The company, Tomorrow.io, uses weather observations gathered by its own satellites and artificial intelligence platforms to provide weather forecasts and data to companies and government agencies like NASA and the National Oceanic and Atmospheric Administration.
Although Tomorrow.io has worked in the past with UAlbany's atmospheric researchers, the school and the company are now embarking on a new project they say could provide much more accurate and actionable weather forecasts for a variety of industries using a combination of high-performance computing and real-time weather data collected from both space and the ground.
That data from the ground is coming from UAlbany, which manages the New York State Mesonet early-detection weather observation system, a network of 127 atmospheric data collection sites placed across the state.
The plan under the partnership is to combine Tomorrow.io's satellite data, UAlbany's Mesonet system data and historical weather data to create AI weather prediction models that more quickly and accurately predict extreme weather events and dangerous winds, which research has shown can have devastating impacts not only on human life but also on economic activity.
A joint release issued by UAlbany and Tomorrow.io says that the new weather forecasting models would be extremely valuable to a variety of industries "such as aviation, energy, logistics, and emergency management to significantly enhance operational preparedness, from crosswind prediction and flight planning to optimizing wind power generation, cargo safety, and storm response."
The New York Mesonet stations, which include a 10-meter metal tower about the size of a telephone pole, are spaced about 19 miles apart from one another and include sensors and other meteorological devices that collect real-time weather conditions data. That data is fed back to a control room at the Atmospheric Sciences Research Center, which is located inside UAlbany's ETEC building at the Harriman State Office Campus.
Cole Swain, vice president of strategy at Tomorrow.io, believes the new partnership with UAlbany marks a "significant milestone" in advancing weather prediction modeling.
"The future of weather forecasting is poised to be fundamentally reshaped," Swain said in a statement. "We are setting the stage for dramatically enhanced precision and practical utility of weather predictions, driving the development of next-generation derivative models for critical decision-making."
Two of the UAlbany researchers working with Tomorrow.io are Jan Woodcock, director of operations of the New York State Center of Excellence for Weather and Climate Analysis, and Sukanta Basu, an Empire Innovation professor at UAlbany who is affiliated with both the Atmospheric Sciences Research Center and the university's Department of Environmental and Sustainable Engineering.
The researchers spoke to the Times Union on Thursday about how vital weather and climate prediction tools have become to virtually every industry and government across the planet, which has spurned a race by private companies to develop the most accurate models that are the most cost-efficient.
Weather, Woodcock said, has more of an outsized impact on the global economy than people might imagine. Extreme weather events can wreak havoc on crops, shipping, the travel and tourism industries, health care. Even global finance as insurers and reinsurers grapple to cover billions of dollars of losses after major weather events.
"Fifteen years ago, 90 percent of all weather analysis was done by governmental agencies," Woodcock said. "What's happened (since) is that weather has become more and more of a critical tool for industry to use to manage more effectively and more efficiently. So weather affects over 35 percent of the gross national product. So it's a very big number, right?"
Take an electric utility, for example. In this age of ubiquitous electronic devices, businesses and consumers alike cannot function without a reliable supply of on-demand electric power. Outages become major emergencies, and so when major weather events are forecast, electric utilities must prepare large teams of linemen and bucket trucks to prepare to restore power wherever the storms may land and cause damage that knocks out the power.
Due to the limitations of current forecasting methods, especially when it comes to predicting the wind, an electric utility is often forced to send large truck crews to two different sites since the models can't exactly pinpoint where the storm will land or where the wind will be the worst, Woodcock explained.
"What they do today is they send trucks to both locations," Woodcock said. "And the one that the storm doesn't show up at is called a false positive. And it costs millions of dollars because each one of those trucks costs a couple thousand dollars a day."
Basu said that AI and high-performance computers are changing weather prediction so quickly because so many more models can be run at the same time that it gives users so much more predictability, especially when more real-time data is used that can detect things on the ground as they are happening.
"What an AI-based system does, it allows you to run thousands of different scenarios. Not just 10 to 50 but rather thousands of different scenarios, and that is actually really, really powerful," Basu said.
Although the exact scope and timeline of the partnership were not revealed in detail, Tomorrow.io said in a release that the goal is to "rapidly deploy, train, and scale advanced machine learning models, with the goal of optimizing performance to surpass traditional physics-based forecasting."
©2025 the Times Union (Albany, N.Y.). Distributed by Tribune Content Agency, LLC.
