[Wrf-users] Re: Wrf-users Digest, Vol 29, Issue 1

[Don Morton]

Thank you all, for those who had the patience to see me
through my moments of confusion!  I think that, indeed,
I was getting small differences, all along.  See below:

On 1/3/07, WRF Help <wrfhelp at ucar.edu> wrote:
> Best way to do single obs test in WRF-Var is to activate
> "pseudo_ob_nl" namelist options.

I was a little reluctant to go off in this direction because
the ultimate goal is to ingest a number of obs (like the
rest of you) in a real-time fashion, so I wanted to start
by ingesting a single obs in the "normal way."

> After successfully running WRF-Var, plot analysis increments
> (analysis-background) to see how this single observation changed the
> initial background (FG) field at different levels. With this setting
> maximum increments is expected at level 10.

This statement is what grabbed my attention, and I decided to use
IDV to view the differences, and it showed a very small difference field
at level 0.942, on the order of something like 1.0E-1 or maybe even 1.0E-2.
It shows up as a rough circle centered right at my point of application.  At the
lowest level, it shows no difference.

In the following, I'm speaking with some uncertainty (I'm the computer
scientist who pretends to understand this stuff), so I'd be appreciative if
someone could correct or verify:

I had intended to apply my temperature perturbation at the surface, so in my
little_r file specified an obs height of 16.2m and -888888 for the pressure:

            70.20100          -148.47100PASC
         Deadhorse                               FM-12 SYNOP
                  ASOS
16.20000         1         0         0         0         0         F
      F         F   -888888   -888888      20061226180000-888888.00000
     0-888888.00000      0-888888.00000      0-888888.00000
0-888888.00000      0-888888.00000      0-888888.00000
0-888888.00000      0-888888.00000      0-888888.00000
0-888888.00000      0-888888.00000      0-888888.00000      0
-888888.00000      0     16.20000      0    245.00000
0-888888.00000      0-888888.00000      0-888888.00000
0-888888.00000      0-888888.00000      0-888888.00000
0-888888.00000      0
-777777.00000      0-777777.00000      0-888888.00000
0-888888.00000      0-888888.00000      0-888888.00000
0-888888.00000      0-888888.00000      0-888888.00000
0-888888.00000      0
      0      0      0

It "appears" that 3dvar_obs.exe tried to translate this height to a
pressure level,
as the data portion of obs_gts.3dvar showed a pressure of 998.15mb:

FM-12 SYNOP  2006-12-26_18:00:00 Deadhorse
        1      70.201               -148.471                 16.200
             PASC
 -888888.000 -88 200.00 -888888.000 -88  0.200
   99815.000  -5 100.00 -888888.000 -88   1.10 -888888.000 -88   5.00
               16.200   0   6.00     245.000   0   2.00 -888888.000
-11   2.00            -888888.000 -11  10.00


The PSFC field in my background was showing a pressure of about
1002.5mb, so I'm
guessing that since obs_gts.3dvar was showing the obs at 998mb, the obs was
actually being applied roughly 100 ft above the surface??  So, here I
was, trying
to look for changes in my T2 variable (2m temperature) and finding nothing.

Thanks again to those who offered insight.  It's an interesting area, and now
I'll move on ingesting multiple obs (eventually COSMIC data).

Don

-- 
Don Morton
Department of Computer Science
The University of Montana - Missoula
http://www.cs.umt.edu/~morton/