVS namelist parameter for a T31 ccm.3.6 run.
>
> The error message
>
> SSTINI: Failed to find dates bracketing mcdate, mcsec= 19680901, 0
>
> is written in stdout.
>
> In the namelists I have specified
> SSTCYC = .False.
>
> As can be seen from above, I am specifying the start as 19680901.
>
The problem relates to the difference between 4-digit years and 2-digit
years (hence my refering to it as the Y2K problem). When you give
a start date as 19680901 it will try to find the dates 19680816 and 19681016.
Now if the SST file stores the dates as 2-digit years, it will have
dates 680816 and 781016, so it won't be able to find the dates it's looking
for. If you had turned cycling on (SSTSYC = .True.) the 2-digit year problem
goes away since it then doesn't care what the year is.
>
> The dataset is at /JEFF/sst-monthly/ccm3/nmc/T42.Dec1949-May1995
> NCAR system. It is a monthly average time variant dataset from
> December of 1949 to May of 1995.
>
If you look at the dataset you will notice that it has the dates stored
as two-digit years.
So in conclusion you need to overwrite the dates on the file as four digit
years rather than two digit years. You can use NCL, NCO, FORTRAN or whatever
your favorite method of editing netCDF are. Another option would be to
ask for the date 680901, the only problem with this is that your output files
look like
"0068-12" rather than "1968-12" which is certainly easier to understand.
> executed 'ccm2nc -r8 -C T42.Dec1949-May1995 T42.Dec1949-May1995.nc'
> to create the netCDF dataset.
>
> executed 'interpsst -x 96 -y 48 T42.Dec1949-May199.nc T31.Dec1949-May1995.nc'
> to interpolate the SST data to T31.
>
The above two lines are exactly what you want to execute. The next line is
not needed.
> executed 'definesurf -t /localdir/topo.nc T31.Dec1949-May1995.nc'
> to rewrite the ORO data.
Running definesurf will add the ORO and PHIS variables to the SST dataset. These
variables are not needed on the SST file, so this step is not needed. "definesurf"
is used only for the initial condition dataset (ie. SEP1.T42.0198.nc).
Erik
Erik Kluzek, (CGD at NCAR)
National Center for Atmospheric Research
Boulder CO, (off) (303)497-1326 (fax) (303)497-1324
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
This archive was generated by hypermail 2b27 : Thu Jun 01 2000 - 09:07:46 MDT