AyashiKaibutsu
Diamond Member
Damn it my college better not bend over and take it from another hurricane : (
-
We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.
Hurricane Jeanne IS heading for Florida (again)
Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Are you asking this as a serious question?
They are never right. You know how many weekends they've ruined by saying its going to be nice and it rains and vice versa? How about all the snow storms they predicted that never materialized or snow storms that dumped a ton of snow that they said would have little or no accumulation. These guys can't even predict one day out it seems.
You guys do realize that the weathermen don't actually predict anything, right? They give you what the computer models spit out.
Depending on the conditions, the models can be right.. or wrong. There are so many variables that it is nearly impossible, unless the weather is in a steady pattern..
The models can generally predict 24 hours into the future realitivly accurately.. I don't have any figures though, that would be an interesting study.
But you'll notice that unless, as I said.. the weather pattern is steady, the later days in the long range forecasts will almost always change each day, sometimes each forecast... the weather is just too turbulent, so to speak.. but the more we predict and are wrong, the more we can correct our mistakes and theoretically the more accurate we will get..
The ultimate will be some sort of weather supercomputer that basically calculates weather by using data from the whole globe. That will probably be a huge advancement.
DrPizza
Administrator Elite Member Goat Whisperer
Not a huge advancement because the other local variables would have more of an effect on the weather 2 days down the road than the weather half way around the world. The problem with weather forecasting is chaos. We cannot make enough precise enough measurements to predict more than a couple of days in advance with any reasonable amount of accuracy. Nonetheless, there's public demand for forecasts 7 or even 10 days in advance. It's pointless to provide those predictions, because random guessing is probably just as accurate... but you've gotta keep the masses happy. Plus, with their short attention spans, they usually forget 5 days later what you predicted 10 days in advance.
apoppin
Lifer
Some places . . . like Hawaii, the weather is relatively easy to predict ("morning and afternoon showers especially for the mountains and windward side" in a 'Trade pattern'; 'Kona Lows' are much harder to predict); the SoCal Desert is another "easy" place - where there are LESS variables. 😉
By the time we have a global model, we will be able to CHANGE weather (not that humans aren't affecting weather now - IMO, hurricanes are a "balancer" - that is why we are seeing more and more powerful ones . . . . of course there are always head-in-the-sand people who don't care about 'global warming'. 😛
:roll:
it's only gonna get worse.
Yeah, that's true I suppose. What I was kinda getting at, though.. is some sort of technology that could watch every square mile of the globe. If you could compute with that ammount of data, accuracy would probably go up a bit.. assuming we know how the things at play actually.. play, and implement that into the models correctly..
When the weather pattern changes suddenly, something has to cause it... I'm probably trying to make it seem easier than it is, though.. it would probably take some pretty massive computing power, assuming it's even possible... I suppose it should be possible, unless it is really "chaotic" and unpredictable, but that would kinda go against modern science or something.
dullard
Elite Member
As far as I know, I'm the only one at Anandtech who actually uses programs similar to weather prediction programs. These programs are not very accurate.
1) We simply don't have the basic theory. Without the theory, how can we write a computer code, and how can anyone expect that code to be highly accurate? A lot of the theory is quite good, but when it comes to turbulence in the atmosphere there is NO good theory.
2) We simply don't have the computer power. A simple sounding calculation like airflow around a building may take days or weeks for typical supercomputers to solve. When it comes to weather we don't have weeks (how useful would a great computer prediction for last weeks weather be?) Thus they have to dramatically oversimplify the calculations - leading to errors even if the theory was perfect.
3) We simply don't have the raw data even if #1 and #2 were perfect. We have official data at every airport - and basically nothing else. Data for wind speed, direction, humidity, temperature isn't known in very many locations. Even worse: most of the data collected is just along the ground very few weather ballons are used to get data in middle or upper atmosphere.
4) Even if we could get the billions+ of data points needed for accurate predictions, we don't have the infastructure to get all those individual data points to every single weatherman's computers.
1) We simply don't have the basic theory. Without the theory, how can we write a computer code, and how can anyone expect that code to be highly accurate? A lot of the theory is quite good, but when it comes to turbulence in the atmosphere there is NO good theory.
2) We simply don't have the computer power. A simple sounding calculation like airflow around a building may take days or weeks for typical supercomputers to solve. When it comes to weather we don't have weeks (how useful would a great computer prediction for last weeks weather be?) Thus they have to dramatically oversimplify the calculations - leading to errors even if the theory was perfect.
