[RegCNET] Questions about CHE files with postproc
Christos Dogras
cntogras at geo.auth.gr
Fri Oct 2 16:31:24 CEST 2009
Let me guess... Only about one in four time entries has a non-zero
value and even those values that are non-zero, make no sense.
The same thing happened to me a few months back. In fact, I am
probably the "someone else" mentioned by Dr Ashfaq in his e-mail. If
your case is similar to mine, then all the others (ATM, RAD, SRF) are
ok.
The temporary solution I came up with, was to do the same
calculations using GrADS. The problem with that, is that it takes too
long, especially for parameters like the Aerosol Optical Depth, since
the computer has to sum the (Extinction CoEf) values of each level and
then average that over every month.
It's a stopgap measure, but it does the job. Good luck with your
problem and if you discover a better alternative, drop us a hint.
Christos Dogras
Graduate Student
Meteorology & Climatology Dept
School of Geology
Aristotle University of Thessaloniki, Greece
_______________________________________________________________________________
Hi RegCM users
The problem was found - when using postproc on CHE files, one needs to
set the number of
tracers in postproc1.param (ntrac) exactly equal to the number of
tracers used in
regcm.in (chtrname). If set incorrectly, postproc will run without
errors, but give you
really weird results!
Chris
Moetasim Ashfaq wrote:
[Hide Quoted Text]
Hello Chris,
On Sep 27, 2009, at 10:15 PM, R. Chris Owen wrote:
Hello fellow RegCM users,
I have a simulation with SO2, BC, and OC. I have used a version of
postproc dated Apr
2009 (at the top of postproc.f) to compute monthly averages from the
CHE files. It
appears to run OK (i.e., no errors given). In the postproc output
files, I have 3
things that seem odd and/or incorrect:
I had an update after Apr 14 which can be downloaded from:
http://web.ics.purdue.edu/~mashfaq/moet/PostProc.tar
<http://web.ics.purdue.edu/%7Emashfaq/moet/PostProc.tar>
1) the units for each tracer are unstated and apparently ~1e9 time
larger. The original
CHE files have each tracer in mass mixing ratio (kg/kg), with values
of an upper limit
in the range of 1 ppb. The postproc files have values with an upper limit of 9
parts/part (an unachievable number). Could this be a conversion from
part/part to ppb?
Or some other conversion (to concentrations or volume mixing ratios)?
There was 1.E9 multiplier in old postproc version. I do not have any
idea why it was
there but I removed that in last update. Regarding units, I have used
the info from CHE
grads control file or old postproc. There can be some mismatch but I
can only be able to
fix that after someone sends me accurate units. No other conversion
has been made in
postproc, therefore, unrealistic values can be due to some potential
bug in CHE
postproc. I will check this.
2) in the postproc files, the tracer at each output level, while
different in the shape
of the contours, is roughly equivalent in magnitude. Intuitively,
species with a
surface emission should be significantly higher at the surface than at
an altitude of
10 km and the original CHE files indicate that the surface
concentrations are 1e3-1e4
time larger than those in the top output level.
I need to check this thing. Couple of months earlier, someone else
also had some issue.
I will respond to this after making a thorough check.
3) the order of the levels from RegCM to posproc files seems to be
reversed. The
original CHE files have level 1 at the surface, while level 1 in the
postproc files is
apparently the upper most model level (though, it's hard to tell from
looking at the
data, since the data is of similar magnitude at each output level).
this seems to be
the case for the other file types as well (ICBC, SURF, ATM, etc.), so
maybe this is
also normal and I just don't know it.
This is normal.
Any thoughts or experience with this? Or maybe know someone there who
may have looked
at the CHE output before and has some insight?
I will upload the update after it is ready. In future, you may bug me
directly for any
postproc related problems.
Many thanks for your help.
Sincerely,
Chris Owen
More information about the RegCNET
mailing list