NOTE: The nws-changes mailing list is no longer active. The list archives are made available for historical reasons.
Posted: Mon, Dec 7, 1992 5:51 PM EST Msg: EGJC-5475-7583
k.mielke(rec),
g.trapp(rec), j.kemper(rec)
CC: l.miller.ucar(rec), t.whittaker(rec)
Subj: 2 Nat'l Tech Info Messages
I sent out 2 Nat'l Tech Info Messages today (both follow).
--------------------------------------------------------------------------
WSHPNSNMC
NOUS41 KWBC 071857
NATIONAL TECHNICAL INFORMATION MESSAGE 92-12
NATIONAL WEATHER SERVICE HEADQUARTERS
345 PM EST MON DEC 7 1992
TO: ALL NWS OFFICES, USERS OF NMC'S MEDIUM RANGE MODEL
FROM: MARY M. GLACKIN
CHIEF, SERVICES DEVELOPMENT BRANCH
SUBJECT: CHANGES TO THE MEDIUM RANGE FORECAST (MRF) SYSTEM
(THE FOLLOWING INFORMATION WAS PROVIDED BY DR. EUGENIA KALNAY OF
THE DEVELOPMENT DIVISION.)
At 0000 UTC, December 9, 1992, the National Meteorological Center
(NMC) will implement an "ensemble" configuration for the Medium
Range Forecast (MRF) global system, replacing the single 10-day MRF
by a number of forecasts generated from slightly perturbed initial
conditions. This change, which involves the truncation of the
resolution of the MRF run from T126 to T62 at day 6 and beyond,
will affect only the "day-8" product (average of the day 6 to 10
forecast), and will be essentially transparent to the users.
The change is based on the need to perform more than a single
forecast for medium range in order to a) improve the forecast by
averaging different solutions, and therefore filtering uncertain
forecast features, and b) provide regional estimates of the
forecast skill, by considering how well the different forecasts
agree with each other. It will also form the basis for an
objective probabilistic forecasting system.
Currently the operational global model, with T126 (105 km)
horizontal resolution, and the lower resolution T62 (210 km)
model are run at 0000 UTC for 10 days. The T62 model runs about
9 times faster than the T126 model, but is slightly less
skillful. However, experiments performed by Dr. Steven Tracton,
from the Climate Analysis Center (CAC), showed clearly that the
use of the high resolution T126 is important for the first 5 days
of the forecast, but that beyond 5 days, the use of the lower
resolution does not degrade the skill or change the forecast.
Based on these considerations, a new "ensemble" configuration has
been designed, truncating the resolution of the MRF run from T126
to T62 at day 6.
Although the December 9 change will open the possibilities of
using advanced techniques for ensemble forecasting, until these
techniques are developed and tested, the change WILL NOT impact
MRF guidance.
SENT - W/OM23 - C. ALEX
----------------------------------------------------------------------
WSHPNSNMC
NOUS41 KWBC 072115
NATIONAL TECHNICAL INFORMATION MESSAGE 92-13
NATIONAL WEATHER SERVICE HEADQUARTERS
515 PM EST MON DEC 7 1992
TO: ALL NWS OFFICES, FACSIMILIE USERS, AND OTHER USERS OF
NMC'S REGIONAL ANALYSIS AND FORECAST SYSTEM (RAFS)
FROM: MARY M. GLACKIN
CHIEF, SERVICES DEVELOPMENT BRANCH
SUBJECT: CHANGE TO THE INITIAL STATE OF SNOW COVER AND SEA ICE
IN THE NESTED GRID MODEL (NGM)
[THE FOLLOWING INFORMATION WAS PROVIDED BY DR. KEN MITCHELL
OF THE REGIONAL AND MESOSCALE MODELING BRANCH OF THE NATIONAL
METEOROLOGICAL CENTER (NMC).]
Beginning with the 1200 UTC cycle of the Regional Analysis and
Forecast System (RAFS) on Tuesday, December 8, 1992, NMC is
implementing new procedures for specifying the initial state of
snow cover and sea ice in the NGM. This summary describes the
reasons for, content of, and impacts of these new procedures.
