Follow Datanami:
October 24, 2016

U.S. Taps Data to Improve Forecasts


As they struggle to keep pace with more frequent outbreaks of severe weather, U.S. forecasters are looking for better ways of incorporating more data into storm and flood predictions.

U.S. storm forecasts have been criticized for relatively low resolution compared to European models that are leveraging exascale computing, analytics and storage. In response, the U.S. National Oceanic and Atmospheric Administration (NOAA) announced $6 million in research grants this week designed to help the National Weather Service (NWS) field new forecasting tools.

“This funding represents another important step to get new tools and technologies more rapidly into the hands of our weather forecasters,” John Cortinas, director of NOAA Research’s Office of Weather and Air Quality, noted in a statement announcing the grant program. The research office is managing the technology transfer effort in coordination with NWS.

Along with funding forecast technology efforts, NOAA said the tech transfer initiative would allow scientists and forecasters to collaborate so new forecasting tools work with existing weather service models and computer infrastructure.

Meanwhile, a grant competition will focus on improved forecasts models, new weather sensors and “better methods of assimilating data into weather forecast models,” NOAA said. Grants would be awarded to private and university researchers in areas such as improving rainfall estimates and flash flood warnings as well as leveraging radar data to issue real-time alerts on possible flooding from streams.

“Forecasters can use this data to enhance the current stream flood monitoring system,” NOAA said.

European weather models are widely viewed as having higher resolution and therefore greater accuracy because they integrate more data on, for example, atmospheric conditions. The U.S. effort also seeks to determine whether adding temperature and moisture data from ground-based instruments can improve thunderstorm prediction.

While previous NOAA grants covered a wide range of meteorological issues ranging from hurricane and tornado forecasting to air quality and precipitation forecasts, the latest research grants stress moving research tools into the field. These include understanding the role uncertainty plays in weather forecasting as well as transitioning automated tools that provide higher resolution data for predicting weather hazards.

Another focus of the technology initiative is leveraging data technologies to improve drought monitoring along with integrating observations of reservoir and lake levels into the National Weather Model.

In response to increased flooding, NOAA announced a new forecasting tool in August used to simulate how water moves through rivers and streams. That model running on a Cray XC40 supercomputer uses data from more than 8,000 stream gauges to simulate conditions in about 2.7 million locations across the country. The model generates hourly forecasts of the nation’s entire network of rivers.

“With a changing climate, we’re experiencing more prolonged droughts and a greater frequency of record-breaking floods across the country, underscoring the nation’s need for expanded water information,” stressed NWS Director Louis Uccellini.

Recent items:

Data ‘Assimilation’ Seen as Weak Link in U.S. Weather Forecasts

Weather Visualization is Powered by Big Data