[Wrf-users] Attention to WRF users- Relative humidities coming out of WPS above about 100 mb (UNCLASSIFIED)

[Dumais, Bob (Civ, ARL/CISD)]

Classification: UNCLASSIFIED
Caveats: NONE



Dear WRF Users,

        Please refer to the email I sent to my colleagues below. What do
you make out of these WPS concerns I have (either 3.0 or 3.1 versions)?
If necessary, I can share one of my met_em* files for you to see. Thx-
hope you all have a great day!

                                                              Bob

-----Original Message-----
From: Dumais, Bob (Civ, ARL/CISD) 
Sent: Thursday, January 28, 2010 10:41 AM
To: Flanigan, Robert (Civ, ARL/CISD); Passner, Jeff (Civ, ARL/CISD);
Kirby, Steve (Civ, ARL/CISD); Haines, Patrick (Civ, ARL/CISD)
Cc: 'Sen Chiao'
Subject: Attention to WRF users- Relative humidities coming out of WPS
above about 100 mb (UNCLASSIFIED)

Classification: UNCLASSIFIED
Caveats: NONE

All-

          I have stumbled quite accidentally onto what may be a problem
(either our end, with WPS or even with NCEP model output itself) with
moisture values being written to the WPS output files (*the met_em*
files) above about 150-100 mb. This seems to be independent whether NAM
or GFS 1 deg or 1/2 deg is used. What happens is the RH values that are
being written out to the met_em* files are very low above about model
level 55 in our configurations (usually 60 levels with top at 50 mb). In
fact, the values should be floating points but are instead written as
integers (0, 1, 2, etc) at those suspect levels, and occasionally are
even slightly negative (ie; -0.22 %). The other model fields (other than
related moisture like qvapor) look fine, and the simulations below about
100 mb seem reasonable and have compared pretty well in our case studies
throughout. The dewpoint seems to be reset to some minimal value around
-84 C or so in WPS at these same levels, and quite often you will find
when TD is plotted in GrADS above level 55 or so (~ 100 mb) the result
Will be a constant value of this -84 (or something close to this
number). I find this behavior very strange, but looking at all my runs
in the past this seems to be constant regardless of WPS version (3.0 or
3.1) and WRF version (3.0.1.1 or 3.1.1). Most of my runs have been on
our mjm or the old jvn linux clusters. If you guys would, please take
some of the met_em* files you have, and do something like "ncdump -vRH
met_em* > output_test" and then "vi + output_test"  . I believe this
will allow you to see the RH values in the met_em* file starting at the
highest pressure level of the NCEP model data (50 mb for NAM and 10 mb
for GFS). Scroll up and look at the numbers for some time. Do you find
them strange? I noticed this only because my RH and qvapor plots above
level 55 have been looking abnormally low for some time. I never
realized it was apparently originating in WPS. Perhaps this is just
simply an issue of actually calculating RH at these levels in the NCEP
models themselves?

              I am going to forward this email to WRF help as well. I
also want to ask them about those strange "netCDF File Attribute Not
Found" messages that pop up in the real.exe rsl.error files (only for
certain parameters that may not be used at all) when one sets the debug
value in namelist.input to a big number such as 500 (if you leave
debug=0, these messages are not written out to the rsl.error files).
These warnings seem to occur all the time, and have no bearing on the
successful execution of real.exe. Finally, I would like to ensure that
soil parameters are sufficient from GFS to properly run the NOAH land
surface model option in namelist.input (they are such for NAM). 


 
Bob  
Classification: UNCLASSIFIED
Caveats: NONE


Classification: UNCLASSIFIED
Caveats: NONE


Classification: UNCLASSIFIED
Caveats: NONE