The security breach plastered high-profile accounts with solicitations for followers to participate in a cryptocurrency scam. It allowed false tweets to be posted on the timelines of politicians and celebrities, and forced Twitter to temporarily limit the functionality of all verified accounts.
Verified accounts were unable to tweet between about 6:20 p.m. and 8:40 p.m. Eastern time on Wednesday. That affected the accounts of National Weather Service forecast offices, some of which were in the midst of severe weather coverage. The National Weather Service office covering parts of Illinois, for example, was tracking a large tornado about 5 p.m. local time.
The NWS’s St. Louis forecast office tweeted several warnings and interacted with followers to take in storm reports through 5:06 p.m. Central time; the next tweet came more than two hours later, at 7:35 p.m.
That office was unable to tweet a tornado warning issued at 5:13 p.m. for portions of Clinton, Monroe, Washington and St. Clair counties in south central Illinois. Fortunately, no tornado touched down with that storm.
Meanwhile, a second dangerous storm was brewing farther north. The St. Louis office issued a tornado warning at 4:38 p.m. Central time for northern Macoupin and Montgomery counties, and was able to tweet that to its nearly 32,000 followers — along with several updates. A tornado was first reported in Macoupin County, which is located in the St. Louis office’s county warning area.
But by the time that storm — now accompanied by a tornado — moved into the territory covered by the NWS forecast office in Lincoln, Ill., that office appeared to be unable to tweet any warnings to its nearly 18,000 followers. Between 4:49 p.m. and 6:53 p.m. Central time, no tweets emanated from the office.
Typically, NWS offices use Twitter to disseminate warnings and also receive some storm reports, but the agency, which is part of the National Oceanic and Atmospheric Administration, views Twitter as a supplemental way to distribute potentially lifesaving weather alerts, rather than a primary way to reach the public.
The tornado affected parts of central Illinois south of Springfield. The National Weather Service in Lincoln issued a tornado warning for portions of Christian and Sangamon at 5:26 p.m. Central time, but the office’s Twitter feed remained silent. Weather spotters had confirmed a damaging tornado before the warning was issued.
“I was doing the warnings yesterday, so when I issued the initial warning, I said, ‘Confirmed spotter reported tornado,’ ” said Ed Shimon, an NWS meteorologist at the Lincoln office.
Tornado warnings were still disseminated via the emergency alert system, wireless emergency alerts and local broadcasters.
The Lincoln office did employ a workaround when partial functionality appeared to allow it to retweet other tweets. The office had previously constructed an unverified Twitter bot that tweets warnings in parallel to whatever the office issues. The main NWS Lincoln verified account retweeted a dozen warnings, updates and storm reports from that unverified account between 6:53 p.m. and 7:36 p.m. Central time.
“The bot is usually our automated way to get it out,” Shimon said. “We had to manually retweet that.”
The tornado doesn’t appear to have caused much damage and has not yet been assigned an intensity rating. “The emergency manager drove all the way through there. He said he didn’t see any corn crop damage,” Shimon said.
Although the Twitter outage was inconvenient in that it prevented weather forecast offices from maximizing the reach of their severe weather alerts, the agency stressed that social media is one of myriad avenues it uses to get warnings to the public.
“People can receive weather forecasts and warnings in a number of ways,” Susan Buchanan, a public affairs specialist at the National Weather Service, wrote in an email statement to The Washington Post. This includes “Wireless Emergency Alerts on their cell phones, weather.gov, the NWS mobile site, NOAA Weather Radio, local television and radio stations and their websites, private weather companies and apps, and on social media.”
Buchanan noted that the Twitter outage “did not affect … forecast operations,” and warnings flowed as normally through other channels.
One official National Weather Service Twitter account, a verified “bot” that tweets computer-generated graphics of tornado watches and warnings, along with their expiration times, appears to have continued tweeting relatively unimpeded during the outage.
The @NWSTornado automated account tweeted the 5:38 p.m. tornado warning the moment it was issued at 5:37 p.m., specifically mentioning Divernon, Ill. — where the large tornado was seen — in the tweet.
The National Weather Service stated that it does not have plans to create any non-verified emergency accounts in response to the Twitter hack.
Earlier this month, NOAA itself suffered a security breach when mPING, a weather-reporting application operated through a joint venture between the NOAA National Severe Storms Laboratory and the University of Oklahoma, was compromised and displayed numerous false weather reports. That service remains temporarily offline.