Kylie Bearse remembers the blizzard of March 2021—and the public ire that ensued when the record-breaking snowstorm arrived half a day later than local meteorologists initially predicted.

“The storm slowed down as it was arriving,” recalls Bearse, a meteorologist for Fox31 Denver who, at the time, worked for 9News. “And people were really angry the morning—I think it was Saturday morning—that it was originally supposed to start.”

When Bearse posted to TikTok explaining that the storm would actually start later than expected, viewers responded with vitriol:

“How many times are you going to be wrong? You can’t even get food at the store because all you idiots lied and scared everyone into hiding”

“Y’all hype up winter storms every year and it never happens.”

“I don’t believe a word you say, crazy witch lady.”

This type of commentary is nothing new for Bearse, who’s been a meteorologist for 11 years. And it’s not unique to Colorado: People love to hate on meterologists in general, as articles in the Washington Post, Forbes, and Philly Mag have explored.

“There is a big portion of Colorado that is very understanding of how difficult it is to forecast here and that things change leading up to the day of a forecast,” Bearse says. “And there’s maybe even a bigger chunk that are quite loud and quite frustrated [when an initial prediction is off].” Bearse is sympathetic to the latter group’s outrage: People alter their plans based on weather forecasts, she says, “so I’m very aware and understanding of the frustrations.”

That said, there’s a lot that goes into predicting a snowstorm, says Andy Stein, a Colorado-based meteorologist and 5280 contributor. The typical process, he explains, is to start looking at computer-generated weather models roughly 10 days in advance and notice what patterns are occurring in the upper levels of the atmosphere. “What happens up there is kind of a direct correlation to what happens here on the surface,” he says.

From there, it’s about monitoring key data points, including a storm’s location, speed, and strength, as well as the influence of other high-pressure systems and weather patterns—for instance, the La Niña weather pattern, Stein says.

Around five days out, a meteorologist might evaluate the geography and topography that a storm might encounter and analyze how those elements could influence the forecast. “In Colorado, that’s deserts and the plains and the Continental Divide and buttes and canyons,” Stein explains. “There’s a lot of different local terrain that can alter what happens with a storm.”

As meteorologists monitor a storm’s progression and the computer-generated models update data, storm predictions are refreshed to reflect the latest information. “[If a forecast] says one thing three days out versus the day before, it’s not because we were like Oops, that was fun, and we just forecasted the numbers willy nilly,” Bearse says. “We have been watching and updating [our predictions] as we’ve gotten a little bit closer and are seeing what the storm system does.”

Several factors can change a prediction up to the last minute, including the strength, speed, and location of a snowstorm, Stein says. Additionally, Bearse adds, small shifts in wind direction can make a big difference in snow totals and whether or not snow even falls in the first place. That’s due to the forces of upslope and downslope. Upslope is when air moves up the side of a mountain, which cools the air and squeezes out its moisture, creating a “snow machine effect,” Bearse explains. Downslope is the reverse: Air travels down the side of the mountain, warming and drying it along the way, “which, as you can imagine, does not create the best snow-making conditions,” Bearse says.

On the Front Range, northeasterly winds create upslope, whereas west and northwesterly winds lead to downslope. Bearse recalls a storm about four years ago where she had predicted two to three inches of snow, but the wind shifted several degrees, resulting in a more easterly direction, which created significantly more upslope and dumped about a foot of snow on Longmont.

“No one gets more frustrated by a forecast that doesn’t pan out than a meteorologist,” says Bearse, who gets anxious with every impending snowstorm. “They’re my favorite to forecast,” she explains, “but there’s a lot at stake,” adding that many people hinge their plans on the weather forecast.

Further complicating the process is the fact there are more than 200 microclimates in Colorado, which means forecasts can vary widely throughout the state, Stein says. A state like Florida, by contrast, might have around 20 microclimates. Moreover, compared to the east coast, where there are “insane amounts” of weather radar coverage, Stein says, Colorado has gaps in its radar system due to fewer radar stations and tall peaks, which can block data towers. In short, it’s hard to verify the predictions with what’s actually happening on the ground.

When initial storm predictions are off—as was the case in March 2021—Bearse braces herself for the inevitable wave of criticism. “In my opinion, Twitter is the meanest social media platform,” she says, though she also sees negative comments on TikTok. The most common critique she hears is some variation of: I wish I could get paid to be wrong.

Luckily, at this point in her career, Bearse is able to laugh off most commentary—and even find a silver lining to it.

“[Meterologists] bring people together no matter what their political party,” Bearse says. “We [provide] an opportunity for everyone to come together and hate the weather person.”

Still, the next time a storm is slightly off its initial prediction, take a moment to consider all that goes into forecasting Colorado’s fickle weather. And instead of rage tweeting at meteorologists, maybe just try to appreciate the conditions outside—whatever they may be.

Interested in what’s ahead for Colorado’s next snowstorm? See updates from Bearse and Stein on TikTok.