20040317: Eta update-

NOTE: The nws-changes mailing list is no longer active. The list archives are made available for historical reasons.

With the 1200 UTC run, the following changes were implemented in the
EDAS/Eta system (running at 0000 UTC, 0600 UTC, 1200 UTC and 1800 UTC).
Eric Rogers was responsible for the overall implementation of this
change package.

Brief Description of Major Changes:
- Use of bias adjustment in precipitation assimilation
- Add use of GOES-12 radiances to 3DVAR
- Fraction of frozen from Ferrier used in land-surface scheme
- Fixed indexing error which kept first-layer cloud (fog) from
       inclusion in the longwave radiation calculations
- 3DVAR code upgraded to use less computer memory
- Use of faster (20%) post-processor code
- Fixed "time" of output solar flux which was off by 30 minutes
- Increased bit-precision for total column cloud / hydrometeor output
       fields
- Added ability to post numerous output fields


Reasons for Changes:
- Assimilated Precip Bias Adjustment

Since July 2001, observed precipitation in the form of gridded analyses
of gauge observations and radar estimates been assimilated in the Eta
Data Assimilation System (more info at
ftp://ftp.ncep.noaa.gov/pub/emc/wd20er/caftimay01/v3_document.htm ).
Those analyses show a slight tendency to underestimate the precip
amounts which, in turn, underestimates the soil moisture values.
A new procedure has been implemented that keeps track of the deficit
between the assimilated analysis and a ground truth field and makes
small adjustments to minimize it. Ying Lin was responsible for this
aspect of the change package.

- 3DVAR changes

Radiances from the GOES-12 geostationary satellite were turned on in
the the global analysis back in November 2003.  Little change was
required to the 3DVAR to turn these data on in the EDAS, but bias
adjustments had to be spun up.  GOES-12 is the eastern satellite so it
was expected that impacts would be minor when measured, as we do, with
CONUS radiosondes.  The 3DVAR was also recoded to reduce its use of
computer memory.  Dave Parrish was responsible for this aspect of the
change package.

- Use of Fraction of Frozen from Ferrier

As pointed out in a study by Lackmann et al. (2002 WAF), the Eta
Model's land-surface model diagnosed precipitation type based only on
the air temperature in the lowest model layer.  The new procedure uses
the field of fraction of frozen precipitation which comes out of the
Ferrier grid scale
cloud & microphysics scheme instead of the simple diagnostic check.
This
leads to 1. warmer surface temperatures in freezing rain events where
the latent heat of freezing keeps the ground and air warmer and 2.
cooler surface temperatures when snow is falling and the surface layer
is
warmer than 0ºC, where melting of snow will cool the ground.  Ken
Mitchell, Mike Ek and Brad Ferrier were responsible for this aspect
of the change package.

- Fix to index in longwave radiation

A problem was discovered in the longwave radiation scheme in the Eta
Model.  An index error kept cloud in the first layer (fog) if present
from being included in the calculations.  Correction of this error
results in increased surface downward longwave fluxes of  <40 W m-2 in
?typical? atmospheric conditions not associated with an inversion; i.e.
impact depends on low-level profiles of temperature, water vapor and
condensate.  Hsin-Mu Lin was responsible for this aspect of the
change package.

- Fix output solar fluxes

The valid time of fluxes from shortwave solar radiation output from the
Eta Model were off by one hour.  This was pointed out to EMC by fellow
scientists at NOAA/OAR/ETL who are gratefully acknowledged.  There was
no problem with radiation calculations in the Eta Model itself.  This
was just in the post-processing labelling.  Jim Wilczak (ETL), Tom
Black and Brad Ferrier were responsible for this aspect of the change
package.

- Fix to output precision of total atmosphere condensate fields

All output fields from the Eta Model are packed into the WMO standard
format for gridded data called GRIB (GRIdded Binary).  Packing involves
setting bit precision and this value was set too low when the total
atmosphere condensate fields were added to the post back in July 2003.
Without sufficient bit precision, there is no way to recover the proper
values when the GRIB is unpacked.  The fact that these were new output
fields meant that few users had started using them and they weren't
yet on the list of fields distributed via AWIPS SBN or NOAAPORT.
Geoff Manikin and Brad Ferrier were responsible for this aspect of the
change package.

