Latest News

29 October 2003

CODE UPDATE:

An error in how the Kain-Fritsch (KF) convective scheme is implemented within the workstation Eta has been reported.  Due to this error, convection was being called much less frequently than desired, with potentially detrimental effects on precipitation forecasts.

If (and only if) you are using the workstation Eta with KF convection, please download a corrected version of the KF routines.  They are located on the ftpprd.ncep.noaa.gov server in the directory /pub/emc/mmb/wrkstn_eta/ as the file kain_fritsch_replace.tar.Z.  Place this file in the worketa_all/eta/src/ directory, and uncompress/untar it.  The files will be placed into the kain_fritsch subdirectory.  To rebuild the KF executable, go to worketa_all/eta/install and issue the "buildall" command.

Any downloads made after 29 October should not require this update.

6 February 2003

CODE UPDATE:

The workstation Eta's land surface physics routine (SFLX.F) was found to be outdated, and responsible for a cold bias in near-surface temperature forecasts over snow.  In the plots below, OLD represents the code released on 11 July 2002, and NEW is the same case run with the current operational version of SFLX.F  The contoured/shaded fields is the forecast 2-m temperatures, and the red numbers are the verifying observations.  This update makes a dramatic difference over relatively shallow snowpacks.

For those who downloaded after 11 July 2002 and before 6 February 2003, you should upgrade your SFLX.F file.  It can be downloaded from ftp://ftpprd.ncep.noaa.gov/pub/emc/mmb/wrkstn_eta/wseta/latest/SFLX.F.  Save your old SFLX.F in the worketa_all directory to SFLX.F_old, and then copy this new version into the directory.  Run "make" to rebuild the executable.

             OLD                                                                NEW

11 July 2002

CODE UPDATE:

Not quite as significant as the 30 March 2001 update, but potentially very important for select users.  It adopts the changes made to the operational Eta code in July 2001 (revised land-surface physics) and late 2001 (new cloud package).  It also includes:

• The ability to generate hourly station output in BUFR format

• The ability to initialize the model using data from NCEP's Global Reanalysis.  Rerun all of your favorite storms!

• The ability to initialize from the new tiled 12km Eta output.

• A crude nesting capability, where a higher-resolution workstation Eta run gets initial and lateral boundary conditions from a "parent" run.

• Better support for extended range (> 256 h) Eta model forecasts.

• Inclusion of significantly better land-sea mask data (30" resolution).  See the comparison below for an ~4 km grid spacing domain over Chesapeake.  Red dots represent model land points.

MINOR BUG FIXES:

Several relatively minor bugs were reported by a user, and another was discovered at NCEP.  These problems only effect users with domains crossing either the Prime Meridian on the International Date Line.  Revised code can be obtained from /pub/emc/wrkstn_eta/recent_fixes/.  The revised versions of dutil.f and SSTHIRES.f go in the worketa_nh/eta/src/prep/initbc/ directory, and etatopo.F goes in worketa_nh/eta/src/sfc/.  Then execute "make" in worketa_nh/eta/install.  Thanks again to Dusan Jovic for providing very useful feedback!

LINUX BUG FIX:

A problem tied to the use of higher-resolution SST data (described in the 30 March 2001 entry) has been resolved.  Revised code can be obtained from /pub/emc/wrkstn_eta/recent_fixes/.  GRIBST.f goes into the worketa_nh/eta/src/prep/initbc/ directory (then must execute "make" in worketa_nh/eta/install), and the get_sfcfields.com script goes into worketa_nh/eta/bin/.  A major thanks to Dusan Jovic for providing a good solution to this bug!!

MAJOR UPDATE:

A new version is available which has a nonhydrostatic option, the ability to run in sigma or eta coordinates, and the option of using the traditional Betts-Miller-Janjic convection scheme or an alternative Kain-Fritsch parameterization.

This version also uses a higher-resolution SST analysis.  Here is a comparison between old and new along the east coast of the United States (in most parts of the world the difference between old and new isn't quite so dramatic!)

               OLD                                            NEW

              OLD                                    NEW

30 June 2000:

Semi-important fix!:

The code processing the land-sea masks was found to be deficient, leading to the land-sea mask being mapped on the order of 10 km too far north and east.  This isn't important for low resolution (say 32 km and coarser) runs, but is very noticeable in the 10-15 km resolution range.

An example along the west coast of Mexico, where the red 3's represent land on the Eta grid.  This grid has approximately 10 km horizontal resolution.  Click on each and compare the GEMPAK map background with what the model believes is land.:

A new version of etatopo.F can be found in /pub/emc/wrkstn_eta/recent_fixes/ on the ftp.ncep.noaa.gov server.  It should be placed in the users' worketa/eta/src/prep/sfc directory and the program should be rebuilt by issuing a  "make" command in worketa/eta/install.

Thanks to Bob Rozumalski and the NWS user community for motivating me to take another look at this problem.

Minor change:  replaced the deta and radiation files.  For runs with 45, 50, and 60 levels the distribution of vertical levels is slightly modified; this creates slight forecast differences compared with the old versions, particularly over topography.  These files can be found in /pub/emc/wrkstn_eta/recent_fixes/deta_rad on the ftp.ncep.noaa.gov server.  LINUX/DEC users will want to grab the dec_linux_data.tar.Z file, while others want the bigend_data.tar.Z file.  Whichever file is grabbed should be placed into the worketa/eta/bin directory, and then the installbin script should be run (either installbin LINUX, installbin DEC, or installbin NORMAL).
 

Updated the release on the server with a couple of bug fixes:

The boundary conditions were causing problems at the end of the forecast in certain situations.  The subroutine boco.f in worketa/eta/src/prep/initbc has been fixed to solve this problem.

The WAFS dprep code was failing for grids straddling 30W.  This oversight has been corrected with a revised version of dgeta2model.f for the worketa/dprep/src/avn_wafs/ directory.

These two files can also be found on the ftp.ncep.noaa.gov server in /pub/emc/wrkstn_eta/recent_fixes/.  If you already have a recent copy of the model you can simply download these files, put them in the directories specified above, and recompile the code.

Created an archive of old snow and SST files.  Currently covers the period between 6/98 and 9/99, with data available about every 3 days.  There are some gaps in the data due to computer problems.  Eventually more data may become available for the winter months to better capture rapidly evolving snowpack conditions.  The data are tarred and compressed in /pub/emc/wrkstn_eta/snow_sst_archive on the ftp.ncep.noaa.gov server.

15 March 2000:

Released the first major upgrade of the workstation Eta since the initial release in May 1999.  This version is known as version 0.2, or the MPI version.  It includes:
 

This form has fewer vertical levels, and "thins" the data near the poles to offset the convergence of meridians. The net result is that these files are about 15-20% the size of full resolution AVN files.

This brings the workstation model up to date with the operational Eta model code, and lets modellers with multiple-CPU machines tackle larger modelling problems.