Why Some Storms Can Be So Hard To Forecast

Several days out from this latest storm, I spoke to a number of meteorologists and was told the weather models are all over the place, so we don’t quite know what’s going to happen. It could hit L.A. hard or peter out.

It was interesting to hear, as I cover climate and weather in Southern California, and forecasts usually seem a bit more confident in the four or five days leading up to a weather event.

Turns out, this storm and the conditions surrounding it were so wildly complicated that weather models were having trouble settling on what was going to happen.

“The dimension, scale and motion of the big feature that’s driving the large-scale weather pattern…that’s what’s different about this,” said Ariel Cohen, the meteorologist in charge at the National Weather Service in Oxnard.

Precipitation on its way to California on February 16, 2024

(

Cooperative Institute for Research in the Atmosphere

/

National Oceanic and Atmospheric Administration

)

Modeling the interaction of various atmospheric features — from the atmospheric river off the west coast to the storm system up near Alaska — can be a struggle. Especially when models are trying to project several days out and anticipate what rainfall is going to be like at a neighborhood scale.

“That storm system that’s up near Alaska — how’s that going to move forward?” said Paul Iñiguez, a meteorologist with the Center for Western Weather and Water Extremes at Scripps Institution of Oceanography. “If it went faster than we expected or it went slower than we expected, that's going to impact how it interacts with that decaying atmospheric river that's laid up on the west coast.”

One set of variables could lead to heavier precipitation in L.A. while another could mean lighter.

People view the Los Angeles River swollen by storm runoff as a powerful long-duration atmospheric river storm, the second in less than a week, continues to impact Southern California on February 5, 2024 in Los Angeles, California.

(

Mario Tama

/

Getty Images North America

)

What are weather models?

Meteorological agencies around the world simulate global weather multiple times a day. The National Oceanic and Atmospheric Administration does it every six hours and projects 15 days into the future.

They can take data from satellites, balloons, airplanes and weather stations, and use equations that represent our best understanding of the physics that govern the atmosphere, the ocean and the cryosphere to project weather on the way. And they switch out variables to calculate different potential outcomes.

“There are fundamental limits to how well you can do it because the atmosphere is chaotic,” said Tapio Schneider, a professor at Caltech. “Chaotic means that what happens later is sensitively dependent on the initial condition, on the state right now. So there's a fundamental limit to how well you can predict the weather, something like two weeks or so, beyond which you cannot say in detail what weather will be like.”

The closer you get to the actual event, the more predictable the outcome.

Though the models have improved substantially over time, even within those two weeks there are big limitations.

The limits to modeling

We don’t completely understand how the weather works or all of the global processes that influence every little thing that happens. Our monitoring equipment around the planet is limited in the amount of data it can give us, as we don’t have probes covering every mile. And there’s a computational limit to just how much data we can process at once.

The models see the planet as a three dimensional grid of data, with the little squares extending from the surface, high up into the atmosphere.

A view of the L.A. River on Monday shows high water levels and a swift moving stream on a river that's often bone dry.

(

Mario Tama

/

Getty Images

)

Just consider for a second the scales of interactions — from clouds swirling over the Gulf of Alaska down to rain falling on Topanga canyon.

“Think about everything that’s smaller than a few kilometers. That’s clouds, cloud droplets. None of these things you can simulate accurately with the current computational power or any computational power we can foresee getting in a decades to come,” said Schneider.

The more localized the projection, the more accurate the forecast, but the higher the computational cost.

Because of that cost, there are limits to how granular the models can get. Which is why there are inaccuracies when forecasting down to a local level, especially in a place like Southern California, which has wildly varied terrain that results in 20 degree temperature differences just a few miles apart.

Mudflows inundate a street in Beverly Hills.

(

Mario Tama

/

Getty Images

)

Should you take forecasts seriously?

Yes. With the understanding that even if you get a warning from the National Weather Service about a hazard like a flash flood, you may go through a whole lot of prep that doesn’t actually need to be used.

At the end of the day though, on the off chance that an extreme scenario does occur, it’s worth getting your stuff together.

I’d rather be disappointed than unprepared and caught off guard.