Network News

X My Profile
View More Activity
The new Washington
Post Weather website
Jump to CWG's
Latest Full Forecast
Outside now? Radar, temps
and more: Weather Wall
Follow us on Twitter (@capitalweather) and become a fan on Facebook
Posted at 11:00 AM ET, 10/ 7/2010

U.S. lags other nations in global weather prediction

By Steve Tracton

Output from the European Centre for Medium-Range Weather Forecasts (ECMWF) weather model. Credit: College of DuPage.

The numbers are high, according to a new National Research Council report I first wrote about last month: 629 weather-related deaths per year between 1999 and 2008. And 96 weather disasters between 1980 and 2009, each causing at least $1 billion in damages with total losses of more than $700 billion.

No doubt, they would've been much higher if it weren't for accurate and timely weather forecasts. The NRC report, "When Weather Matters: Science and Service to Meet Critical Societal Needs," puts the average annual value of public weather forecasts and warnings in the United States at about $31.5 billion. Compare that to the the $5.1 billion cost of providing the forecasts and warnings (probably the most bang for the buck of any government investment!). And while (to the best of my knowledge) there are no reliable official statistics on the lives saved by weather forecasts, it's obvious that numerous souls owe their lives to tornado and hurricane warnings alone.

Despite the numerous lives and billions of dollars saved by weather forecasts every year -- not to mention the significant progress in observing, understanding and predicting weather during the past 15 years -- the report concludes that the "United States has failed to match or surpass progress in operational numerical weather prediction achieved by other nations and failed to realize its prediction potential; as a result, the nation is not mitigating weather impacts to the extent possible." ("numerical weather prediction", or NWP , is a reference to the computer models that meteorologists use as guidance when making forecasts).

The National Weather Service's National Centers for Environmental Prediction (NCEP) is the primary agency responsible for the development and operational implementation of NWP in the United States. While NCEP has made steady progress in global model prediction, so too have other countries.

In particular, the NRC notes that the "gap in model performance" between NCEP's Global Forecast System (GFS) and the global model of the European Centre for Medium-Range Weather Forecasts (ECMWF), universally recognized as the preeminent global model, "has not narrowed in the past 15 years." GFS is one of the main models that meteorologists in the United States use to help guide their forecasts. It also powers many computer-generated forecasts that appear online and elsewhere.

On average, a five-day ECMWF forecast is at least as good as a 4.5-day GFS forecast. This advantage may seem small, but can be quite consequential in providing advance and accurate notice of an impending major weather system. Furthermore, the GFS experiences more significant reductions in performance -- known technically as "dropouts" and colloquially as "forecast busts" -- than the ECMWF and other international models. In the words of a recent external review of NCEP:

"It is patently unacceptable for the United States - given its extraordinary need for accurate weather and climate information across all sectors of society - to operate a global forecast system that lags well behind those of other nations..."

As detailed in the NCEP review, the reasons for this are many and complex. An overarching issue is "woefully inadequate" human and computer resources.

Model development and running models routinely in real time, of course, is extremely computer intensive. In this arena, ECMWF has maintained a considerable advantage. ECMWF's computing resources, adjusted for NCEP's wider mission requirements, exceed NCEP's by an estimated factor of 6. While ECMWF's computer and human resources are focused on a single model framework, EMC's mission requires it to spread far fewer resources beyond the GFS to include regional models (e.g., the North American Model, NAM) and ocean prediction systems (e.g., WAVEWATCH).

Note, too, that ECMWF is scheduled for a major operational computer upgrade in 2011, while NCEP's next operational upgrade may not be until 2013. With an advantage in computer resources, the ECMWF model can run at a higher resolution than the GFS. That is, it can resolve and forecast weather phenomena (e.g., winter storms) and parameters (e.g., temperature and humidity) with greater detail than the GFS.

In regard to human resources, ECMWF also has a distinct advantage. Budget constraints impose limits on "permanent" (civil service) scientist positions at NCEP'S Environmental Modeling Center (EMC), which is the entity most responsible for model development and improvement. About half of EMC's staff is dependent on "soft money" (funding external to the National Weather Service). Consequently, support for model and model-related development is transient and can bring demands not totally consistent with EMC's priorities, thus complicating execution of long-term plans.

While NCEP's staff is larger in absolute terms, it is spread over a broader array of missions and responsibilities. The end result is that ECMWF has twice the staff working on its global model than does NCEP, consisting of a cadre of more-or-less permanent employees complemented by a predictable rotation of scientists from ECMWF member countries.

Not surprisingly, the NCEP review argues for EMC computer capabilities at least comparable to those of other major international centers and adequate human resources to achieve its mandated operational forecast mission. A prime example of the constraints imposed by insufficient computer resources is the inability to upgrade GFS ensemble products, which help to inform decision-making by better estimating the uncertainty in forecasts. (Note: the same handicap applies to NCEP's regional ensemble system, SREF)

The review also recommends that NCEP embrace a flexible, unified modeling approach that can be specialized for specific applications, and engaging (more so than now) partners from other agencies, academia and the private sector.

Also not surprisingly, the review goes on to find that any significant change in the status quo at EMC will require a substantial increase in NOAA funding of NCEP. Federal budget constraints likely preclude such an increase from coming to fruition anytime soon. This is most unfortunate, for the NCEP review concludes, "It is both urgent and absolutely essential to begin today in order to advance U.S. capability to an acceptable level in the decade to come."

Personal note: As a former research meteorologist at EMC, I couldn't agree more with the NCEP review's finding that "the level of EMC accomplishment is remarkable and laudable considering the parsimony of the resources made available" and that a staff of "significantly stressed" world-class scientists is continually asked/required to do more with less.

(CWG's Dan Stillman contributed to this post)

By Steve Tracton  | October 7, 2010; 11:00 AM ET
Categories:  Reports, Tracton  
Save & Share:  Send E-mail   Facebook   Twitter   Digg   Yahoo Buzz   StumbleUpon   Technorati   Google Buzz   Previous: Forecast: Sunshine and mild temps finally return
Next: Deluges dent dryness & drought


I've noticed recently that the "Euro" [ECMWF] gets more attention than the GFS when NWS issues its 3 to 7 day preliminay [morning] and final [afternoon] outlooks each day, though all models and model ensembles get taken into consideration.

However in certain localized situations the GFS and other models perform with greater accuracy than the Euro...there are also a few times when ALL the models "blow" a forecast [perhaps due to unavailable or inaccurate data at a critical point].

Posted by: Bombo47jea | October 7, 2010 1:22 PM | Report abuse


You are on target. In any given run other models, including the GFS, can prove more accurate, but on average the ECMWF is more likely to come out on top.

Yes, all models may experience a complete bust. And, as you suggest, more often than not this related to data issues

-Steve, CWG

Posted by: ensemblemean | October 7, 2010 2:50 PM | Report abuse

Amen brother. From your keyboard to NOAA's spreadsheet.

Posted by: imback | October 7, 2010 8:10 PM | Report abuse

The comments to this entry are closed.

RSS Feed
Subscribe to The Post

© 2012 The Washington Post Company