3) We simply don't have the raw data even if #1 and #2 were perfect. We have official data at every airport - and basically nothing else. Data for wind speed, direction, humidity, temperature isn't known in very many locations. Even worse: most of the data collected is just along the ground very few weather ballons are used to get data in middle or upper atmosphere.
4) Even if we could get the billions+ of data points needed for accurate predictions, we don't have the infastructure to get all those individual data points to every single weatherman's computers.
SP33Demon
Lifer
I was sarcastic because usually forecasters are dead on regarding the weather within at least a day. To expect them to predict something like a hurricane, that even our supercomputers don't have all the variables for, and its 3-5 day forecasted path is ridiculous. Even within a day we cannot say 100% what a hurricane will do or go, as demonstrated by Ivan. Usually they are right 80%-90% of the time with the weather forecast for the next day, so cut them a break.
apoppin
Lifer
Actually the programs are quite accurate . . . for the data inputted. Looking back (in hindsight) we can see where the models went "wrong" and often it is the intrepretation of them. 😉
Hurricane predicting IS getting better . . . there just isn't enough resources devoted to them (or the weather in general)
Anyway, here's the latest (11am):
Notice the uncertainties BUT ALSO notice where the "Models agree". 😉
SP33Demon
Lifer
Yes, chaos theory is very interesting and I'm sure it has some validity to it. The theory is that a butterfly somewhere flaps its wings, and this acts as the catalyst for a chain of events that eventually leads to a tornado. While I personally am skeptical about this, I think other things such as our NOX gases play a humongous role. The fact that the ocean is rising and cities are usually 5-10 degrees hotter ("heat islands") accentuates this fact.
Again, I completely disagree. If the weather is clear, they are usually right. If the weather is uncertain, I'd say they are wrong almost all the time even 1 day out. I am not kidding when I say they were wrong almost every single weekend this year with their predictions even one day out.
dullard
Elite Member
Accuracy is relative. Take the snow question above as an example. It comes down to resolution. If you have great data from a few points which may be hundreds of miles apart, it becomes very difficult to get the weather accurate down to the mile resolution that affects people. They often predict a storm very accurately in many aspects (size, intensity, amount of precipitation) but they may be off by 10 miles to the south. In one sense, that is a huge accomplishment - they succeeded with so many things. But in another sense, they predicted 4-12 inches of snow (which probably occured ~10 miles away) for the user who saw 36 inches. On a personal scale, accuracy is miserable. In the big picture, accuracy is quite good.
To this day, there are NO theoretical models that can tell me what will happen to air downstream from any complicated shaped objects. None. Nada. Zilch. Heck, even most computer models of air downstream from a cylinder (the simplest shape in this case) use very crude empirical estimations. If they can't have any theory what happens downwind from any object, how can they model around buildings, hills, mountains, etc?
In that sence, hurricanes are often easier to predict (my comments were meant for weather forcasting in general). Being 10 miles off on a hurricane doesn't matter - as you still will have heavy winds, heavy rains, etc. Hurricanes don't have mountains to cross as they travel along the ocean.
SP33Demon
Lifer
Do you pay attention when they say 10, 20, or 30% chance of rain even though they forecast it to be cloudy? Even if it rained, I would still consider that to be "right" because they said there was a chance of it, so you should plan accordingly. If you are saying that they forecasted, on Friday, that the next day would be sunny and then it rained on Saturday, then that would be "wrong". In my old age I have rarely seen forecasters get the weather wrong for the next day.
Come on.
Of course I look at it. When it says 70-80% chance of rain and it doesn't rain or it says 10-20% chance of rain and it rains are they right? Maybe technically but its still wrong. Its like saying the Red Sox have a good chance of winning tomorrow but they could lose. Well, duh.
dullard
Elite Member
24 hr is a lot better than 3 day, 5 day, etc.
I don't have national statistics. The best I can do is this: My local weatherman that I watch most often had a long running promotion. If the temperature is off by more than 4°F, he gives away $42 to a random viewer. Now these aren't full 24 hour predictions in most cases. Since he did the weather at 9 pm and the high is typically mid afternoon the next day, he really was only doing 15-20 hour prediction comparisons. He averaged 5 hits and 2 misses each week in my memory of times I watched. That may sound good, but think about it, he had a 9 degree range where he didn't have to pay out. (1°F for a correct temp, +4°F, and -4°F). If I gave you a season, a location, today's temperature, and the temperatures of locations upwind of you, I bet you'd have a damn good chance getting the temperature within a 9°F range. Heck, in summer here, if I just say the high will be (81°F to 89°F) I bet I too would be right nearly 5 out of 7 times without any models and without any data.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 25K
-
-
-
