We read and hear alot of scary things about AI.
I've been using it almost every day, all day for more than 4 decades since I become a meteorologist just like all professional meteorologists and in fact, many other people who use weather products.
WxFollower, who is a self taught weather expert has likely used these AI products many thousands of times.
Super computers using millions of individual data values fed into hundreds of thousands of mathematical equations that define the physical laws of the atmosphere that are processed around the clock with results that provide humans with a million+ specific solutions around the entire planet.
During a 2 hour global weather model run, the super computer will do 10,000,000,000,000,000 (10 quadrillion) individual calculations.
The smartest mathematician couldn't do that many calculations working 24 hours a day on it their entire lives!
https://www.noaa.gov/jetstream/upper-air-charts/weather-models
The motion of air in the atmosphere is extremely complicated. From large-scale synoptic motion we see high and low pressure systems at the surface and ridges and troughs in the air aloft. Embedded in synoptic motion is mesoscale motion from which we see sea breezes, drylines, and squall lines, etc.
Even smaller than mesoscale is microscale, which is typically too small to be depicted on weather maps. This scale includes things like fair weather cumulus clouds (small clouds) that are here one minute and gone the next.
All weather is caused by atmospheric motions that can be described by mathematical equations. From these equations, we can calculate future motions. Numerical weather prediction is the use of computers to model the atmosphere and predict how atmospheric motions change both horizontally and vertically with time.
What is AI?
AI, or artificial intelligence, refers to the development of intelligent systems that can perform tasks requiring human-like intelligence. It encompasses various approaches, of which the most relevant for weather forecasting is machine learning.
Machine learning is a specific subset of AI that involves training algorithms to learn from data and perform specific tasks without explicit programming. It focuses on improving the accuracy and efficiency of these tasks by leveraging patterns and correlations in the data. In short, AI encompasses the broader field of creating intelligent systems, while machine learning is a specific technique within AI that enables systems to learn from data and perform specific functions without explicit programming.
How is AI used in weather forecasting?
Both weather broadcasters and viewers have greatly benefited from the integration of AI technologies, revolutionizing how predictions are made and communicated. Here are some of the most common applications of AI within the field
Using AI for weather prediction is not a new innovation and has been in use since the 1970s. The weather models that broadcasters rely on to make accurate forecasts consist of complex algorithms run on supercomputers. Machine-learning techniques enhance these models by making them more applicable and precise.
The AI tells us what the temperature will be every hour, for the next 384 hours at any place on the planet, for example. There are several different global models that use slightly different equations that do the exact same thing.
And each model has a set of up to 50 different ensembles that run simultaneously that do the exact thing but have a tiny variation in 1 parameter/equation because the perfect set of equations to accurately represent the atmosphere still doesn't exist. And because of chaos. If 1 element is off a tiny bit, errors growing huge with time.
So what we do is average the 50 individual ensembles to get an ensemble mean that averages out the extremes and provides the most reliable solution, which is usually in the middle somewhere.
But the product of this analysis comes out a step farther with numerical forecasts that can be accessed by no nothing about meteorology humans to find out what the weather will be.
Here's a perfect example of the end product that a person who knows nothing about meteorology can use today that just came out.
This is from the NWS hourly forecast for Evansville IN for 29 different elements(you can pick them) going out the next 6 days.
A person that knows nothing about meteorology can go to the last link every day and get the same numerical values in the forecast as the NWS meteorologist uses.
Thanks to AI!
https://forecast.weather.gov/MapClick.php?x=285&y=83&site=pah&zmx=&zmy=&map_x=285&map_y=83
https://forecast.weather.gov/MapClick.php?lat=38.05&lon=-87.53&unit=0&lg=english&FcstType=graphical
| Point Forecast: Evansville Regional Airport IN 38.06N 87.52W | Last Update: 6:02 am CST Nov 17, 2023 |
 | |
[dashes/dots] | [b/w] | [hide menu] | ||||
| ||||
Sky Cover (%) Precipitation Potential (%) Relative Humidity (%) | Rain Thunder Snow Freezing Rain Sleet Fog | Quantitative Precipitation |
0.100.250.501.00
Snowfall
| |
|
| ||
|
