Sick Man Complains About Weather Forecasting Shocker
by John Walker on Dec.19, 2010, under The Rest
As I graduate from a flu so strong that I needed help to sit up, to a cold where my lungs and throat are in more pain than I know what to do with, leaning on my desk to stay upright simply because I’ll overdose myself on the pharmacy of drugs surrounding my bed if I have to lie down any longer, it’s comforting to know that I can still get furious about the mind-numbing incompetence of weather forecasting.
Sure, I’ve gone on about this a lot, and certainly I appear to be among very few who care, but in a world where people are increasingly calling bullshit on the various woo remedies and snake oils, it infuriates me that weather soothsaying escapes any scrutiny.
I think the only way to communicate the level to which it is palpable nonsense is to show not how the weather is incorrectly predicted, but how the predictions don’t even agree with themselves from the same source. So for today’s example, here’s the BBC’s current predictions for the weather for Bath tomorrow, Monday 20th December. First of all, the day’s breakdown:
So that’s a cloudy, chilly day, but no chance of disruption.
Let’s look slightly further down the page at the satellite map on which this information is based:
Um, well, unless they’ve changed their key for non-stop snow all day to mean “white cloud”, this is getting confusing.
Let’s try to clear things up by watching the television weather bulletin on the same page:
What’s this? A heavy snow warning for the South West? But maybe not Bath, because it’s going to get white cloud all day.
Oh, no, directly on top of Bath. (Compare that 15:00 map with the above 15:00 map to see how little they have in common too.)
So can we all please now start not accepting this? Whatever weather we do get tomorrow, I guess the BBC can claim they predicted it.
December 19th, 2010 on 11:43
I hope you received my photo the other day, where my computer screen was telling me how brilliantly sunny it was, but when I looked outside my window, I saw a massive blizzard.
Look, we all know that weather forecasting is, at best, warm reading (ho ho) with the aid of millions of pounds of supercomputers. But really, is it too much to ask that they report CURRENT weather accurately and consistently?
Now, this *is* important, because the data that’s being used to predict hurricanes and global warming is being fed by the same ridiculously inaccurate sources.
December 19th, 2010 on 12:11
*can’t be bothered to give the whole speech about chaos theory and just asks John to look it up*
I’ll give you the observation stuff though. I use Weather Channel on my Blackberry anyway.
December 19th, 2010 on 12:34
Can’t you just do what the rest of us do and not bother looking it up or paying attention to that bit of the news and just see what happens?
December 19th, 2010 on 12:48
DCD – that is the equivalent of posting on a blog entry against homeopathy and saying, “*can’t be bothered to give the whole speech about how water doesn’t have a memory and just asks John to look it up*”
I am aware that it’s impossible to predict weather. And I’m aware why. Which would, how can I put this?… BE THE POINT.
Dave – Ignoring lies and deceit which profit millions of pounds every year (including license fee money) may seem like an elegant solution to you. To me it seems spiteful.
December 19th, 2010 on 14:22
I shouldn’t have to say this to a sceptic, but the plural of anecdote is not data. The weather forecast will certainly fail, because its making millions of probabilistic predictions every day. As I understand it we have some of the best forecasting there is, although the fact that we are an island can often lead forecast astray. Its true the media rarely talks in terms of probabilities and errors, but thats usually the media’s fault rather than the statisticians.
December 19th, 2010 on 14:41
I’m sorry to you you’ve been so sick! *hugs*
Also this is the part where I’d post the Lewis Black weathermen lie bit, but I’m too lazy to google it.
December 19th, 2010 on 15:18
Mister K: the Met Office itself admits that its forecasting record is around chance (perhaps less than chance, if I remember correctly – John and I pulled the data some time ago).
So, in this instance, anecdote is indeed the plural of data.
Anecdotally, the weather forecast has not once been correct when it mattered for me in the last 5 years or so.
