Nieman Foundation at Harvard
As the warming climate causes extreme weather events to become more frequent and severe, your everyday concern about weather risks may be growing. And you might not only be preoccupied with the chances of extreme weather where you live, but also in places you’re traveling, or where your friends and loved ones reside.
What if you received a brief email alerting you to a chance of extreme weather, and explaining the level and type of risk, where your daughter is working a summer job, or where your father lives in a retirement community, or where you’re planning to travel for a weekend? And what if you could get a single email summarizing any low, moderate, or high level of extreme weather risk in all of those places?
The New York Times believes such a system could be useful to you; this week, the organization launched a newsletter that allows readers to select up to four places in the continental United States and receive a morning email if there’s a risk of any of four types of extreme weather in the next three days — excessive rain, tornadoes, high winds, or hail.
Readers do not have to be paid Times subscribers to get the Your Places: Extreme Weather newsletter. It’s the latest example of the Times’ push into personalized newsletters, and of the Times’ experimentation with and emphasis on weather data.
The New York Times has launched Your Places: Extreme Weather, a newsletter about extreme weather risks in places important to you. Sign up today and we’ll let you know when forecasters see a risk of tornadoes, hail, high winds or excessive rain. https://t.co/kNq0ixXLbc pic.twitter.com/ki0V6FlDbO
— The New York Times (@nytimes) May 31, 2023
This newsletter is the brainchild of the Times Weather Data team, which the organization began building last year by hiring John Keefe as editor of weather data last summer to lead “a new team in the newsroom focused on making The Times a destination for extreme weather coverage.”
Keefe, who was previously a senior data and visuals editor at CNN (and who has also written for Nieman Lab), told me that he recalled talking with Times meteorologist Judson Jones about the National Weather Service’s “multi-day outlooks,” which the service uses to predict severe weather in what Keefe called “blobs” of regions across the country. “We were talking about how incredibly accurate they are, and how no one is really using that data to its full potential,” Keefe said. While anyone in the weather business relies on this data constantly — and the Times would use the data internally to determine whether and when to check in with reporters across the country — it was not being distilled and brought directly to non-experts, Keefe explained. “We just thought, hey, this would be great to bring the readers,” he said.
“To our knowledge, we weren’t aware of anybody doing it like that — where you could say, ‘Hey, let me know when one of these blobs, basically, is over a city I care about,” Keefe added.
When he pitched the idea to the Times, Keefe said he used an example of New Orleans, which the National Weather Service showed having an elevated risk of tornadoes that day. Several hours after he made that pitch, a tornado hit the city’s outskirts. “It was just a perfect example to say, ‘look, they’re good at forecasting these things; we can help bring that information to our readers,’” Keefe said.
The goal of the newsletter is to inform planning and enhance “situational awareness” of extreme weather risks in any combination of places readers care about, “whether it’s their vacation places, or where they have kids or parents or loved ones,” Keefe said. He emphasized that these emails are not substitutes for highly time-sensitive emergency alerts, like the kind you’d receive on your phone; because the email is only sent out once a day on days when a risk is detected, readers will not receive alerts about evolutions in weather patterns later in the day, though they can check the Times’ Extreme Weather dashboard for updated information.
“It’s not an immediate ‘Oh, I need to head to the basement’; it is, ‘Huh, we have this picnic planned for the day after tomorrow, maybe I should reconsider that’ — or travel or whatever it might be, just to be aware,” Keefe said. But these alerts, by heightening general awareness, could make a reader more likely to look out for developments later in the day, or to watch for an emergency notification, he added.
The team settled on allowing the selection of a maximum of four places to try to achieve a balance “between what we could design for the page and in the email and for mobile, versus giving enough of a selection” to be useful, Keefe said. To define and standardize the three tiers of weather risk, the Times consulted with experts about which terms to use and researched how factors contribute to the National Weather Service’s more varied categories. The team tried to make its three risk level categories “a little bit more streamlined” and easy for readers to grasp, Keefe said. Essentially, the lower tier means “at least some chance of extreme weather”; medium means “it is likely that damaging weather will happen”; and higher means “extreme, dangerous weather is expected” in a given area, per the Times’ dashboard.
The weather data team at the Times worked “in earnest over the past many weeks” to develop the newsletter in collaboration with other teams, including the graphics, digital news design, and interactive news teams as well as the messaging and personalization team on the product side, Keefe said.
The newsletter is “100% automated” — the “geospatial cross” of designated extreme weather patterns with any of a user’s selected locations automatically triggers a customized email — but Keefe emphasized that it does not rely on any artificial intelligence.
“The base concept is pretty straightforward,” Keefe said. “Pick a place; is one of your places in the zones that ha[ve] this risk; send their emails.” But “operationally, it’s far more complicated” in terms of infrastructure and design challenges — requiring bringing in massive quantities of data, distributing tailored summaries to a “New York Times-sized audiences” and making sometimes complex and dense weather data understandable.
The emails are all signed by Jones, the meteorologist. As part of the preparation for creating this product, Jones drafted several descriptors that could be plugged into the email by the automation process — descriptors for a lower risk in one place and higher in two other places, for instance. The only manual tweaking of these automated emails, Keefe said, will come for the high-risk weather patterns — Jones might manually add context about these, and link to Times articles, because that’s where readers are most likely to want and need additional information.
The weather newsletter concept also builds off the Times’ geographically customized Covid-tracking email, Keefe said — though that email was sent on a regular schedule regardless of Covid levels in each place, whereas the weather team is “only sending emails to people where the risk exists” and is detected by the National Weather Service. This presented its own implementation challenges and “turned out to be a puzzle that needed to be solved,” Keefe said.
He described the new email as a “next step in personalization” where the sending schedule is determined by data conditions rather than, for instance, when a writer finishes a column.
The team hopes to expand the service to also include data for places beyond the continental U.S. and internationally, and to track other types of extreme weather, Keefe said. But the focus was getting the service up and running now, and expanding later.
Weather alerts, Keefe noted, have long been the purview of local news organizations. But “a lot of readers across the country are in situations where…the local news that’s available to them is dwindling.” In that context, the team agreed that “providing more geographic diversity in our service” is a great benefit of this project.
