The afternoon before the blizzard predicted for New York City fizzled out, the National Weather Service knew the storm might not produce the epic snow amounts it had forecast. But because it didn’t want to “confuse” the public, it decided to continue emphasizing the worst-case scenario.
This was a well-intentioned but flawed decision that has the potential to damage public trust in weather forecasts.
The Weather Service has a responsibility to put out the best possible forecast. If forecasts are changing and it knows there is substantial risk they could be wrong, it needs to revise them and communicate the changes immediately.
On Monday afternoon, many computer-model forecasts had presented a clear trend toward less snow in the big cities from Washington to New York.
Most of them showed the storm tracking closer to the coast, which would draw in milder air and switch snow to sleet, ice and rain. However, the Weather Service’s Global Forecast System (GFS) model held onto very high snow amounts.
Rather than hedge, the Weather Service doubled down on its predictions for snowfall totals of 3 to 6 inches in Washington, 8 to 12 inches in Philadelphia and 12 to 18 inches in New York.
“[Weather Service] offices decided it was best to remain on the high-side of the forecast snowfall in the event the rain/snow line remained just east of the major cities, which was still a possibility,” it said in a statement.
It added that irrespective of the specific amount of snow, the combination of snow and sleet would be hazardous, and it didn’t want to signal a less-severe storm by lowering amounts. “Making a dramatic change in the snowfall forecast and risking having to flip back and forth could produce an unwelcome result of less readiness and vigilance,” it said.
The models forecasting less snow proved correct. 2.0 inches fell in Washington, 6.0 inches in Philadelphia, and 7.6 inches in New York — about half of what the Weather Service predicted.
The lower-than-predicted snow totals did not go unnoticed.
Politicians and thought leaders pounced. “I don’t know how much we should be paying these weather guys,” said New Jersey Gov. Chris Christie (R). “I’ve had my fill of the National Weather Service after seven and a half years, to tell you the truth.”
— Jacob DeFlitch (@WxDeFlitch) March 15, 2017
The influential media aggregator Matt Drudge tweeted, “What is going on with National Weather Service? Lots of misses piling up.” He added: “Overreaction by govts, bad forecasting … very troubling trend.”
Even before the storm hit, President Trump expressed skepticism about the forecast and was proved right, at least in Washington, Philadelphia and New York. “Let’s hope it’s not going to be as bad as some people are predicting,” Trump said. “Usually it isn’t.”
Trump: FEMA is ready for the blizzard
— Bradd Jaffy (@BraddJaffy) March 13, 2017
Of course, the Christie, Trump and Drudge statements were sweeping and unfair generalizations that totally failed to appreciate the excellent Weather Service forecasts inland from the big cities and great work the agency does day in and day out.
“I am deeply disappointed in the governor’s comments,” said Gary Szatkowski, who ran the Weather Service office serving Philadelphia before retiring last year. “They are unfair to meteorologists in general, and to the National Weather Service in particular.”
But such statements show missed forecasts in major metropolitan areas get noticed by thought leaders and have consequences. This is particularly unfortunate considering how well the Weather Service predicted the historic snowfall totals in interior portions of New York State and elsewhere.
The Weather Service, which has a mission to protect life and property, may have felt it was best serving the public by stressing the worst-case scenario for the big cities. But it’s a risky strategy that can cost credibility.
Trust is so important in weather prediction because, when it is eroded, the public may take forecasts less seriously in life-or-death situations.
The Weather Service doesn’t have to limit itself to communicating the worst-case scenario for the public to pay attention to a high-stakes forecast. The public is smarter than it is given credit for; it can understand uncertainty if it is explained well; and it appreciates knowing about changes to the forecast.
When Atlanta broadcast meteorologist Glenn Burns asked his viewers about the Associated Press report that the Weather Service decided not to revise its forecast even when presented with new information, many were insulted.
“We are not children,” said Jill Nelmark. “Give the most accurate forecast and accurate update.”
“It makes the NWS look less reliable for future events,” said Josh Walls.
“Give me the facts and trust me to make an intelligent decision,” said Kris Chandler.
“I think they should have been honest and said that it might not be as bad. But to still prepare in case it was,” said Suzanne Blanton.
The New York news blog Gothamist reacted to the AP report with this snarky headline: National Weather Service: Sorry, You’re Too Stupid To Trust With The REAL Forecast
When the weather is serious, the Weather Service has a captive audience eager to consume the best information. It must have tools in its toolbox to satisfy this appetite, including the ability to rapidly update the public on changing situations, explain the uncertainty and express the severity of the hazards.
This holds true if the evolving weather situation is growing more severe or less severe — and shouldn’t vary for either case. Would the Weather Service have resisted updating its forecast for this nor’easter had it been trending more severe in the big cities?
Meteorologists I spoke to agreed about the Weather Service’s responsibility to not withhold information.
“We should strive to provide our most current, up-to-date thinking, even if — perhaps especially if — there is a dramatic, last-minute shift in that thinking,” said Nate Johnson, a broadcast meteorologist in Raleigh. “Transparency in both our process and our communication builds trust between us and the people who use our forecasts, even when our forecasts don’t pan out. ”
“It’s important to tell everyone what you know. We must tell everyone what we know,” said Matt Lanza, a meteorologist who works in the energy sector in Houston. “That’s a lesson moving forward”
But Erica Grow, a broadcast meteorologist in New York, said she understood the Weather Service’s decision not to backtrack on amounts and continue to emphasize the storm’s severity.
“I personally think there is WAY too much emphasis on snow totals when it comes to winter weather forecasting,” Grow said. “Impact-based forecasts are more effective, and I think the Weather Service did their best to issue an impact-based forecast given the constraints of their current system.”
I understand Grow’s sentiment, and I think it’s really important to prioritize impacts over inches.
But, ultimately, inches are what are logged in record books and how storms are remembered and compared with one another. And the Weather Service is judged by how close it is to getting the inches right.
The Weather Service needs to try to be both accurate and communicate the storm hazards. It’s a cop-out to say you can’t do those two things at the same time.
Public perception is so important for the Weather Service right now. It is an agency run and staffed by extremely talented, committed people who do outstanding work. Yet its value is often underappreciated, especially by politicians. A 2014 Commerce Department report found “the valuation people placed on the weather forecasts they consumed was 6.2 times as high as the total expenditure on producing forecasts.”
The Trump administration is expected to propose steep budget cuts at its parent organization, the National Oceanic and Atmospheric Administration. A four-page memo outlining the proposed cuts, obtained by The Washington Post, showed a $60 million cut for the Weather Service.
The Weather Service must prioritize issuing the most timely, accurate and complete forecasts possible. The consequences of the appearance of missing high-stakes forecasts and lacking transparency in how it communicates are too huge not to.