The National Center for Atmospheric Research | UCAR | UOP
photo Home Our Organization Our Research News Center Education Libraries Community Tools

Quarterly logo

Extraordinary ensemble

TIGGE unites a world of weather-model output

by Bob Henson

Weather modelers on five continents have joined forces to produce a treasure trove for researchers. The THORPEX Interactive Grand Global Ensemble (TIGGE) involves ten operational weather forecasting centers around the globe (see map). Up to four times a day, each TIGGE center provides a selection of the numerical output from each member of its short- and medium-range ensemble weather prediction systems.

The result is an impressive “ensemble of ensembles,” with as many as 259 separate model runs from the ten centers. These are made available to researchers 48 hours after each forecast’s initial time. The built-in delay can be bypassed for major projects of special interest to THORPEX, though it can still take up to 36 hours after model initialization for large volumes of data to arrive. Once submitted, data requests typically take a few hours to fulfill, allowing time to process and transmit the data.

The three TIGGE archive centers (red dots) receive data from ten participating forecast agencies: Australia’s Bureau of Meteorology (BOM), China Meteorological Administration (CMA), Canadian Meteorological Centre (CMC), Brazil’s Centro de Previsão de Tempo e Estudos Climáticos (CPTEC), European Centre for Medium-Range Weather Forecasts (ECMWF), Japan Meteorological Agency (JMA), Korea Meteorological Administration (KMA), Météo France, US National Centers for Environmental Prediction (NCEP), and UK Met Office (UKMO). (Illustration courtesy Douglas Schuster, NCAR.)

TIGGE was designed at a workshop hosted in 2005 by the European Centre for Medium-Range Weather Forecasts, with the ten centers officially joining between late 2006 and early 2008. “We wanted to establish closer cooperation between the academic and operational worlds,” says ECMWF’s Philippe Bougeault. The output is now available from data archive ­centers at ECMWF, the China Meteorological Agency, and NCAR (see “On the Web”).

Each day some 240 gigabytes (billions of bytes) of data flow from the operational centers into the TIGGE system in a ­framework organized by ECMWF’s Baudouin Raoult. The archive now holds more than 100 terabytes (trillions of bytes). The interface between the centers and the TIGGE archive is based on the Internet Data Distribution/Local Data Manager protocols from UOP’s Unidata program. This interface has been specially configured for TIGGE by Unidata’s Steve Emmerson, Steven Chiswell, and Tom Yoksas, based on interoperable software provided to each data center.

To date, 70 users have downloaded more than a terabyte of data from NCAR’s TIGGE Archive Center, according to Douglas Schuster, who manages NCAR’s TIGGE hub. The access rate at NCAR should get a boost once a set of upgrades goes online this summer, says Schuster. He’s been ­working on the project with a number of other staff in NCAR’s Computational and Information Systems Laboratory, including David Brown, Luca Cinquini, Richard Grubin, David Stepaniak, Hannah Wilcox, and Nathan Wilhelmi.

By mid-2008, users of the NCAR portal will be able to specify grid resolution and spatial area across multiple models. “Once this becomes available, we expect our usage to increase markedly,” says Schuster. Later on, as part of TIGGE’s second phase (which still awaits funding), a data-access mechanism will allow users to download data directly from each of the ten participating centers using a common interface.

Young-Youn Park (Korea Meteorological Agency) carried out some of the first analyses of TIGGE’s grand ensemble while visiting Roberto Buizza at ECMWF. Park found that the tropics are the main area where forecasts will benefit from TIGGE. As summarized by ECMWF’s Bougeault, “Single-model ensembles underestimate the spread of actual conditions in the tropics. Multi-model ensembles such as TIGGE do much better.”

Happily for researchers, all of the TIGGE data are available for research and education at no cost through a simple electronic registration process, as David Parsons notes. Now on leave from NCAR, Parsons is working at the World Meteorological Organization in Geneva and serving as chief of the World Weather Research Program, which includes THORPEX and this summer’s THORPEX Pacific Asian Regional Campaign (T-PARC).
T-PARC, the first major field program to draw on TIGGE, will include a vast net of observations stretching across much of the western Pacific and eastern Asia. The study will examine the impact of typhoons that originate in the northwest Pacific, the world’s most prolific breeder of tropical cyclones. As these storms curve north and east, they often influence weather across the North Pacific, North America, and beyond. Forecasters will call on TIGGE’s compendium of model output to help guide data collection during T-PARC.

The ultimate goal of TIGGE, says Parsons, is to pave the way for even closer linkages between national and ­international forecasting centers, perhaps even including a full-scale ­prediction effort that incorporates the work of each center. Planners envision this Global Interactive Forecast System (GIFS), as it’s been labeled, as building on the first two phases of TIGGE, with an end-to-end system possible as soon as the mid-2010s.

Daniel Wildcat
David Parsons.
(Photo by Carlye Calvin.)

A GIFS working group has described the system’s potential benefits in a draft plan led by Zoltan Toth (NOAA). For instance, the system might find a 10% chance that a major typhoon will strike a densely populated Asian coastline eight days out. By marshalling a wide range of observing and modeling resources, the system would then zero in on the threat, issuing special forecasts with enhanced resolution. Even if the typhoon were to eventually miss the coast, this coordinated approach might be able to detect the change in course and sound the “all clear” up to 36 hours sooner than would otherwise be possible. That, in turn, could save millions of people from the expense and inconvenience of needless evacuation.

Achieving this ideal will take time as well as dedication. “It was not a trivial task to organize TIGGE,” says Parsons. He notes that the forecasting centers had to agree to share data, while research and operational scientists had to agree on what fields to provide and address various formatting issues.

Overall, says Parsons, “TIGGE is a success story for the World Weather Research Program and THORPEX. We hope it will help improve prediction, advance knowledge of the behavior of the Earth system, provide users with advanced weather products, and serve as a model for future cooperative efforts on other ­predictive problems, such as seasonal time scales.” ♦

On the Web

TIGGE (NCAR portal)




Also in this issue...
Untitled Document