Giant: Why U.S.
and Prediction are
Not Achieving their
Atmospheric Sciences The U.S. Weather Prediction
Enterprise is very big but
University of Washington uncoordinated.
April Bulletin of the AMS
The U.S. Weather Prediction Enterprise
Has Accomplished a Great Deal During
the Past Decades
Dramatic Improvements in
Numerical Weather Prediction
NCEP operational S1 scores at 36 and 72 hr
over North America (500 hPa)
36 hr forecast
72 hr forecast
35 10-20 years
1950 1960 1970 1980 1990 2000
Large Advances in Weather
Substantial Improvements in Weather Sensors
The Development of a Vigorous
Aggressive Use of the Internet and
other New Dissemination Approaches.
Substantial Advances in
Understanding of Weather Systems
and Basic Processes.
But Even With These Advances We
Have Accomplished Far Less Than Our
Potential and in a Number of Areas the
U.S. Have Lost World Leadership
This talk will suggest that our
progress has been undermined by
the inability of the major sectors of
the weather prediction community
to work together effectively.
• The skill of the GFS, the main U.S. global weather
prediction model, lags that of others (e.g., ECMWF).
500 mb Height--24h
The U.S. Weather Research Program—whose
goal was to coordinate and support the nation’s
weather research--is essentially dead, ended by a
lack of funding, vision, and interest.
• Major deficiencies exist in key forecast model
parameterizations (e.g., planetary boundary layer,
microphysics) and there is minimal community
coordination and joint research to deal with them.
• Disturbingly, the intellectual resources needed to
deal with such major problems are declining.
(How many PBL researchers are in our
department now versus 30 years ago?)
prediction is essentially
community has held to
paradigm, failing to
provide our users with
A new forecast preparation/dissemination approach
(IFPS) was developed by the NWS
that is completely deterministic and which burdens
forecasters with producing single renditions of
reality every six hours for seven days.
Exact Forecasts Out Several
Days at Every Location!
Starving Probabilistic Prediction
• U.S. efforts to create short-term, mesoscale
ensembles have been underfunded and
• Virtually no post-processing of ensembles
to produce reliable probabilistic forecasts
The Communication Deficit
• There has been virtually no research on how to
best communicate weather information.
• Current icons are inconsistent and often irrational.
• Icons and other approaches are used with little
understanding of how people interpret them.
And what does probability of precipitation really mean?
A large proportion of the population doesn’t really know.
Short-Term Forecasting: Our
• With improving radar, mesonets, short-term NWP,
and physical understanding, meteorologists know a
great deal of what is happening now and during the
next few hours….that is not communicated.
• Example: radar and MM5 indicate a profound rain-
shadow that will persist for the remainder of the day.
NWS and media provide forecast of “scattered
• There has never been as much discrepancy between
what meteorologists know and what information the
Example 2: Puget Sound Rain
Camano Island Radar 1-h Precip. Total
High Resolution MM5 Does Very Well With Rainshadows
NWS Forecasts Miss It Completely
EVERETT AND VICINITY-SEATTLE METROPOLITAN AREA-TACOMA
AREA-INCLUDING THE CITIES OF...EVERETT... EDMONDS...
DERAL WAY... KENT...RENTON…TACOMA ...LAKEWOOD...
330 AM PST SUN NOV 6 2005
.TODAY...SHOWERS LIKELY. HIGHS IN THE UPPER 40S. SOUTHWEST
WIND 10 TO 20 MPH.
TEMPERATURE / PRECIPITATION
EDMONDS 49 40 47 / 60 60 60
EVERETT 48 40 45 / 60 60 60
SEATTLE 48 41 46 / 60 60 60
BELLEVUE 49 41 48 / 60 60 60
PUYALLUP 49 40 47 / 60 60 60
TACOMA 48 39 46 / 60 60 60
Increasing Tensions Among the
• Tensions have developed among the
community, particularly at the interface
between the private and governmental
• The private sector feels that the NWS is
invading their territory with Internet and
wireless delivery, and is providing
“custom forecasts” to potential clients.
• A group of private sector companies Penn. Senator
convinced Senator Rick Santorum to Santorum
introduce legislation that would limit the (a.k.a., the Senator
NWS to warnings and collecting from Accuweather)
And Other Tensions
• NCAR and NCEP over the development of
a new national mesoscale model (WRF)
Inefficient and Poorly Organized
Research Structure in NOAA
• The government research laboratories that are
tasked to supply new scientific and technical
advances for the NWS are OUTSIDE of the
organization, under NOAA’s Office of
Atmospheric and Oceanographic Research
• Without direct management by the “user” agency
and with research tasks balkanized over many
labs, the research agenda has been inefficient,
developing technologies that are unneeded or
redundant, and not dealing with acute problems.
Lack of Critical Mass: Too Many
Weather Prediction Models
• NWS NCEP: Eta, GFS, Regional Spectral, WRF-
• Navy: COAMPS, NOGAPS, NORAPS
• NCAR: MM5, WRF-ARW
• Oklahoma: ARPS
• Colorado: RAMS
• MESO Inc: MASS model
• …and several more!
Too Many Models
• Some competition is good, and we don’t want to
put all our eggs in one basket, but we have clearly
divided the community between too many
modeling systems for our own good.
