Tuesday’s severe weather forecast was grim. So much so that schools in Oklahoma closed for the safety of students and faculty. A moderate risk of severe storms was in place, and a “potentially dangerous situation” tornado watch was issued late in the afternoon. One local station in Oklahoma City said that on a scale from zero to 10, Tuesday’s tornado risk was a “10.” EF-4 tornadoes were possible, they said.
That threat proved to be unfounded.
First, to be clear, this is not just a problem for the National Weather Service or the Storm Prediction Center. They are in charge of the outlooks, watches and warnings, but this issue spans the entire weather enterprise. Technically, the forecasts were correct and they verified — there was only a 10 percent tornado risk yesterday. They weren’t calling for widespread long-track tornadoes. They emphasized large hail and wind over the tornado threat.
But these details do not matter. If people in the Plains were expecting a huge tornado outbreak on Tuesday, then the weather enterprise did a horrible job communicating the forecast.
Things started to go off the rails as early as last week. Tuesday’s severe weather risk was in the forecast for days. The Storm Prediction Center’s long-range severe weather outlook is a great forecast product that allows meteorologists to start getting the message out early. The sooner we have a handle on the risk, the sooner people can start to prepare.
— Jim Cantore (@JimCantore) April 21, 2016
But there is a drawback to issuing such a strong outlook six days in advance. It allows the hype train to gain a lot of speed right off the bat. Trusted weather organizations were calling for the possibility of “wedge” tornadoes, nearly a week in advance. They took what should have been a simple heads up on possible severe weather — not just tornadoes, but large hail and dangerous winds — and morphed it into a tornado outbreak.
“The storms on Tuesday have the greatest potential to produce numerous tornadoes, including some strong (wedge) tornadoes that could be on the ground for an extended period of time,” AccuWeather wrote on Thursday (emphasis mine). This was a forecast of micro-scale storm structure and evolution made five days in advance.
Once a forecast like this is out there, meteorologists tend to double down on it. They want it to be right, because they want their forecast to verify, which only leads to communicating the extremes and not the uncertainty.
— Marshall Moss (@MarshallMoss) April 22, 2016
Once Monday rolled around, the Storm Prediction Center issued a moderate risk of severe weather for Tuesday afternoon but hinted at the possibility that they were considering upgrading to a high risk — the highest on the five-level scale. They never did make that leap, but just the whiff of a high risk is enough to send people into a frenzy.
There was another problem with the moderate risk, too. Since it came out the day before the event, no specific probability forecast on tornadoes, wind or hail was issued. This is standard — the Storm Prediction Center does not issue the forecast for, say, tornado risk, until the actual day of the event. What that leaves people with, then, is a pretty dire-looking forecast without the uncertainty context (unless you dig into the technical discussion weeds).
Even on Tuesday, when forecast models were indicating that a widespread, long-track tornado outbreak was not likely, and the Storm Prediction Center had issued their 10 percent tornado risk, one local media outlet was wildly, irresponsibly sounding the alarms, going against what is even feasible in the current state of the science.
SVR/TOR Outlook next 10 days K4 Viewing area. 1PM Tuesday Update pic.twitter.com/0Dx5CcsgHY
— Mike Morgan (@MikeMorganKFOR) April 26, 2016
My LATEST TOR Outlook. 1PM F4 TOR Psbl Today, ESP in Bright RED. OKC Threat 6PM-10PM Stay weather aware and 4warned! pic.twitter.com/UWMkEHosHh
— Mike Morgan (@MikeMorganKFOR) April 26, 2016
The final nail in the communication coffin was the PDS tornado watch — a “particularly dangerous situation” — issued late in the afternoon Tuesday.
Earlier in the day during a Twitter Q&A, someone asked the Storm Prediction Center what it would take for them to issue a PDS watch. The center responded, “high confidence in tornadoes EF-2 or stronger, with EF-4 or EF-5 possible, or large area of winds in excess of 75 mph.”
When people read that, they gravitate toward “high confidence” and “EF-4 or EF-5.” They interpret the watch exactly how it sounds, “a particularly dangerous situation” — more dangerous than your average tornado watch, which is quite frankly already dangerous enough.
Tuesday’s severe weather outbreak turned out to be more strong winds and large hail than tornadoes. In all, five tornadoes were were reported to the National Weather Service by Wednesday morning, although that number may increase by a few as teams go out to survey the damage.
Bill Bunting, the operations chief at the Storm Prediction Center, agrees that there was an issue with communication. But he also thinks it was a reflection of high expectations in situations where a moderate risk is issued.
“It was a reminder of, ‘How do you balance uncertainty with the appropriate risk?'” Bunting told the Capital Weather Gang.
The Storm Prediction Center’s forecast did verify. Severe thunderstorms popped in almost every case within their outlook. But they also realize that there’s a lot to learn from Tuesday’s challenging situation, including the PDS tornado watch — something that they thought the weather actually called for at the time.
“Obviously, in hindsight, had we known the outcome we would have done something different,” Bunting said. “We learn from every event, and I’m not going to go back and second-guess the decision that was made with the information we had.”
Rick Smith, the warning coordination meteorologist at the National Weather Service in Norman, Okla., stands by the forecast and information his office provided but is concerned the signal was getting lost.
“I think it gets to some point where people are not only looking at us, but they’re looking at TV and Twitter and every available piece of weather information they can get their hands on,” Smith said. “And there are different messages.”
That is the essence of the problem. Too many conflicting messages. Too much noise. It’s in these uncertain, life-threatening events that the weather community needs to come together to deliver a consistent outlook and communicate uncertainty.
I know hindsight is 20/20, but we have to do a better job. Unfortunately, some people will remember this event as a total forecast failure, and they just might be less inclined to believe us in the future.