December 19th, 2010 on 15:38
My coat trick has worked mostly perfectly more then BBC whether out, oh it looks a bit overcast I think i will bring my coat and what do you know it rains. And vice versa though sometimes the skies can fool you into thinking it will be a nice day and then it puts you under a torrential rain 1 hour later.
December 19th, 2010 on 16:25
The BBC’s forecast for Kilburn has been accurate for the past week, so far as I’ve been able to tell. Driving from London to Basingstoke on Saturday morning, my mother told me that snow had been predicted to hit the town at 9 in the morning. We in fact encountered it still a few miles away at about 8:40. I’m not sure if this counts as incompetence on the weather forecaster’s part or not. Certainly, if we’d left twenty minutes earlier we’d have made it to our destination before the snow arrived.
December 19th, 2010 on 18:00
Here’s Freakonomics on this issue:
http://freakonomics.blogs.nytimes.com/2008/04/21/how-valid-are-tv-weather-forecasts/
December 19th, 2010 on 18:12
THis is also pretty good – a guy has written an automated script comparing the BBCs “prediction” of today’s weather against hat they said it would be yesterday and checking for consistency:
http://weather.slimyhorror.com/
December 19th, 2010 on 18:34
The Met Office’s own pages on the issue are laughable, relying on the old “if we show some graphs that point upwards then it means we’re improving” gambit so beloved of pseudoscience everywhere.
http://www.metoffice.gov.uk/about-us/who/how/forecast-accuracy/uknwp
It would be nice if that page explained a solitary damn thing about what the “NWP Index” actually is, eh? Still, they’re hitting their target, whatever it means, so yay for them I guess.
A bit more digging reveals the following graphs:
http://www.metoffice.gov.uk/about-us/who/how/forecast-accuracy/temp
http://www.metoffice.gov.uk/about-us/who/how/forecast-accuracy/rain
December 19th, 2010 on 19:42
Weather forecasting here in the Midwestern US is pretty reliable two days out, but there’s no Gulf Stream and no turbulent straits or seas anywhere nearby to stir things up. The Great Lakes are there, but they’re fairly predictable. If you’re east of one when moisture comes from the west, you’ll get lots snow.
December 20th, 2010 on 10:15
hmm, if you are correct Nick thats rather surprising- I honestly expect the met to do better than that. Sadly such information is not readily available from a quick google search. I may be making the mistake of assuming that, considering that they choose to report it, the 5 day forecast is better than chance. That said they’re more likely to be correct a day in advance, and usually are.
December 20th, 2010 on 11:01
Mister K: their forecasts are on average slightly worse than 50% accurate. One might say that this is nevertheless more accurate than chance, because of the number of variables that one needs to get correct. A more interesting comparison would be with an amateur’s educated local’s attempts to predict the weather compared with super-computered-multi-boffined Met Office. I think such data would be rather revealing.
December 21st, 2010 on 00:04
Life becomes a lot simpler and more accurate when you ignore all weather forecasts entirely, and simply stand outside for a bit and hazard a guess about the rest of the day. Your success rate will likely be a lot higher than the telly’s anyway.
Also, my sympathies. I believe I may have had the same thing you did – bloody awful day yesterday. 5 straight hours lying still clutching my head and gently sobbing.
December 21st, 2010 on 08:21
That freakonomics post was interesting, Thomas. A local radio station here runs a contest tied to their weather forecasts, and the prize can be won… yes, on days when their meteorologist is off by more than three degrees.
December 21st, 2010 on 08:33
Alex, it’s interesting that they use temperature – rather loading the dice in their favour. Three degrees of latitude either way in a particular season gets you a band of around 80% accuracy before you actually do any work! This is thr sleight of hand scam of a mentalist conjourer
December 21st, 2010 on 22:19
The main thing is that … I hope you are feeling better John.
December 24th, 2010 on 06:44
Alex, it’s interesting that they use temperature – rather loading the dice in their favour. Three degrees of latitude either way in a particular season gets you a band of around 80% accuracy before you actually do any work! This is thr sleight of hand scam of a mentalist conjourer