Over the years, we’ve repeatedly discussed the difficulties in forecasting thunderstorms and, in particular, severe thunderstorms. But even though it’s tough, we have an obligation to do it. These storms can be life-threatening.

Yet any time we forecast the chance of “severe” storms, it sets off alarms. People worry and sometimes change their plans — even when we aren’t totally confident that dangerous storms will develop.

Sunday was a case in point. For several days, we and other forecasters warned of the potential for violent storms in the D.C. area. But they missed to the north. Meanwhile, the Taste of Wheaton was canceled and no doubt many other outdoor events.

A problem is that some forecast consumers only pay attention to the headlines, i.e. the “potential for severe storms” and the scary-sounding impacts — like damaging winds, flash flooding, large hail and tornadoes. They don’t consult the fine print.

Consider all of the qualifiers we included in our discussions:

On Thursday: “Before anyone gets too alarmed, remember that forecasting severe storms in the Mid-Atlantic is challenging within a day of the actual event, much less four days out. It is certainly possible ingredients will not come together for widespread severe weather.”

On Friday: “Thunderstorm forecasting always carries with it uncertainty and severe weather is certainly not guaranteed. Storms, when they develop, tend to be hit or miss. … Unfortunately, whether days such as Sunday end up being big severe weather events, or big ‘busts,’ often boils down to nowcasting the day-of.”

On Saturday: “As always, thunderstorm forecasting is complex. Thunderstorms tend to focus their fury on small areas, so it can be difficult to give a broad look at expectations.”

On Sunday: “The main questions are whether storms will be scattered such that not everyone will see a big storm, or widespread such that the vast majority of locations get in on the action, and how many storms will reach severe category.”

Despite all of the qualifiers, some people who changed their plans based on the forecast were pretty frustrated when the storms largely missed. Some commenters expressed their outrage in our blog comment area. Others took to Twitter:

Addressing the problem

As much as we communicated the uncertainty, the forecast cannot be considered a success if the message we were trying to send — that big storms were possible but wouldn’t ruin everyone’s day — did not reach some people.

When our forecasts lead to bad decisions like they did Sunday, we need to address that by improving the way we communicate.

Traffic lights

A common reaction people have when they hear about the risk of severe storms is that the day is shot and they should change plans. But they may not realize thunderstorms are usually short in duration and hit or miss.

We could introduce a traffic light system to help people understand when they can go (green), should “stop” (red), or proceed cautiously (yellow) on thunderstorm days.

The yellow signal would display on the overwhelming majority of thunderstorm days, when outdoor activities are typically possible but weather-awareness is required. We would emphasize the importance of a Plan B on such a day. That is, have a designated place of shelter that can be reached quickly for outdoor activities.

On the rare day when we had high confidence of a washout and multiple rounds of storms, we could display a red light.

If we implemented such a system, I think it would be useful to break the day into at least two segments since morning and early afternoon hours may well support green light signals on many thunderstorm days.

Scenarios

It also may be useful — especially for complex thunderstorm events like yesterday’s — to develop scenarios and assign likelihoods for each scenario, like we do for winter storms. This would help people eager for more detail understand the full range of possibilities.

The challenge of lead time

Lastly, we wonder, given the uncertainty in forecasting thunderstorms, how far in advance should we be doing deep dives on the possibility? How soon is too soon for people to worry about them?

As we wrote about the potential for Sunday’s storms way back on Thursday, perhaps the repeated drumbeat built up an expectation that exceeded the actual threat.

Of course, our hand was forced in a way as soon as the National Weather Service Storm Prediction Center (SPC) wrote, that Thursday, that “a fairly widespread damaging wind event may evolve” Sunday. Could we reasonably ignore that?

When an authority like SPC sounds the alarm so early, we feel compelled as a media outlet covering weather to carry that message — in part because we want folks to be aware, but also because we know other weather outlets will start covering if we don’t. The information will be out there, but we wouldn’t be addressing it. The truth is, when other media outlets in the region begin to cover an approaching event, it forces our hand.

CWG severe weather expert Jeff Halverson described the situation as a “no-win” in an email:

This is a high impact hazard, one that kills. Part of the problem is that we are all now slaves to social media – word of a potential outbreak gets out there 3-5 days in advance, and no meteorologist in her/his right mind would talk about such things that far in advance. But social media forces us into early damage control, and talking about the matter earlier than we would feel comfortable with. SPC throws out an Enhanced 4-5 days prior for us, and Accuweather, The Weather Channel are going to run with it. Soon, everything is ablaze.

And believe it or not, we are facing the exact same challenge today. The Storm Prediction Center has highlighted the risk of severe thunderstorms in our region for this coming Saturday — six days away!


Severe storm threat outlook for Saturday. (NWS SPC)

It writes: “All modes of severe will be possible including hail, wind, and tornadoes.” But it also cautions: “This severe area is likely to change shape or location in subsequent outlooks, and predictability is always an issue this far out.”

We’ve heard all of this before.

So when is the right time to start messaging on this? And do we introduce scenarios and traffic lights?

We welcome your feedback.

Dan Stillman, Jeff Halverson, Ian Livingston and Angela Fritz contributed to this post.