- Additional output products:
Responding to requests (listed in parentheses) for additional output of
gridded products, the the ability to generate the following output
parameters has been added to the Eta Model post-processor:
- surface wind gust  (requested by OPC & AWC)
- lowest wet bulb zero height  (HPC & SPC)
- the pressure level of the lcl (SPC)
- a 3rd cin computed by lifting a mixed layer near surface (SPC)
- a 4th cape computed by searching for parcel in lowest 300 mb (SPC)
- a 4th cin computed by searching for parcel in lowest 300 mb (SPC)
- the pressure level of parcel lifted in 4th cape & 4th cin (SPC)
- height of pbl (Air Quality)
- rime factor from Ferrier scheme (ER, HPC & AWC)
Geoff Manikin, Brad Ferrier (rime factor) and Marina Tsidulko (pbl
height) were responsible for this aspect of the change package.
Contingent upon available resources on the NCEP and NWS TOC servers,
these grids may be accessible via anonymous ftp from these servers
at a future date.  Newly mandated advance notification will precede
any additions to datasets available online.  There were no changes
to any Eta model output available on AWIPS or NOAAPORT.  On-line
inventories of output grids remain available from
http://wwwt.emc.ncep.noaa.gov/mmb/mmbpll/etax.gridinv/ .

Jim Tuccillo (IBM) was responsible for the faster post-processor code.


Schedule for change:
These changes were implemented on 16 March 2004.


Description of testing:
Most components of this change package were tested individually or in
combination using EMC's EDAS/Eta parallel system at either 32 km or 12
km resolution. Eric Rogers maintains a comprehensive change log
at http://wwwt.emc.ncep.noaa.gov/mmb/mmbpll/paralog/  On this page one
can find links to the web sites with output and statistical summaries
for the different tests.

Statistical summaries:
Summaries of statistics on QPF, fit to raobs and fit to surface obs
can be found at
http://wwwt.emc.ncep.noaa.gov/mmb/mmbpll/pll12stats.etx_winter2004/

Anticipated impact of forecasts:
Based on both our own and outside subjective evaluations of forecasts
and the objective verification statistics, the changes described above
will lead to a) no discernible impact on QPF skill or bias, b) no
discernible impact on fit of forecasts to surface observations of
temperature, wind and RH and c) slight improvements in fit of forecasts
to upper air observations of temperatures, heights, RH, and winds.

Field evaluation:
Datasets and webpages were available for download and review,
respectively.  Feedback from the NCEP service centers participating in
the evaluation are quoted in the Director's briefing package


Points of Contact:

Eric.Rogers@xxxxxxxx      301-763-8000 ext 7227
Brad.Ferrier@xxxxxxxx     301-763-8000 ext 7290
Geoff.DiMego@xxxxxxxx     301-763-8000 ext 7221

Approvals:

NCEP                        EMC                     NCO
Director:                   Director:               Director:

Future changes:

I.  On or about 20 April 2004 we will modify our production schedule
to produce the 84 hour forecast in the same time period as we are
now producing the 60 hour forecast.  This will result in earlier
delivery of all Eta products.

II.  Another upgrade package is being readied for later in 2004.  It
will include an upgrade to the Land-Surface Model, a new shortwave
radiation scheme (same as that used in the GFS), modifications to
boundary layer and convective treatments (both deep and shallow).
A 3DVAR upgrade will also be associated with this change package in
which we turn back on the use of surface temperature observations
over land and begin use of new sources of observations: e.g. surface
mesonets, GPS IPW and boundary layer profilers. Implementation of
these and other changes are planned for the summer of 2004 and will
represent the final change package for the EDAS/Eta.  By the end of
FY2005 (30 September 2004), we expect to replace the 12 km EDAS/Eta
system with a 10 km system based entirely on WRF components.



  • 2004 messages navigation, sorted by:
    1. Thread
    2. Subject
    3. Author
    4. Date
    5. ↑ Table Of Contents
  • Search the nws-changes archives: