Share this post:
Want to know the amount of rain two days from now in a specific location, down to a fraction of an inch, or the intensity of the rain? Our Deep Thunder weather technology was designed to deliver this type of hyper-local, specific weather forecast with upwards of 96 percent accuracy. And it’s about to become even more powerful.
Yesterday, our colleagues at The Weather Company (Weather) announced plans to advance the precision and accuracy of weather forecasting by combining Deep Thunder’s hyper-local, short-term custom forecasts with their own global forecast model. The combination of the two models will keep the name, Deep Thunder, and also use historical weather data to train machine learning models that will help businesses predict actual impact of weather events. Deep Thunder will become a unified weather model software platform for Weather that can be used for regional and targeted applications, on-demand and customized deployments, as well as on-going weather research and development to improve precision and accuracy of weather forecasting on a global scale.
With this new powerful model, utility companies that want to more accurately determine wind energy to meet customer demands, or farmers who require soil moisture forecasts to conserve water used in irrigation, will have the capability to use Deep Thunder, no matter where they are located.
Weather data everywhere
The Weather Company’s global weather modeling capabilities will help to expand Deep Thunder’s custom precision across almost any time zone and geography because it can pull data from almost anywhere: standard weather stations, weather radars, airplanes and spacecraft, even the 195,000 personal weather stations that people all over the world connect to the Weather Underground network. Their models ingest about 100 terabytes of third-party data every day. There isn’t a geography on Earth that Weather can’t report on.
If The Weather Company is the big picture of global weather, Deep Thunder is the cell under a microscope. It uses advanced physics to produce 24-84 hour forecasts for areas as small as 0.1 square kilometers). Plus, Deep Thunder can retrospectively forecast the weather – or “hindcast” as we atmospheric scientists would say. This enables the training of weather impact prediction models that use machine learning and cognitive techniques to forecast electricity outages from storms, or renewable power. It also helps companies better prepare for and understand weather’s impact on business.
This news marks an exciting opportunity to accelerate the science of weather forecasting precision and accuracy, and expand the Deep Thunder model to new challenges around the globe. Stay tuned as I write more about weather science advances and new Deep Thunder applications and capabilities in the coming weeks.