Why Aren't We Better At Predicting Weather?

[Spooner]

I can understand not being able to know what's truly going to happen next week (with varying pressure system, winds, and such)... but why is my goddamn weatherman completely WRONG like 75% of the time? It's 2002. I should be able to wake up in the morning, turn on the TV, and know if I need to take an umbrella to work... or know if I'm going to need a shovel to dig my car out after work.

Probably as something to do with living in new england, eh?
Damn weather changes more often than watch's second hand.

 

[Elledan]

I got two words for ya: Chaos Theory.

After more than three days it becomes virtually impossible to 'predict' the weather.

 

[HappyFace]

Sometimes I feel that these guys are making nothing more then an educated guess.

 

[Random Variable]

<< Sometimes I feel that these guys are making nothing more then an educated guess. >>


That's why it's called "weather forecasting." No human or computer can definitely predict the future.

 

[dullard]

One of the best weather predictions ever used: Todays weather will be the exact same as yesterdays. With all our advanced computers and models we really can't beat that prediction by more than 1°F or so.

Elledan has the correct answer. If you say today's temperature is 80.0°F a model may say tomorrow's temperature will be 92.4°F. If instead you plug in that today's temperature is 80.1°F, the model might say that tomorrow the temperature will be 102.3°F (assuming you kept all other input data the same). See how chaotic the output is to minor input changes? That is why we can't do better. At least yet...

We cannot accurately measure the current conditions. We only measure official temperatures/wind speeds/etc at major airports. You know what? That isn't enough data. Plus the measurements are altered by the "city effect". Large cities are generally up to 2°F hotter than the surroundings. When we take so many measurements from large cities and 0 measurements from difficult to access areas, we get incorrect input data. Sure downtown Chicago may be 90°F, but the rest of Illinois is 88°F. The incorrect input combined with the chaotic output results in bad predictions.

Weather forcasters typically run 3-4 computer models. Then they estimate which one is more accurate. This leads to inconsistancies. One day they may report Model #1's predictions. The next day they may report Model #2's predictions. This is also why different forecasters will often give drastically different forcasts.

 

[Spooner]

but, CMON, we should be able to tell what's going to happen for the rest of the DAY

all i want to know in the morning is how the traffic is and what the weather is going to be all day.

I don't think that's too much to ask.

 

[Nemesis77]

"A butterfly beating her wings in Australia can cause a thunderstorm in North America"

Predicting the weather is one of THE most difficult things you can do! Guess what they use those massive supercomputers for?

 

[Jzero]

Don't worry--I heard they're using chemtrails to change the weather. Soon, my friend....

 

[BoberFett]

And yet scientists want us to believe they can predict what the human affect is on the environment by modelling it out 50 or more years.

 

[dullard]

<< "A butterfly beating her wings in Australia can cause a thunderstorm in North America"

Predicting the weather is one of THE most difficult things you can do! Guess what they use those massive supercomputers for? >>



Exactly I use those supercomputers for my work (similar to predicting the weather). Computers are about 1/10th the speed that I would like them for my simple problems. I bet complex weather modelling requires computers 100 times more powerful than we currently have. That and our models aren't perfect to begin with. And the input data is too sparce and often incorrect. And weather people just randomly pick one model to report... No wonder they aren't accurate.

 

[dullard]

The most common problem is that they accurately predict the temperature/precipitiation - but they are 25 miles off. For example last year they predicted one of the biggest snowstorms ever recorded will hit some big cities in the Northeast. It didn't happen (the huge storm occured but it just missed the big cities by a few miles). So the millions of people in the cities were angry saying that their predictions were terrible!

 

[jahawkin]

<< And yet scientists want us to believe they can predict what the human affect is on the environment by modelling it out 50 or more years. >>



Global climate models are much different than operational models used to predict short term weather. So, scientists can predict the climate in 50 years with accuracy. They are usually predicting global average temperature, where as the short term models are used to predict temperature, precip, etc in a very localized area.

