Why are the weathermen so often wrong?

I was thinking of this as the NY area hunkered down in anticipation of a predicted once-in-a-lifetime blizzard that never came. As I see it, there are two possibilities:

  1. The science. Possibly the science of meteorology is at a stage at which the forces that control the weather are still poorly understood. Thus scientists may know that under a given set of conditions the weather may have a 60% chance of doing X and a 40% chance of doing Y, but not fully understand what differentiates the times that it does X from the times that it does Y.

  2. The data. Even if the weather was perfectly understood, it may be that the weather is shaped by so many factors that it is not practical to gather up (or to build a forecasting model that would assimilate) the mass of data that would be necessary to make the predictions more accurate.

Or some combination of the two.

The significance of the question is in what the likelihood is that we will see some significant improvement in these predictions. If its mostly a lack of science, I imagine progress will march on. But if its a data issue, this may be the best we can get.

Perhaps there is a perception issue here? You expect the weatherman to get it right, so it sticks in your memory when they get it badly wrong. ( a little like the mind-reader’s trick in reverse – you ignore the bad guesses and are duly impressed when they get it right, or even close)

There is a classic example from over the pond. In 1987 (IIRC) a weatherman by the name of Michael Fish announced live on air that he had received a letter from a psychic asking him to warn the nation that the country was about to get hit by the most appalling storm. Pride comes before a fall, so he laughed and reassured his viewers that this was not going to happen. We all woke up the next morning to the scene of the worst devastation in living memory, roofs blown off houses, trees fallen all over the place.
I couldn’t provide a cite, but I recall reading that in the UK weathermen get it right 80% of the time. You’d be right 75% of the time if you said the weather tomorrow will be the same as the weather today.

I remember the incident quite well - I was watching the bulletin when he said it. If it was a psychic he did not say so on air. The letter said it was going to be a hurricane and in a technical sense he was correct, it wasn’t. He was too gracious to use that in his defence however, as he was still spectacularly wrong. Funnily enough the week-long forecast for farmers on the previous Sunday (the storm was early Friday), which is generally vague and inaccurate past Wednesday, did mention a possible deep low in exactly the correct place.

Aside from your points 1 and 2, there is another factor:

If you predict snow of biblical proportions and no snow comes, everybody is relieved & you get a little bit of bad press.

If you fail to predict snow and the whole eastern seaboard gets 3 feet, you get fired, so forecasters always err on the side of snow.

Forecasting the weather seems to be a lot like predicting what the stock market will do. People often get the general direction of the weather right but fail when it comes to the details. Also it’s probably not so easy to apply well-understood laws of fizziks & thermodynamics (and whatever else applies to the pressure & heat of weather systems) because you can’t get a good grip on the quantities of air & an exact measure of all of the energies involved.

Only after the storm failed to materialize can it be realized that the large cold air mass imported from Canada did not have the personality we thought it did.

(1) Weathermen are not scientists; they are historians.

(2) TV weathermen want to trump the other guy. The public is more apt to watch the weatherman who is predicting 18" rather than the weatherman who is predicting 8". There is a frenzy to see who can predict the most.

Apologies, Izzy, but this seems like a good place for a hijak.
Since we’ve established that weather forcasts are imperfect for a number of reasons, who does the best job? I’m not in the Northeast, but I’ve been burned by the Weather Channel a few too many times and am searching for an alternative source. I got a little application called Weather Bug in hopes of getting more up to date and accurate forcasts, but all it seems to do is channel a local TV station’s forcast to my desktop. Where can a poor pedestrian turn for proper prognostication? (meteorologically speaking, of course) Are they all pretty much the same?

The news report I heard said that there was a possibility of a huge storm. Just because the storm didn’t occur doesn’t mean they were wrong. I, personally, would rather hear about potential storms a ways off, with the understanding they may not materialise, than only find out when they are 100 percent sure, but the storm is on top of me.

Apart from that, I believe your point 2 is a big factor. Lack of computing power is also a factor. I recall much more detailed micro-predictions being made for some event several years ago (Might have been the Atlanta Olympics, might have been for some computer get-together in Silicon Valley. Probably by IBM at any rate.), using more data and a finer gridding of the atmosphere. Putting together a detailed prediction for a localised area is one thing. Doing the same over most of the East Coast is another.

Hey, it ain’t our cold, dry air that’s responsible; it’s your warm, moist stuff from the Gulf of Mexico. Cold, dry air doesn’t snow.

Warm moist air from the Gulf had nothing to do with the NE storm. There’s plenty of warm moist air right off the coast there, in the Atlantic. A “nor’easter” is caused by a low pressure developing and deepening off the coast of the mid-Atlantic states. This latest storm had many meteorological features that indicated such a low would not only form, but stall: another low pressure moving down from Canada and another one moving up from the Gulf.

I live in SC so I’m not familiar as to which cities, if any, got a lot of snow, but I understand many did. I also understand that temperatures remained a little too high in many places for it to snow, and it began as rain. In addition, even where it was below freezing, it was above freezing at cloud level, and they had sleet or freezing rain. That was a very tricky forecast, as it depended upon the exact position of the Low. Apparently, it was in a position that threw a lot of warm air from the ocean, enough to raise the temperatures sufficiently.

I don’t think you can fault the computers completely. It’s the data that is imputed or the program. I know here that during the hurricane system, meteorologists use over half a dozen computers to predict the path of a hurricane, and it is rare that all of them agree. Weathermen get a lot of flack down here, too, when a predicted hurricane doesn’t hit us, or remains in the ocean. Then everybody reacts the same as you guys: Hey, why did they scare us so? You caused us to board up the house, stock up, etc. for nothing. I say better safe than sorry.