• The NWS and universities have generally used
different models. Eta has been the main
mesoscale model for the NWS, MM5 (and now
WRF) for the academic community. A real
problem for the transfer of research results to
The WRF Model Saga
• The WRF (Weather Research and Forecasting)
model was conceived as a way to bring the
academic and operational communities together
under one model infrastructure for mesoscale
• Plug compatible physics would be easy to switch
• Completely new model with better, numerics,
clean code and easily parallelized.
• The NWS and NCAR went in separate directions with
different dynamic cores.
– NWS: NMM core
– NCAR/Universities: ARW core
• The plug compatible physics that would allow various
cores to use the same physics has not panned out.
• The infrastructure…the software framework that it all
works in has proven difficult to modify and
• Decision making has been limited to NCAR and a few
• NWS and NCAR going separate ways on data
These Issues Have
Developed at the
Same Time as the
• The private sector has grown rapidly and
stands as an equal to the government and
• The National Weather Service is now a
much smaller piece of the pie.
• The boundaries between the sectors have
become more diffuse.
The Boundaries Are Fading
• All sectors now do NWP
• All sectors disseminate information to the public
– (even the universities through their web pages
and involvement with the media—e.g., Penn.
State produced weather page in the NY Times)
• All sectors are involved in data collection.
• Such overlap is really a very healthy development,
promoting creativity and cross-fertilization,
assuming it does not lead to conflict and tensions.
Still a Giant
• We (the U.S.) still have many things going for us:
– The world’s largest meteorological academic
– The largest and most successful private sector
– The largest governmental research community.
– Demonstrated great creativity.
• But we have not effectively combined our resources…both
intellectual and financial…to make the rapid progress of
which we are capable.
• Often overseas competitors take our technologies and beat us
to operational implementation (e.g., 4DVAR)
A Number of Groups Have
Realized We Have a Problem
Pielke and Carbone 2002
• The goals of the weather prediction
enterprise “are unlikely to be reached if the
community proceeds in a balkanized
fashion that has characterized it in the past.”
• “No organization or entity has embraced the
collective measure of responsibility for
improving forecast processes.”
National Research Council: 1998 Report on the
Future of the Atmospheric Sciences
“Today, there is reason for considerable concern about
planning for atmospheric research. No one sets the
priorities; no one fashions the agenda…. Thus,
BASC believe that a national research environment
requires a strong disciplinary planning mechanism.
... all partners in the atmospheric enterprise –in
government, in universities, and in a variety of
commercial undertakings– must join together as an
effective team focused on the future.”
NRC FairWeather Report (2003)
• Suggests that the various
components of the
needs to work together
• Provides some potential
approaches (e.g., AMS
set up a neutral arbiter)
Weather Coalition: Private
Sector/University “Lobbying Group” for the
New AMS Commission on the
Community Meeting on the
Weather Enterprise (July 2005)
How Do We
Some Concrete Suggestions
• NWS should establish an advisory committee that
is widely representative of the weather community
• NWS should amend its partnership policy not to
support forecast products for specific industries or
individuals unless dictated by law.
• The private sector must acknowledge the NWS
needs to use modern communication technologies.
• The AMS Commission must expand its role to
provide venues to discuss Weather Enterprise
The Community Must Develop Better
Mechanisms for Organizing and Coordinating
Development and Research for Operational
and User Needs.
• What are the deficiencies of the current
observing, modeling and data assimilation
systems and how will the resources, both
financial and personnel, be found to deal with
• How can duplication of effort be reduced?
One Approach: Expand the Role
of the Developmental Test Center
• The DTC was set up as “a facility where the
numerical weather prediction research and
operational communities will interact to accelerate
testing and evaluation of new models and
techniques both to improve the technology and for
operational implementation. “
• Supported by a number of groups (NCAR, Air
Force, NOAA, Navy) and is seen as a relatively
• Underfunded now, but could serve as the nucleus
of a true community effort.
• Establish an oversight board representative
of the entire community.
• Would hold workshops on important
mesoscale modeling topics
• Would establish active working groups on
model physics, data assimilation, etc. that
would prioritize and coordinate U.S. efforts.
“The fault, dear Brutus, is not in
our stars, but in ourselves”
The Northwest Example
• In the Pacific Northwest, a wide-ranging
group of Federal, state and local agencies,
academic institutions, and private sector
entities have combined resources to build a
regional weather prediction system, using
both high resolution modeling (down to 4-
km grid spacing) and mesoscale ensembles.
Northwest Modeling Consortium
• National Weather Service
• University of Washington
• USDA Forest Service
• Port of Seattle
• United States Navy
• U.S. Environmental Protection Agency
• Washington State Department of Ecology
• Puget Sound Clean Air Agency
• Washington State Department of Natural Resources
• Washington State Department of Transportation
• Seattle City Light
• KING TV
• Resources are pooled and decisions are made as a
• It took a few years of intense discussion for the
various sectors to understand and trust each other,
but this was achieved with very positive results.
• Working together the group has also created a
wide range of innovative applications in
Transportation Weather, Hydrology, Air Quality
Forecasting, and other areas.
Washington State DOT Traveler Information System