AT&T dives deep into climate data

AT&T Broadband truck on the road
Shutterstock Sundry Photography
AT&T Broadband service vans driving on the freeway.

Adapted from the VERGE Weekly newsletter, published Wednesdays.

Did you know that AT&T has more than 30 full-time staffers in the United States and the United Kingdom alone — in charge of a fleet of close to 300 self-contained trailers — that can be deployed to respond to man-made and "naturally caused" disasters?

Indeed, the company has invested more than $600 million in its response program since 1992, alongside "billions" of dollars of investments in the telecommunications equipment network itself, according to the website describing the program. 

Now, the $170 billion telecommunications company is embarking on an ambitious data analytics initiative to paint a clearer picture, literally, of where its operations are most vulnerable to the future impacts of climate change.

As opposed to investments in renewable energy or energy efficiency that fall under AT&T’s efforts to mitigate its environmental impact, "this is part of our adaption strategy," Shannon Carroll, director of environmental sustainability at AT&T, told me.

At the center of the project, a collaboration with Argonne National Laboratory is a "climate change analysis tool" that layers maps of AT&T’s vast network of copper lines, fiber cable, cellular communications sites and network switching offices with climate modeling information from Argonne. The mashup has been used to create a graphic representation of how the predicted impacts of climate change — such as sea-level rises, high-intensity winds, and coastal and inland flooding — might affect operations 30 years into the future, providing insights on an hour-by-hour basis.

The company is piloting the tool in Florida, Georgia, North Carolina and South Carolina, where storms such as Hurricane Michael in 2018 wreaked havoc on mobile, land-line and internet services there.

"We knew it was time for AT&T to take a deep dive into what we were doing as a company to plan for the long-term impacts of climate change," said Carroll in a white paper with more details about the partnership. "We received an independent, unbiased and credible point of view. It provided validation of some thoughts and challenged others, which is a good thing."

The project, unusual in the industry for the proactive nature of its focus, actually was born in the 2015 timeframe, and it involved a cross-discipline team representing AT&T’s corporate social responsibility group, strategic planning, operations, network support and from within the company’s "chief data office." The Argonne scientists tapped to deliver the climate data needed for the project included experts in radar data, inland and coastal hydrology, advanced statistics, infrastructure modeling and engineering, and decision science.

Why look outside AT&T for this information?

Hint: It takes a lot of very powerful computers to crunch out models of this nature, which is where Argonne’s high-performance computing technology comes in. (Running the simulations for all four states for just one year would take more than 4,000 hours on a "standard" computer, according to AT&T.) The company paid for those services, but it hasn’t disclosed how much. 

So, what has AT&T discovered so far?

  1. The water levels during inland floods in all four states will be an average of 5 percent higher by mid-century if the world keeps emitting as much greenhouse gas as it currently does.
  2. Wind speeds of up to 90 miles per hour are more likely across a large part of southern Florida, both during hurricanes and other weather events, in the same time frame.
  3. Both "nuisance-level" and severe flooding will become more commonplace across all four states.

And that's just three generalizations. (Check out the white paper (PDF) for way more detail.)

While Carroll and his AT&T counterparts weren’t necessarily surprised by some of those revelations, it was eye-opening to have their suspicions validated and visualized on a map. It helps the organization more quickly see not just which equipment could be at risk in the future, but also which unexpected locations might represent opportunities for investment. 

"It really starts to become real for people," said Kim Keating, vice president of data science, machine learning and artificial intelligence at AT&T. "This will really help our network planning team think about how we grow in the future as we continue to invest in our infrastructure."

There’s more work to be done, but Carroll and Keating both suggest that the tool could be made available to other regions — and, in the future, wildfires could be part of the data set.

I haven't heard any of AT&T's peers discuss this sort of project publicly, so I poked around to get a sense of how the industry writ large is planning for climate change. While I didn't find any specific statistics, it's worth pointing out that a survey last year of big data center operators conducted by the Uptime Institute found that just 11 percent were taking steps to plan for increased flood risks, and only 14 percent were re-evaluating site location decisions with higher temperatures or water concerns in mind. That sort of complacency seems very dangerous indeed.