In the hurricane scenario, the same data is inputed to different computers. It is how the computers are programmed to interpret that data that makes the difference.

Just a few months ago, a marathon in Raleigh was postponed because a terrific snowstorm was predicted. The organizers were in a quandary. Would it better to go ahead and try to hold the race with so many people coming from all over the country? The police said they could not patrol the course if they had a snowstorm. So they postponed it. And everybody with prepaid rooms, etc. had to go home. You guessed it. It didn’t snow.

Weathermen say they are right 80% of the time. But what is their criteria? If they predict p.c. skies with 40% chance of rain, and it is mostly fair with no rain, are they wrong? Or if it is p.c. and it doesn’t rain, are they 60% correct?

As to which is better, the Weather Bureau or some private organization, such as Accuweather, I haven’t noticed much difference. If you access Accuweather on the web, you will notice that they predict the temp., sky condition, and precipitation (if any) for every hour in the day. I’ve relied upon their temp. forecast to wear the appropriate clothing at a tennis game, much to my chagrin, when they were totally wrong. I have not noticed that any of their hourly predicted temperatures are accurate.

If you access http://www.weather.com you will see that they now have an extended extended forecast, similar to what Accuweather has had. You can take those forecasts to the bank. :smiley:

Pity the poor meterologist.

I live in Missouri, the area that prompted Mark Twain to say “If you don’t like the weather, just wait awhile.” I have seen many many occasions where the predicted storm misses us by 40 miles.

But hey, the meterologist is predicting what it’s going to do within about a 75 mile radius (that’s our market area). If the storm winds up hitting 40 miles from downtown, the poor schnuck still got it right. It’s just that 90 percent of the viewers didn’t get dumped on.

And I’ve seen viewers get downright nasty when the TV stations interrupt for tornado warnings two counties away. First, the people two counties away watch that TV station too. Second, have these people ever BEEN in a tornado? I have, and I’d rather sit through a warning than sit through a tornado.

I’ve read that to get an accurate forecast, there’d have to be a weather station for every ten square feet of land. If you’ve read about Chaos theory, you might appreciate how difficult it is to predict the weather. Correct me if I’m wrong but a butterfly flapping it’s wings in China will necessarily change the weather in Los Angeles. A hurricane probably won’t result, but it states that even the most minute changes to a chaotic system (as the weather is) will make a large difference to the overall system. I’d like to know more about Chaos but I’ve not got a mathematical brain and bits of it are hard to get my head around. But it’s complex - and that’s just the explanation of how its worked out…

Yeah, I remember Chaos Theory. I was going to post it, too, it’s just that I had 5 minutes before class started when I was reading this. yeah, that’s it…

There is no way that every possible variable could be measured. However, even if there was some magical way to measure every variable, we just don’t have the computing power to deal with all of them. Even if you could build a computer powerful enough to compute the data, the heat it would generate would change the weather. :smiley:

That would be ironic. The power needed to cool that beast would use loads of electricity pumping even more greenhouse gases into the atmosphere. If you can’t predict it, just change it to suit you.

On another point - and I was planning to ask this before I saw this thread. How do weathermen calculate the sun’s strength into their temerature predictions. Do they have a chart which tells how strong the sun is at a particular time of year in a particular place, or do they jyst extrapolate form the previous day’s temerature? If the latter, what if it’s been cloudy for the previous month?

They don’t. Temperature predictions are for the shade. In the summer time, it could be 30 degs warmer in the sun. The temp. predictions are based on the air masses and the temperatures of those masses.

when the weatherman says that there’s a forty percent chance of rain, what he means is that in forty percent of the past situations like this for which we have data, forty percent of those days had rain.

bizerta is right; it has more to do with statistics and history than with anything else.
jb

p.s.- how come statistics with weather are taken with a grain of salt, but when the same methods are applied to the election, everyone swallows them down like a cosmic load? Damn Americans, the whole lot of us…

I don’t think so. When a 40% chance of rain is predicted that means that a weather system is approaching which contains rain in 40% of the area over which it passes. If you look at the Weather Channel and note the precipitation pattern of an approaching system, and 40% chance of rain is predicted, you will see that over 40% of the area rain is falling.

40% can mean one of two things: 40% of the area will get rain or there is a 4 in 10 chance you will get rain. However, it amounts to the same thing as both are correct.

As a side note to my prior note, when does the Weather Bureau count a forecast as correct? How close do they have to come? For example, if they predict a high of 70, but it only gets to 67, is that correct?

Yeah but in the summer it’s warmer than in the winter. This is because of the Earth’s tilt with Northern or Southern hemisphere towards or away from the sun. So obviously the sun has an effect on temperature depending on the time of year…

Cecil Adams on What does it mean when “X percent chance of rain” is predicted?

Let’s say you’re about to take your turn in a game of backgammon. For no reason, I state that you are most likely to roll a seven. I’m making a statement of probability. That is true whether you roll a seven or double sixes (or anything else).

A weatherman does the same thing. He looks at current conditions, compares them to times in the past when conditions were the same or similar, and uses this information to determine what the weather will most likely be like tomorrow (or the next day, or whenever). Of course, I think most the work is done by the computer.

Just like when I state that the most likely result will be a roll of seven, I am not wrong when you roll a six, the weatherman states that there is a possibility of a storm, he is not wrong if the storm doesn’t happen.

What’s that? You don’t want to take my word for it? Maybe you’ll take Cecil’s.

From this column: What does it mean when “X percent chance of rain” is predicted?

manhattan just stole my thunder!