US weather forecasters triple supercomputing oomph with latest machines Predicting the weather is a notoriously tricky enterprise, but that’s never held back America's National Oceanic and Atmospheric Administration (NOAA). After more than two years of development, the agency brought a pair of supercomputers online this week that it says are three times as powerful as the machines they replace, …
I don't think weather forecasting has gotten much better in the past forty years.
Perhaps with the prediction and accuracy of tornadoes? But run-of-the-mill temperature, precipitation chance, humidity, wind - not much seems to have changed and that's not saying a whole lot.
Still loved it when I was in Los Angeles several years back and the local news channels were all advertising that they had the most powerful Doppler radar in the area. Can't find the commercial, but here's LA ABC7's Mega Doppler 7000 HD page... eyes rolling. (It's stated in the title of the page.)
Oh, and by the way - Los Angeles receives on average 14 inches of rain a year, with 92% of that coming during the rainy season (November to April). Big whoop. [0].
[0] - https://en.wikipedia.org/wiki/Climate_of_Los_Angeles
It feels like it is more accurate than it was a couple decades ago. Sure, there are always times it is off, but usually it is stuff like being off on where the weather front will develop - they say it will rain where I live but the front develops 100 miles west and peters out by the time it gets here. Or they get the timing wrong and a storm they predict a few days out comes a half day earlier or later than projected.
Last Saturday my girlfriend had a backyard party. She had scheduled it about a month ago and when it was two weeks out and that day showed up in the extended forecast on my weather app (Accuweather) it showed 80% chance of rain for that day. I told her about it and she said "they're never right, it'll be fine". It rained 2" starting at 3am Saturday morning until 11am, her party started at noon. That's not bad for a prediction two weeks out.
Obviously there are plenty of examples where they screw up, but I remember when it seemed like they were almost never right, and more and more I'm surprised about how accurate it turns out well over a week in advance. Even if the success rate was only 40%, if it was previously 10% that's a big improvement.
It may have something to do with where you live, as I'm sure some locales must have more difficult to predict weather than others.
Where I live just under a small mountain and not that far from some effing big ones, the weather can sometimes change dramatically in a very short time. Even then, local predictions are amazingly good, they map out big storms usually 2-3 days ahead, and are maybe 2-3 hours off the start/end times of the storms.
Of course as you say, they don't always get it right, but I would say more often than not they do by quite some margin (maybe 65-35). Considering the (apparent) volatility, that's a pretty high strike rate.
I think what really has greatly improved in the last few years is the resolution of the models. I remember seeing forecast maps covering whole countries with maybe 5 or 6 regions, going there's sunny, there's rainy etc with each region covering hundreds of kms. My weather app today can give me a pretty accurate individual forecast for every town in the region.
Isn't that comment in the report equivalent to "hugely small" or a "massive quantum leap"?
The best I can do from the dictionary definitions is "beyond local" which is ... err ... what's more local than local?
At present I believe the Met Office work to about 10km forecast cells (did that drop to 1Km when they upgraded?) ... yet it still can't accurately forecast whether it's going to rain (heavily) in the whole of East Anglia (roughly 10Km square) ...
Perhaps general accuracy is more important than the localisation of forecast as, like another commenter has said, the forecast itself does not seem much more accurate than it was 20 years ago. "50% chance of rain tomorrow" still means "in similar conditions to those predicted for tomorrow, historically there was a 50% chance of rain." The model predicting temperature, pressure, wind speed etc is slightly more accurate but they still rely on old statistical data to forecast whether it will rain ...
I wish they would expand on precipitation forecasts. Give me a chance there will be any rain during the time period, the percentage of that time period it will be raining, and the amount of rain expected during that time period. For larger regions you might want something like the percentage of the area that will see any rain during the time period.
There's a big difference between a 50% chance of any rain during the day versus rain being expected for 50% of the daylight hours of that day. There's also a big difference if that predicted rain event will total 0.05 inches versus totaling 3.50 inches.
It may be that is not what you want it to mean, but that is how various weather forecasters predict rain. It's a difficult to understand definition, but they've struggled over the tears to come up with something better.
I used to help look after the IBM supes at the Met. Office some years back before Cray got the gig, and I believed that the UK forecast was down to 4KM cells when I left, and they had the capacity to do 1.5KM for special forecasts (I remember when the Power7 775's were being commissioned just before the 2012 Olympics, they produced a forecast with an exceptionally high resolution - I think it was 800M, but it could have been higher, for the location of the sailing events, just because they could with the extra power). The organisers were quite happy, as they accurately forecast difficult conditions for the first scheduled day, so they delayed the event until the next day, and that produced much better racing.
I think that generally, the forecasts they produce are reasonably accurate at larger scale, but the local forecasts appear less than accurate. They also change a lot as you get closer, probably because the predicted weather mostly happens, but not necessarily where they said it would, but it still happened somewhere close by. But I feel they're less accurate now than they were 10 years ago. And this is not just the Met. Office forecasts, but the MeteoGroup ones used by the BBC as well.
But it may just be the way it is presented. The TV forecast has to be much more general than the local forecasts served up by the weather apps, and sometimes it gives a better overall picture.
I recall the "Weather Rock".
If the rock is wet, it's raining.
If the rock is swinging, the wind is blowing.
If the rock casts a shadow, the sun is shining.
If the rock does not cast a shadow and is not wet, the sky is cloudy.
If the rock is difficult to see, it is foggy.
If the rock is white, it is snowing.
If the rock is coated with ice, there is a frost.
If the ice is thick, it's a heavy frost.
If the rock is bouncing, there is an earthquake.
If the rock is under water, there is a flood.
If the rock is warm, it is sunny.
If the rock is missing, there was a tornado.
If the rock is wet and swinging violently, there is a hurricane.
If the rock can be felt but not seen, it is night time.
If the rock has white splats on it, watch out for birds.
If there are two rocks, stop drinking, you are drunk.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