 

[BoberFett]

<< They are usually predicting global average temperature, where as the short term models are used to predict temperature, precip, etc in a very localized area. >>

And those local temperatures, precip, etc. all affect the future. Butterfly flapping its wings, man.

 

[Spooner]

"Nobody ever suspects the butterfly..."

 

[Beau]

"A butterfly beating her wings in Australia can cause a thunderstorm in North America"

A butterfly flaps its wings in Peking, and in New York it rains instead of sunshine. It just points out that there are far too many minute variables. In this instance the butterfly flapped its wings, causing dust particles to be inhaled by a single cow, causing it to sneeze, which startled the rest of the herd, causing a stampede, which changed the immediate wind patterns: dominio effect with the jetstream. Eventually it effects the North American jetstream, offsetting the low and high pressure zones and causing the storm that was in ontario the day before to miss New York.

Now tell me, what super computer could have predicted that? Jurassic Park had some great, more lay descriptions of the chaos theory. Good read if you haven't done so already.

Beau

 

[EagleKeeper]

You want a good idea of what the weather will be the next day.

Get the local area weather map. Look at the weather 200-300 miles west of you.
Look if any fronts are pushing through and what/where the jet stream direction is.

Most most areas of the country, that will give you a clue what to expect the next morning.

With this info, you can equal your local TV/Radio weatherperson.

 

[AndrewR]

Sure, we have the butterfly saying, but I think it boils down to this, written above: "And the input data is too sparce and often incorrect". In order to accurately predict the weather, we need to input a great amount of information, like surface temperatures with a high density of coverage. However, sensors are limited, and computing power is limited so the complete picture is never available.

Maybe someday.

 

[tcsenter]

<< Sometimes I feel that these guys are making nothing more then an educated guess. >>

Your feelings are correct, because that is PRECISELY what weather forecasters are doing, even they will admit that. No secret there.

 

[Logix]

As that old Enron commercial said, "Welcome analysts.... everyone knows you can't predict the weather...."

 

[singh]

Don't worry... it will get A LOT better when Quantum computers become mainstream.

 

[Hayabusa Rider]

Spooner- I see you live in Boston. I'm in Quincy. Think about this. We live right on the coast. At this time of year the water is warmer than the air. The air above the ocean is therefore warmer and moister. A difference of a mile or two makes a great difference. The cable stations here show Norwood temps. (for those not familiar with this area, Norwood is about 7 or so miles SW of Boston). In the winter, the lows can be 5-15 degrees cooler. Sometimes more. With such great temperature and moisture variations occuring in such a small area, you can expect unpredictable weather changes in a timescale of a few hours on occasion. Just the way our geography and topography are.

 

[kgraeme]

Nah, the weather reporters just drop acid and look at the pretty colors on the map/screen.

"Dude! It's snowing!"
"Phil, it's July."
"No dude, it's snowing!"

 

[rudder]

We can now predict the weather with a 90% accuracy rate (for the next 24 hours), then its is something like 75% accurate for a 72 hour forecast and basically a 50% guess for a 5 day and beyond forecast.

Cloud formations are different everyday which results in the solar heating of the earth's surface to vary and never be the same. Thats just one reason, but there are too many unpredictatble factors that come into play. Much of forecasting is based on history and how the current atmospheric settings behaved in the past. When a forecaster says we have a 40% chance of rain, they are actually saying that under these specific conditions atmospheric conditions 40% of the time it has rained in the past.

 

[EagleKeeper]

Those of you on the Mass shore, track the weather in Albany and south of NYC.

Compare what they had yesterday to what you get today and tomorrow.

You will pick up a pattern quickly and can use that to account for the differences.

Many moons ago, when I lived in central Mass, for a HS science class, we were graded on our predictions.
By using this method I was able to come within 2 degrees most times for the fall semester and identify wind rates and reasonable precip rates fairly closely.

Around the central area of 128 vs south shore vs the NH border there will be differences, but the differences will usally be consistent.