How do people predict weather?

679 views

It amazes me that people can predict weather e.g. rain, humidity, wind even to the hour. How accurate is it and how does weather forecast happen?

In: Mathematics

7 Answers

Anonymous 0 Comments

Low pressure causes moist, surface air to rise (causing moisture to condense), and high pressure causes dry, cold air to descend. Literally everything else results from these two processes. The rest is data refinement.

Anonymous 0 Comments

Predicting the weather is actually pretty easy if you know the state of every single molecule in the atmosphere – such as it’s position, velocity, what substance it is, where it has been previously, and so on. Naturally, this is impossible, and even the smallest error will grow exponentially in the long run. We also don’t have computers that could possibly keep track of that many molecules and their states.

We can approximate, though! If the pressure, which we can measure, in a certain area is particularly high while it is particularly low in another, the air will want to flow from the high pressure area to the low pressure area. Perhaps the air carries a lot of water? We can measure that too. If moist air is cooled down, say by mixing with cold air (temperature is also measurable) coming down from the mountains, the water will condense and clouds will form. The bigger the clouds, the higher the probability it will rain – unless it’s really cold because then it’ll snow, and so on.

In the end, predicting the weather just comes down to these guesstimates. Meteorologists use lots of weather stations like [this one](https://ceptu.com/wp-content/uploads/2019/10/weather-station.png) to obtain information about the states mentioned above from lots of different places. They also combine this with what we know in general – that we get more sun in the summer than in the winter, for example. Then they do lots and lots of calculations, and in the end they tell you the most likely outcome on the weather report based on the data available.

As with all kinds of real life physics, your prediction cannot be more accurate than your measures. Since meteorologists can’t possibly know everything there is to know, some errors will make their way through. This is why sometimes the weatherman is wrong, and you get rain when you planned on going to the beach. As more time passes, the predictions get more and more off. This is why you can know today’s weather almost for certain, have a pretty good guess on what the coming week will look like, and be clueless as the what the weather will look like in a month or a year.

Anonymous 0 Comments

They have satellites and hundreds of weather stations with radar towers giving them data. By looking through past data and finding patterns they can predict what will happen next by finding which start of a pattern fits current conditions best. Also using wind direction and conditions upstream you can predict conditions downstream by just saying it will be similar in a few hours.

Anonymous 0 Comments

I cant speak for all meteorologists but one was born with the gift. Their chest was able to tell them of it was going to rain…. well when it was already raining more accurately.

Anonymous 0 Comments

Satellites. Radiosondes. Weather balloons. Ground stations. RADAR. Readings from aircraft (less so at present, for obvious reasons). Any data they can get, basically. Then mathematical models running on supercomputers.

The problem is that the weather is, in the mathematical sense, a chaotic system. What that means is that, even if your model is perfect, you need to know exactly – EXACTLY – what state the entire system is in to corrrectly predict what it will do. Even the tiniest imperfections in your data build over time into big errors in your predictions, and sometimes quite quickly. And those can make for the difference between predicting a heat-wave and a blizzard. And weather data is far, far from complete, so there ARE going to be errors in your predictions. Normally they don’t grow big enough to worry about for a few days, even weeks, but just occasionally it’s a bit like the weather is balanced on a knife edge, and it’s unclear which way it’s going to go, or what it will do when it does. So – how do you deal with that?

The thing about chaotic systems is that, even if they’re not predictable, similar states tend (only “tend”) to follow roughly similar trajectories. So the meteorologists run large numbers of simulations based on the data they have, varying it by small amounts each time, to see whether there are any obvous trends. Quite often the different runs will mostly produce broadly similar results, and that’s the headline data that gets onto the media. Sometimes the weather’s in a more unstable state, and you’ll find several different possibilities tending to pop up quite a lot, some of which run off in totally different directions – and that’s when they’ll tend to hedge their bets more.

Anonymous 0 Comments

It’s also true that some sources barely break the 50% mark, especially in regions near the oceans where weather is generally much less stable and several microclimates conflict. For SF for example:[https://www.forecastadvisor.com/California/SanFrancisco/94134/](https://www.forecastadvisor.com/California/SanFrancisco/94134/)

Anonymous 0 Comments

Take a map and divide it into squares. You take data such as climate data going back as far as you can, and the current conditions for that square. The climate data tells you what the weather was in the past and what were the conditions that led up to that. Then you take all of that data and feed it into a super computer. The computer then predicts based on what happened before and what’s happening now what will probably happen in the future. Then do that for every square and you get a forecast for each of those squares. That’s of course the 20,000 foot level description.