We brace for the worst when hurricanes make landfall.

Powerful winds trailing inland and torrential rain from storms that don't even have an eye.

The threat isn't just at the coast anymore, it's everywhere.

And the problem is these storms are getting stronger and the strong ones are what forecasters find the hardest to predict.

- For those kind of events, you need to have accurate forecasts.

- Tropical storm Chantal dropped nearly 10 inches of rain on parts of Central North Carolina.

Here at the Eno River, the flood from Hurricane Fran in 1996 came just shy of this sign.

But Chantal, the water rose even higher.

The flood officials now call a 500 to a 1,000 year flood event.

That doesn't mean once in a thousand years, it means rare.

But it's getting more common now and it could happen again and sooner than we think.

At NC State, scientists look beyond radar.

They use powerful computers to build 3D models of storms.

Models shaped not just by satellites and service data, but by aircraft that fly straight into the heart of hurricanes.

Those onboard sensors and those they drop deep inside send back real-time measurements, feeding those models and sharpening the forecast.

Dr. Lacherman, what are the models still getting wrong?

- The most extreme weather events are the ones where they struggle the most.

And those are also the highest stakes forecasts.

So when you have extremely heavy rain, hurricane, strong wind, atmospheric river, models struggle with the extremes, but that's when you need to get the forecast right the most.

- Radar observation-- - Getting the forecast right has taken decades of scientific research, starting in the 1950s when computers were first used to simulate what a storm was going to do.

- UNIVAC can take the past histories of thousands and thousands of storms, analyze them, compare them with developing conditions, and make predictions.

- Since then, forecasting has advanced dramatically.

Today's models capture wind, rain, and storm surge with a level of detail and precision that early scientists never thought possible.

We looked at an image from Hurricane Isabel that hit North Carolina in 2003.

- If you look within the eye, you can get what are called eye wall vortices, where you get little circulations that are embedded within the eye, and those can concentrate damage when a hurricane makes landfall.

- Researchers now run massive models of Trekker Storm's path.

While a second high-resolution model zooms in, - What we had, we have kind of a large, large-scale model that shows kind of the whole Western North Atlantic, and these lines tell us where Hurricane Isabel is, and then the colors are the rain that it left behind in its wake.

But then to get more detail, we have another part of the model that follows the storm.

- But even with better modeling, like what happens when a hurricane makes landfall, remains one of forecasting's greatest gaps.

And Cameron Macielo is trying to close it.

- So this is actually one of the model simulations that I use a lot for my research.

And so what this is, is a numerical representation of a hurricane.

So what you see here is this kind of black line.

This represents the coastline in my numerical simulation.

So everything to the north of it is over land, everything to the south of it is over water.

The blue colors to green are tropical storm force intensity.

Where you're seeing these yellow and reds, these are like hurricane force intensity winds.

So these would be like Cat 3, Cat 4 winds here.

And so this is kind of what we think of as a one minute sustained wind.

Doesn't tell us kind of the turbulence or the gusts what are happening.

- It's those fleeting violent blasts that can destroy life and property and remain one of the hardest things for models to predict.

- What that means is we get outputs that are extremely math intensive.

- Since wind is a fluid, Macielo says they use what's called the Navier-Stokes equation, which is a math formula that helps determine how air moves.

And without it, it would be difficult to accurately predict the weather.

- In a hurricane, there's kind of a primary and secondary circulation.

So the primary circulation, you know, is those swirling winds, which we call tangential or azimuthal winds.

And then there's the secondary circulation of winds kind of going in towards the hurricane center, up and then out.

And so you're kind of seeing right here at the top of the atmosphere, you're seeing the outflow where you see the kind of the striations moving radially away from the TC center or the hurricane center right there.

That is kind of outflow, basically vecting mass away from hurricane center.

- No single model can predict a storm with perfect accuracy.

Each run will get something wrong.

So they run multiple models.

Only then can they see patterns emerge.

Those patterns help forecast more than just hurricane winds.

They reveal so-called atmospheric rivers, which are narrow bands of moisture that can carry more water than the Mississippi.

- We are worried about them intensifying to a point where they go from being beneficial to more destructive.

- While they provide valued rainfall in places like the West Coast, they can have a different impact elsewhere in the East.

- So here we have imagery of the integrated water vapor for the whole world.

Integrated water vapor is just the moisture from the surface to the top of the atmosphere.

And where we start to see these streamers really condense, that's where we would be looking for those are atmospheric rivers.

- Even in the Southeast, those streamers can stretch from the Gulf of Mexico, feeding moisture into powerful storms.

They're part of the Earth's water cycle, but when they grow stronger, the balance shifts.

East Tom says some models can get confused when weather hits barriers like mountains or ridges, like you'd see in the North Carolina Blue Ridge.

And that can affect speed to the system in some areas.

- So what I'm doing is I'm taking what we call a re-forecast.

That's when we just rerun the forecast model with historical data and try to diagnose where it went wrong with these atmospheric river forecasts.

And then I was able to diagnose where we were seeing the primary errors, which was either the forecasts were too fast or too slow, or they were too intense, or they were too weak.

So for, actually for a lot of the too fast cases, we saw this anomalous trough.

And then conversely, with the too slow cases, we saw this really predominant ridge here, and that really inhibited the landfall of those atmospheric rivers.

- These findings help explain why long-range forecasts can often miss the mark, but offer clues for improving future predictions.

At NOAA, researchers are flying into storms and conducting atmospheric recon missions, and have increased precipitation accuracy by up to 12%.

And these missions that work at NC State, much of it depends on funding.

- This research, everything that goes into these observations and these weather models is funded by taxpayer dollars and science funding to help make all these capabilities possible.

If that goes away, it's going to be a tremendous loss of investment.

- And possibly valuable warnings on extreme weather.

And so, in a future of warming seas and stronger storms, the American forecast may depend not just on the science, but on whether we choose to invest in it.

(upbeat music)