Meteorologists simulate issuing warnings with experimental software at the National Weather Service's Hazardous Weather Testbed. (NOAA)

We’ve all heard them — the blaring alerts that activate our cellphones or television when a severe weather warning is issued.

Perhaps our favorite weather app sent us a push notification, or we saw a television meteorologist pointing at vibrant boxes on a weather map. Whatever the medium, weather warnings have a way of finding us, especially whenever a severe thunderstorm is close by. Now those warnings, specifically the way in which they’re generated, are in the process of getting a makeover.

Severe weather warnings are issued for individual thunderstorms; before 2007, entire counties would be alerted at once. Over the years, weather warnings have become more targeted — but one warning can still cover an expansive area. Moreover, conditions can vary wildly even within the region enclosed by a single warning.

The National Weather Service is hoping to change that.


A photo from National Weather Service training material shows forecaster guidance on how and when to issue severe weather warnings. (NOAA/NWS)

Kodi Berry leads the program that’s updating warnings at the National Severe Storms Laboratory in Norman, Okla. The Forecasting a Continuum of Environmental Threats program, or FACETs, is an endeavor the National Weather Service is pursuing to communicate the hazards posed by severe thunderstorms on a hyperlocal level.

Berry says the goal is provide a more continuous flow of information for those who need it the most.

More precise warnings

According to the National Severe Storms Laboratory, FACETs aims to improve weather watches and warnings to provide “detailed hazard information through the use of ‘threat grids’ that are monitored and adjusted as new information becomes available.”

Typical weather warnings are issued in the form of polygons digitally drawn on a map. If you’re within the polygon, you’re alerted and urged to take action — such as seeking shelter. But just a stone’s throw away, a neighboring home outside the polygon may not be given any special instructions. The state of weather warnings is binary, akin to a “yes” or “no” to severe weather.

Berry’s team is hoping to improve that by creating a product that reflects the gray area in between. They are experimenting with displaying probabilities to reflect the range of possible outcomes in a rapidly evolving severe weather event.

“There has been a lot of social science research that shows that, given probabilistic information, people make better decisions,” Berry said. “If we appropriately define these probabilities and what they mean, people can use them to make better decisions.”


A look at the difference between conventional weather warnings and the probabilistic grid-based forecasts that will be central to the FACETs program. (NOAA/NSSL)

Imagine you work in a nursing home 20 miles downwind of a tornado-producing thunderstorm. An existing tornado warning extends only 15 miles downstream, so you’re not under a warning — yet.

But you know it takes half an hour or more to move all the residents to shelter. Do you start now? Or do you wait until a warning is (or is not) issued?

Berry’s team found the one-size-fits-all binary nature of warnings doesn’t necessarily fit all consumers. “Some people may need a little more time than what the warning provides,” Berry said. “They may have a lower personal probability threshold.”

A probabilistic approach

Adding probabilities will not replace existing weather warnings but rather offer more context for people around the warning itself. The probabilities will be assigned on a gridded map, much like most weather forecasts, and will update by the minute in real time.

Probabilities will be greatest along the center of the storm’s predicted path, diminishing radially outward as well as farther downwind. Berry’s office compares the so-called plume to the probabilistic wind speed graphics issued by the National Hurricane Center.


A look at the experimental ProbSevere interface. (NOAA/CIMSS)

Severe thunderstorm and tornado warnings themselves are getting an overhaul, too. Warnings will now “move” with storms, growing downwind if a storm looks to hold together. The back edge of the warning will “drop out” behind the storm, too — akin to an “all clear” message once the danger has passed.

“I think the most beneficial thing is the more equitable lead time,” Berry said. “[In the past], people near the downstream edge of the warning [got] much less lead time if they [weren’t] weather aware.”

The warnings themselves will also be updated more frequently. “I think the National Weather Service policy [currently] is that a tornado warning should be updated … once every 15 minutes,” explained Berry. “We’ve tested one-minute updates, two-minute updates. … We started to notice a big difference when we got to the five or two minute [intervals].”

Streamlining the process

All this updating could dramatically heighten a forecaster’s workload, particularly in environments with multiple storms occurring simultaneously. That’s where automation comes in.

“[Meteorologists] are getting some automated guidance that isn’t solely radar,” Berry said. While details are hazy as to what this computer software guidance might look like, it would likely ingest data from surface observations, satellite products, lightning mapping arrays and more.

That means some severe weather warning updates could theoretically be entirely computer-generated. But that doesn’t mean anything is being left on autopilot.

“There’s a lot of forecaster value that I don’t think can be replaced by automation,” Berry said. “One of the features that we included [in an online interface] was to be able to graphically tell which ones were automated versus which ones had been touched by the forecaster."

The shape of the warning could also be changed by automated software packages based on severe weather probabilities churned out by high-resolution computer models. Berry’s team is working on a proposal regarding best practices to prevent fluctuations in the forecast to result in an “expanding and contracting [warning] with time.”

“You don’t want people going in or out of the warning,” Berry said. “We’re working to create more consistency with the warning.”

Berry estimates these changes could take up to five years to implement. By then, atmospheric scientists are hoping to overhaul their strategy for issuing weather warnings — making calls based on forecasts, rather than detection.

Warn on forecast

In the current system, a severe thunderstorm warning isn’t issued until a storm meets severe thunderstorm criteria — the capability of producing damaging wind or hail larger than the size of quarters. The same is true with tornado warnings — rotation must be spotted within a storm.

With more advanced high-resolution computer models, NOAA aims to model individual thunderstorms before they become severe or generate a tornado, issuing warnings based on the forecast of severe weather. Such modeling would test the limits of current forecasting, since they would have to detect weather features at local scales many current models miss.

There may even come a day when you’ll get a severe thunderstorm or tornado warning while standing beneath a blue sky — awaiting a storm that has yet to develop.