Eli5: How do meteorologists predict the amount of snow/rain that will fall?


Watching the weather report this morning and they said there will likely be 15cm of new snow. How do they work this out?

In: 7

There is a lot of historical data that is used to make a model of weather systems on a computer. The more data, the more accurate the prediction. However there is a practical limit of about 10 days out where the error probability is really high and its accuracy falls sharply.

Computer models track and predict the movement and development of individual cells of atmosphere. Are they going to rise or fall? Warm up or cool down? Gain or lose water? Merge with other air masses? Exceed the “dew point” where moisture condenses or have too little turbulence or convection to keep droplets suspended? All that feeds into models of likely outcomes based on past history.

Because weather is highly chaotic system (in the scientific sense) some of the most powerful supercomputers are used by meteorological organisations to do this.


On Thursdays, the drive home from work takes you about 30 minutes.

But on Thursdays when the local NFL team is playing that night, the drive home takes you 50 minutes. All that extra traffic from people driving to the game, slows you down.

Every time there’s a local NFL game, it takes longer for you to get home.

So pretend it’s Thursday morning, and your wife asks what time you’ll be home after work.

Normally, you’d say you’d be home in 30 minutes. But you check the game schedule, and you see your team is playing tonight.

Knowing this, you tell your wife it will take you 50 minutes to get home.

IN A SIMPLE WAY, it’s like that with meteorology.

“Most of the time, when it’s been 40 degrees with 71% humidity on a Tuesday, it rains on Wednesday”.

They literally study the weather characteristics, and try to see patterns. When the team plays, *usually* traffic is worse. And when there’s certain weather conditions on a Tuesday, *usually* this specific thing happens on Wednesday.

It’s a lot more complicated than that. They take into account temperature, humidity, wind, sun/clouds, and a lot more. They also consider the last several days, maybe even the last two weeks, not just the day before.

But in a nutshell, that’s it. They see repeated patterns over and over, and they can then use those patterns to predict what will most likely happen.

Lots of complex multivariable calculus 👍 but like not doing it by hand, done by computer models and simulations