Cooperative Institute for Research in Environmental Sciences at the University of Colorado Boulder

Near-term weather forecasts get powerful boost from new computer model

Starting today, a sophisticated new weather forecast computer model developed by scientists from the Cooperative Institute for Research in Environmental Sciences (CIRES) and NOAA will help improve predictions of quickly developing severe weather events such as thunderstorms, winter storms and dangerous air turbulence.

The Rapid Refresh model was developed at NOAA’s Earth System Research Laboratories (ESRL) in Boulder, Colo. in collaboration with NOAA’s National Centers for Environmental Prediction (NCEP) in Camp Springs, Md. It provides NOAA’s most rapidly updated weather forecast, updating every hour with a new forecast extending out 18-hours for North America. Such forecasts are especially important in aviation, where fast-developing weather conditions can affect safety and efficiency, but they are also important for severe weather and energy-related forecasting.

“When accurate and timely weather modeling is needed most, the new Rapid Refresh model delivers,” said Louis Uccellini, Ph.D., director, NOAA’s National Centers for Environmental Prediction, a part of NOAA’s National Weather Service. “This new tool ensures that forecasts are the best they can be by using the latest science and computer techniques in effort to create a more weather-ready nation.”

The United States is the only country in the world that updates computer model forecasts every hour using the latest observations from an extensive network of ground- and satellite-based sensors, radars and aircraft, said CIRES Fellow Stan Benjamin, lead developer of the new model and a research meteorologist at ESRL.

The new model was tested extensively, running experimentally for 22 months at NOAA’s NCEP, and will replace the older rapidly updated model, Rapid Update Cycle (RUC). In comparisons with that older model, the new Rapid Refresh (RAP) performed well. “Overall, RAP provides equal or better forecasts than RUC for all variables, from winds to precipitation,” Benjamin said.

For example, RAP performed significantly better than its predecessor in forecasting heavy rain that pounded the Midwest in June last year. For one 24-hour period of comparison, RAPs forecasts correctly projected more of the regions that received 2 inches or more of precipitation.

RAP’s skillful forecasts derive from three key improvements over the earlier model: RAP is based on a significantly more advanced numerical weather prediction model, the Weather Research and Forecasting (WRF) model. WRF was created through a collaboration of NOAA, the National Center for Atmospheric Research, the Air Force Weather Agency, and dozens of other research institutions; RAP uses an innovative technique for “assimilating” current observations used to start the forecast model. The newer assimilation technique, developed through a NOAA-NASA research partnership led by NOAA’s NCEP, improves short-range forecasts; and RAP extends the geographical coverage of NOAA’s weather situational awareness information to all of North America, not just the contiguous U.S. as was the case for the RUC model.

The addition of Alaska, where air travel is common, is particularly important. “The majority of Alaska is not connected to a road network, so the only means of transportation to many locations is by air,” said Gene Petrescu, science and technology transfer meteorologist at NOAA’s National Weather Service in Anchorage, Alaska. “We are hearing from our forecast offices that they see value in these hourly forecasts.”

CIRES scientists instrumental in developing the model along with Stan Benjamin are Curtis Alexander, Ming Hu, Tanya Smirnova, Joe Olson, Patrick Hofmann, Eric James, Bill Moninger, Xue Wei, Steven Peckham Media Contact: Karin Vergoth,, 303-497-5125

Recent Stories