[Wrf-users] WRFDA Single Observation Tests

[Crowell, Sean M.]

Hi,

I am using version 3.3.1 of WRF, WPS and WRFDA (on a different machine than the one I recently emailed about), and I'm trying to get a sense of the correlation length scales in the be.dat that I generated with obsproc.  The tutorial says to use Pseudo-Single Observation Tests to examine the structure of the background error covariance.  I am following the instructions in the tutorial, setting

num_pseudo = 1

and

pseudo_var = "Q"

and then running da_wrfvar.exe.  When I check the rsl.out.0000 file, WRFDA isn't applying any pseudo obs to the analysis.  The initial and final cost function values are zero.

Here's the strange part, when I change the variable to "U" and the values to "1" and "1", WRFDA applies the correction and generates an increment.  I checked the files in the da_obs and found the fortran code that sets up the pseudo observations, and "Q" is the correct letter for the moisture variable.  Incidentally, this bad behavior is repeated when I try "T" and "V" also.


As a workaround, I generated a obs*.3DVAR file with a single observation that is the model RH value perturbed by a value equal to the error, and WRFDA saw it, but changed its QC code to -88.

Any help with this would be useful.  I'm trying to do a fair comparison with an EnKF experiment, and so I need to know how long to make the correlation length scales to be fair.

======================================
Sean Crowell
Postdoctoral Research Associate
College of Atmospheric and Geographic Sciences
University of Oklahoma
Office: NWC 5420
Phone: (405)325-1089
scrowell at ou.edu<mailto:scrowell at ou.edu>





-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.ucar.edu/pipermail/wrf-users/attachments/20120817/efe67f2f/attachment.html