IBM’s Develops World’s Most Accurate Weather Forecast System

  • IBM develops the Global High-Resolution Atmospheric Forecasting System. 
  • It provides 3 times more accurate results than existing global forecasting models.
  • It offers forecasts of up to 15 hours for 25 million locations across the world.

The modern age of weather forecasting began with the invention of the electric telegraph in 1835. More than a century later, advances in the understanding of atmospheric physics led to the foundation of numerical weather prediction.

Today, we collect data from weather stations, buoys, ships, satellites, and other sources, and generate forecasts based on comprehensive regional perspective about the current weather, rather than just observing at a single point. However, this all leads to less accurate and low-resolution outcomes.

This lack of accuracy is due to mediocre weather instruments and systems that update infrequently, at resolutions that can’t capture several weather phenomena.

As a solution, IBM has developed the Global High-Resolution Atmospheric Forecasting System (GRAF) that updates hourly, at a 3-kilometer resolution. It offers three times more accurate results than existing global forecasting models.

Although we have had high-resolution weather models for a while, they can only provide short-term weather forecasts down to the country level. Up until now, a few areas in the developed countries had access to these high-resolution weather models. IBM GRAF brings that kind of detailed resolution to the entire world.

What Sets IBM GRAF Apart?

The two most popular global forecasting models — the United States GFS model and the European ECMWF model — split the Earth’s atmosphere into grid boxes that average 13 km and 9 km on one side. These models then solve the atmospheric flow equations to create a forecast for individual grid cells.

GRAF, on the other hand, employs a variable-resolution grid that can be configured with higher resolution over regions of particular interest. The grid looks like a honeycomb and its elements have an average resolution of 3 km over land areas, which is 3 to 4 times greater than the US and European models.

Indian monsoon forecast generated by a global weather model vs GRAF | Credit: IBM 

To make a top-notch model, IBM partnered with the National Center for Atmospheric Research, a US federally funded research and development center that uses start-of-the-art science to predict the weather at a truly global scale.

Source: IBM 

GRAF runs on an IBM POWER9-based supercomputer to handle update frequency and detailed resolution. It will use 3.5 petabytes of capacity from the IBM Elastic Storage Server to provide forecasts of up to 15 hours for 25 million locations across the world.

Applications and Availability

Regardless of where people live, GRAF can give a clear sense of exactly when the weather will hit around the world so that they can make better decisions. For instance,

  • It can help farmers know when rainfall is likely to occur so they can use pesticides and fertilizer wisely.
  • Airlines can efficiently route around turbulence.
  • Utility companies can effectively position outage repair crews for a storm.
  • Disaster management departments can better prepare for worse weather and maximize relief resources.

Read: ClimaCell Tracks Weather In Every 500-Meter Space Of The Earth’s Surface

Like other weather forecasting models, GRAF will not be directly available for public use. Instead, it will benefit people through the services offered by The Weather Company and IBM, such as Wunderground and The Weather Channel.

Written by
Varun Kumar

I am a professional technology and business research analyst with more than a decade of experience in the field. My main areas of expertise include software technologies, business strategies, competitive analysis, and staying up-to-date with market trends.

I hold a Master's degree in computer science from GGSIPU University. If you'd like to learn more about my latest projects and insights, please don't hesitate to reach out to me via email at [email protected].

View all articles
Leave a reply