Hurricane Florence and Hurricane Fake News have been jostling for dominance these last few days. And that's dangerous. Because in times of crisis, trust and truth can become one and the same thing.

President Donald Trump’s refusal to accept that nearly 3,000 Puerto Ricans died last year in the aftermath of Hurricane Maria — a subject he raised as Florence neared landfall — is part of the clash.

“FEMA doesn’t count deaths,” emergency management chief Brock Long said in defense of his boss. “And if you take what’s going on with Florence, the deaths that are verified by the local county coroners are the ones that we take.”

Of course, FEMA does count deaths — and everything else, too, from bottles of water to blankets to generators. That’s what bureaucracies do. What Long missed was an opportunity to explain how deaths are assessed, as with Maria, when hurricanes blow coroners away, too.

Then there was Donald Trump Jr.'s apparent attempt on Twitter to use an old photo to accuse CNN's Anderson Cooper of staging a flood photo. The image was from Hurricane Ike in 2008, and was part of a legitimate focus on the hidden dangers of receding waters.

It wasn’t a first. Bogus photos peddled on social media during disasters — like that shark swimming along a highway — have become such a widespread problem that many journalistic outlets set up debunking squads during crises.

But if you want a sense of the frustration that comes with meteorological messaging in an era of fake news, plug into the Twitter feed of Marshall Shepherd, director of the University of Georgia's atmospheric sciences program.

In the weather world, he has become a central figure in the pushback against deniers who say that predictions of Florence’s landfall were off-target (they weren’t) or that meteorologists didn’t warn of the flooding that would become the storm’s primary danger (they did).

“Not paying attention not = to no warning,” Shepherd wrote in perfect Twitterese, as Florence churned on Sunday.

We spoke on Monday. “People will so often assume that if they didn’t see it in their Twitter feed or their Facebook feed, it didn’t happen,” Shepherd said. “That’s something we’re really struggling with, with weather messaging.”

Shepherd divides deniers into two camps. First, there’s the “everything’s local” crowd. If they don’t see more than six inches of rain from a hurricane, they find it hard to believe that someone 10 miles away experienced 30 inches. But rain bands behave like that.

Then there are those distrustful of expertise in all its forms. Look deep enough, and you’ll find that most are climate change skeptics. “I don’t think those two can be decoupled,” the UGA professor said.

This is important. One of the failings that Florence has pointed out in the meteorological media world, Shepherd said, is an overemphasis on wind as a weather gauge for public worry. We sweat over the high winds of a Category 4 hurricane, then relax when it’s downgraded to a Category 1.

But the real damage is done by the water dumped by these storms, not the wind. Harvey, Sandy and Florence — the last three storms to cause tremendous damage on the mainland — were “wet” storms, not “wind” storms.

To explain that rain is the new — or at least, newly prominent — villain in hurricanes, one also needs to convey the role of global warming. Heated ocean water makes more water vapor available to storms.

“The concept of saying ‘downgraded’ or ’weakened should be forever banished,” Shepherd would tell the next journalist he spoke with. “With Florence, I felt it was more dangerous after it was lowered to Category 2.”

But before we hung up on each other, Shepherd said that if I really wanted to understand where natural (or man-made) disasters and fake news intersect, I needed to talk to one of his colleagues.

Sarah DeYoung is an assistant professor at UGA’s Institute for Disaster Management. One of her specialties is how and when people listen to evacuation warnings.

That false alarm last January — when residents of Hawaii were told that a nuclear missile was headed their way? She’s got a paper coming out on it. Rather than seek immediate shelter, she found, most residents engaged in “social milling” – checking the warning’s credibility with other sources.

If you’re in the emergency management business, and seconds count, that isn’t necessarily a good thing. Trust matters. So does visual — or olfactory — evidence.

Take wildfire-plagued California. “If they can smell the smoke, we have an easier time getting people to evacuate,” DeYoung said. “It becomes a problem if the wind is blowing in the wrong direction. The fallback for that is making sure the local citizens trust their local government and first responders who are sending out messages.”

But the most interesting thing to come from DeYoung was a 2016 study of North Carolina residents — the same people now swimming in Florence’s aftermath — and what it would take for them to leave their homes in the face of a storm.

Women observed evacuation orders quicker than men. Renters quicker than homeowners. Not a surprise.

But DeYoung and her associates — she was the lead author — also found evidence of a shifting trend among African-Americans in North Carolina. In past studies, minorities required more persuasion, not less, than white residents when pushed to seek alternative shelter.

In this 2016 study, African-Americans required less persuasion than white residents.

One reason may have been Hurricane Katrina in 2005, and the abandonment felt by many African-American residents of New Orleans. “Minority residents may evacuate at lower thresholds of storm categories, if they feel that there will be insufficient emergency services and support if the hazard is severe,” the DeYoung paper stated.

Another possibility is the tension between black communities and law enforcement. “Send the red, not the blue,” has become an emergency management mantra, DeYoung said in our interview. In other words, rely on fire departments, not the police, when dealing with certain populations.

“There’s an issue with law enforcement shooting people of color on a regular basis,” she said. “Now go into a public evacuation shelter scenario. The sense of public safety in a public shelter is then different between black and white.”

In emergencies, we draw from life’s day-to-day lessons. For better and worse, sometimes the lack of trust wins.

Now consider the fact that on Oct. 3, the Trump administration will send out a “presidential alert” to all cellphones in the U.S., testing a new alert system carrying warnings of national emergencies.

The test had been scheduled for Thursday, but was postponed out of respect for Hurricane Florence. The message is to begin with “This is a test,” and end with the phrase, “No action is needed.” We do not know if President Trump’s name will appear in between.

If it does, the question must be asked: Will you give more or less weight to an emergency message from a world leader who carries on a highly casual affair with actual facts?

“I think it needs to be made clear to the public that it’s not actually coming from him,” DeYoung said. “Trust is very important when it comes to heeding warnings and orders.”

If it’s not there, and clouds are rolling in, then we have a problem.