Since first operational on March 27, 1985, the NGM has used a
National Environmental Satellite , Data, and Information Service
(NESDIS) weekly analysis of northern hemisphere snow cover and
sea ice (performed on Mondays) as the source of its initial state
of snow and ice. The primary problem with the NESDIS analysis is
its weekly update interval, which frequently misses those
significant changes in snow cover that can occur over large
regions on daily time scales. The resulting erroneous presence
(absence) of snow cover in the NGM contributed to negative
(positive) temperature forecast errors in the lowest several
layers of the NGM (e.g. the T1 and T3 fields in the NGM FOUS
guidance). Such errors hinder, for example, the proper
designation of the rain/snow boundaries by forecasters.
A secondary problem with the NESDIS analysis is its 190.5 km
resolution (true at 60 N, versus 90.75 km at 60 N in the NGM).
This coarse resolution led to special problems in the NGM in
coastal areas, where interpolation of the NESDIS analysis to the
NGM grid using the NGM land/sea geography resulted in snow cover
over land being spread as sea ice over nearby water and vice
versa. This problem was especially prevalent over the Great
Lakes.
The above problems are described further in the June 1989 issue
of "Weather and Forecasting" (published by the American
Meteorological Society) in the NMC Note by R. Petersen and J.
Hoke.
Beginning on December 8, the NGM's initial snow-cover analysis
will be derived from the U.S. Air Force daily northern hemisphere
snow cover analysis, produced operationally by Air Force Global
Weather Central on a polar stereographic grid with a resolution
of 47.6 km at 60 N. This daily analysis, valid at 1200 UTC, is
completed around 1700 UTC and received at NMC in time for use
first by the next NGM 0000 UTC cycle. Previous tests of the NGM
using the daily, high resolution Air Force snow-cover analysis
have demonstrated improvements in the NGM low-level temperature
forecasts.
Since sea ice does not change as rapidly as snow cover in
general, the weekly NESDIS analysis will continue to be the
source of the NGM's initial sea ice. A strength of the NESDIS
analysis is its routine input of the detailed sea-ice analyses of
the NOAA/NAVY Joint Ice Center in Suitland, MD.
However, the interpolation of the NESDIS sea-ice analysis to the
NGM grid has been been improved significantly. Use of land/sea
geography files on both the NESDIS and Air Force analysis grids
are being added to the codes that interpolate these fields to the
NGM grid. Their use in the new interpolation codes, along with a
modest correction of the NGM land/sea geography field over the
Great Lakes, virtually eliminates in the NGM the erroneous
spreading of coastal snow cover as sea-ice and vice versa.
(Similarly, recent changes by NESDIS in their analysis procedures
has greatly reduced a similar coastal spreading problem in the
generation of the original NESDIS analysis.)
The above NGM changes are were made to an NGM preprocessing step
that generates many of the NGM input surface fields. No changes
were made to the NGM forecast code, consistent with the fact that
the configuration of the NGM forecast model was "frozen" on
August 6, 1991. In particular, the NGM still keeps its initial
snow cover and sea-ice fields fixed throughout the 48-hour NGM
forecast.
In a similar context, the above changes are not expected to
change the NGM's systematic behavior (seasonal biases) inherent
in the NGM Model Output Statisitics (MOS) guidance (the FWCs on
AFOS/ FOUS14 KWBC for external users). Previous NGM use of the
weekly NESDIS snow-cover analysis was equally likely to result in
either erroneous snow presence or absence. The switch to the
daily Air Force snow cover will reduce the standard deviation of
the error of NGM low-level temperature forecasts, but not their
systematic bias over a season.
The NMC global spectral model (both Aviation and Medium Range
runs) and the NMC experimental ETA model continue to use the
NESDIS weekly analysis for their source of snow cover and sea
ice. Use of the daily Air Force snow analysis is expected in the
ETA model in about a month, and in the global model within a
year.
SENT - W/OM23 - C. ALEX
---------------------------------------------------------------------
nws-changes archives: