# DART software - Copyright UCAR. This open source software is provided
# by UCAR, "as is", without charge, subject to all terms of use at
# http://www.image.ucar.edu/DAReS/DART/DART_download
#
# DART $Id$

This is a WRF 3.0.1.1 regression test case, which can be used to test
that code changes do not break existing functions, as well as testing
a wrf/dart initial setup.

This is a global domain; no boundary conditions required.
The grid is 128 by 64 with 40 levels.

The observation files are the standard prepbufr NCEP observations, from 2007/04/26.
   
There are enough files to run with 20 ensemble members.
   
The smaller files are here in this directory.
   
The large data files (observations, initial conditions and WRF tarball)
are on the DART web server, at this http address: 
   
http://www.image.ucar.edu/pub/DART/WRF/regression/Global-V3
   
Details of the files follow.


This the Table of Contents for the WRF3 global DART regression case

1. wrfinput_d01.skamarock :: I got this file from Bill Skamarock will
   I was visiting Boulder in October 2007.  I presume that it was
   generated from the WPS from real meteorological data, but I myself
   did not perform that operation.  As ncdump will show, the domain
   is: 128 by 64 by 40. It is self-described for being valid at 12Z,
   18 Oct 2007.

   As I recall, I used wrf_to_dart to convert this wrfinput_d01 file
   into "dart_wrf_vector", which I then simply renamed "perfect_ics".
   I used the executable filter's ability to add small random noise to
   generate an ensemble from perfect_ics.  I then integrated this
   ensemble within DART using filter driven by a bogus obs_seq file
   for about two months to yield an approximately climatological
   ensemble valid at 12Z, 31 Dec 2007.  

   We were going to use actual NWP obs valid beginning 0Z, 1 Jan 2003,
   and running through till 0Z, 10 Jan 2003 (I think), so I used the
   executable restart_utility to reset the timestamp of my ensemble to
   12Z, 31 Dec 2002.

2. wrf_global.tar.gz.skamarock :: This is gzipped tar file of the
   global WRF3 code that I ran.  I got this from Bill Skamarock will
   out in Boulder in Oct 2007.  I don't know how similiar or different
   it is from what was released officially.  Hopefully they're
   identical, but in case they are not, I am including this.

3. namelist.input.skamarock :: This is the namelist.input that goes
   along with the wrfinput_d01 file I got from Bill Skamarock.

4. filter_ics.20021231120000 :: This is the DART input file of my
   spun-up WRF ensemble valid at 12Z, 31 Dec 2002.  It is stored in
   binary (native to RedHat Linux, but I don't know whether big or
   little endian or what?).  It has an ensemble size of 20.  Note that
   one should use "advance_global_model.csh" for 'adv_ens_command'.

5. input.nml.2002Dec31 :: This is a copy of the input.nml file I used
   when beginning the executable filter for the first two days of
   obs_seq files.  

   Note that 'outlier_threshold' = 3.0.  We found that this rejects
   many of the obs, but gives a better evolution of the analysis.

   Also, we did use adaptive inflation through the assimilation as
   well as a localization cutoff of 0.2.

6. obs_seq2003010112 - obs_seq2003010924 :: These are NOT included
   here because they are rather large (like 40 MB each) and I
   originally got them from Nancy.  Apparently, they are a classic set
   of obs_seq files to test global GCMs for their NWP prowess.  I can
   make them availabl if necessary.  Note that used the merge_obs_seq
   tool to combine four of these files into a single 2-day file (an
   example of this operation is included merge_obs_seq_nml within the
   included input.nml file).

7. namelist.input.wrf3.0.1.1 :: Apparently, the "dynamics" namelist
   options have changed from between the time of Bill Skamarock's
   original global WRF3 (see wrf_global.tar.gz.skamarock) to what is
   now officially released from NCAR.  Here is a copy of
   namelist.input with the dynamics namelist options worked out so it
   will run with WRFV3.0.1.1.  Note that I had to set km_opt .ne. 0 in
   order to get it to run, but this is clearly different than what
   Bill Skamarock originally gave me.

8. WRFV3.0.1.1.TAR.gz :: This is a copy of the tar and gzipped code
   bundle that I got from NCAR this morning.  This is officially WRF
   3. I'm including it simply for convenience as I assume anyone
   working with this will either already have it or can easily get it.

# <next few lines under version control, do not edit>
# $URL$
# $Revision$
# $Date$
