19991022: 32 km Eta runs available on web (fwd)

Subject: 32 km Eta runs available on web

EMC has begun twice daily real-time runs of a full resolution 32 km
Eta Model prediction and Eta Data Assimilation System (EDAS) on the 
developmental SGI Origin 2000 computer system.  With the inoperable
status of the Cray C-90, resulting in a degraded Eta (no EDAS and
run at 80 km) backup version run operationally on a Cray J-90, the
entire suite of codes/scripts (preprocessing, analysis, prediction
model, postprocessing) was ported to the Origin 2000. The purpose
of this work is two-fold:

1) Continue to provide the highest quality mesoscale guidance to 
the field (if logistically possible).

2) Maintain the fully cycled EDAS (unbroken since June 1998 on the
Cray C-90).  The cycled EDAS has led to a reduction in model biases
of soil moisture, 2-m specific humidity, and skin temperature. To 
maintain these positive benefits it is considered crucial to maintain
the self-cycling of soil temperature/moisture in the EDAS.

After preliminary testing, the Origin 2000 EDAS/Eta-32 began realtime
forecasts at 0000 UTC 10/15/99. The Eta-32 forecast runs at the
same speed as it did on the Cray C-90 and finishes 30-60 minutes 
before the Eta-80 backup on the Cray J-90. Because of this result, NCO
personnel are currently working on a robust interface which will 
substitute Origin 2000 Eta-32 output for Eta-80 output if it is
available in time.  If these logistical problems can be overcome, 
then all output files from the Origin 2000 Eta-32 will be used in the
NCEP Early suite instead of those from the Eta-80.  This would 
eliminate both sources of current degradation of the early Eta runs 
at 00z and 12z, i.e. resolution and lack of an EDAS.  The one 
exception is the hourly forecast station profiles in BUFR.  The BUFR
sounding output from the Origin 2000 run is not produced due to time
constraints and must continue to come from the J-90 backup run of the
80 km Eta.

Web sites have been setup in the meantime to view the 32 km runs, 
they can be found at:

http://sgi62.wwb.noaa.gov:8080/ETA32/ (CONUS)

http://sgi62.wwb.noaa.gov:8080/ETA32AK/ (ALASKA)

http://sgi62.wwb.noaa.gov:8080/ETA32EPAC/ (East Pacific)

In the near future, we will create a web page showing forecast height
errors compared among the Eta-80 backup, the Eta-32 Origin run, and 
the AVN and will attempt to compute them from all the previous runs
that have been made (hopefully).  This will be setup at:

http://sgi62.wwb.noaa.gov:8080/ZVERIF_ETA32/

These efforts have been largely due to Eric Rogers with help form
Matt Pyle, Dennis Keyser and Bill Collins.

---------- End Forwarded Message ----------



Linda Miller - address@hidden
External Liaison, Unidata
University Corporation for Atmospheric Research
P.O. Box 3000
Boulder, CO 80307-3000
303 497-8646 fax: 303-497-8690
URL:  http://www.unidata.ucar.edu/staff/lmiller/un.act.html