Guest post by Steven Goddard
On July 23, 2009 the UK Met Office issued their infamous winter forecast, ahead of the coldest winter in 50 years. It read:
“…Early indications are that winter temperatures are likely to be near or above average over much of Europe including the UK. For the UK, Winter 2009/10 is likely to be milder *(and wetter) than last year “.
This was recorded by Piers Corbyn at Weather Action and several other sites on July 23.
Source:
http://www.metoffice.gov.uk/science/creating/monthsahead/seasonal/2009/winter.html (released 23 July)
I remember reading the article on the Met Office web site at the time. But something funny happened on December 30, 2009. The Met Office over wrote that link with a new article titled “Forecast for the rest of Winter 2009/10” which has no mention of the original prediction. It now reads:
…for the rest of winter, over northern Europe including the UK, the chance of colder conditions is now 45%; there is a 30% chance of average and a 25% chance of milder conditions.
Their original warm winter forecast seems to have been scrubbed from the web site, and there are no longer any press releases dated July 23.
http://www.metoffice.gov.uk/corporate/pressoffice/2009/index.html
Other sites which noted the July 23 Met Office article and link include:
http://climaterealists.com/index.php?id=3758
According to The Independent, the winter forecast seems to have been updated on September 29, but the Met Office no longer has any press releases with that date either.
The Met Office came under tremendous fire as a result of their disastrously bad winter prediction
The Big Question: Should the BBC drop the Met Office as its official weather forecaster? By Michael McCarthy, Environment Editor
And this lead them to drop their seasonal forecasts, which have been notoriously poor in recent years. What could have motivated them to destroy their original winter forecast?
======================
[From AW– Note: unlike government services in the USA, the UK Met Office gets bonuses, see their benefits package:
http://www.metoffice.gov.uk/corporate/recruitment/benefits.html So, this might be an incentive to remove poor work products.
The Times did a story about it last August after the BBQ summer fiasco: It’s raining bonuses at the Met Office
And the Met Office Chief, despite botched forecasts, got a 25% pay increase in January 2010, according to this Telegraph story:
Met Office chief receives 25 pc pay rise
The head of the Met Office, the national weather service which has been heavily criticised for getting its forecasts wrong, is now paid more than the Prime Minister, after receiving a 25 per cent pay rise.
]
======================
———————
For the record, here are a few of of their other classic mis-predictions:
2007 – forecast to be the warmest year yet Wrong – la Nina hit and temperatures plummeted.
Met Office forecast for Summer 2007 Hot summer – Wrong – it was the wettest summer on record with cold daytime temperatures.
A typical British summer 2008 Wrong – it was the second wettest summer on record with cold daytime temperatures.
Trend of mild winters continues 2008 Wrong – it was the coldest winter in 15 years.
Summer forecast 2009 “Barbecue Summer” 2009 Wrong – another miserable washout of a summer.
Warming could push Greenland ice sheet beyond ‘tipping points’ Complete nonsense
And if they are weren´t trying, how would they ever learn to make better seasonal forcasts?
🙂
“Steve Goddard (19:35:05) :
Both are usually less accurate than a random number generator.”
LOL A random mumber generator is very accurate, every time it runs in fact.
These bozos have been using the global warming models and adjusted data and then have the gall to wonder why they have been wrong 9 out of the last 10 years. Doh!
Mattias, Sweden (20:05:34) :
I check the CPC forecasts all the time and they are usually inverted from what happened. They forecast a warm winter for the central and southern US twelve months ago.
There is no reason to believe that the models will ever be accurate more than a few days out, due to chaos. Weather modelers understand this and it is surprising that climate modelers don’t. Climate models are basically weather models with a few more considerations (like atmospheric content.)
Patrick Davis (20:07:53) :
A good random number generator is one which generates a nearly perfect gaussian, i.e. it is completely unpredictable for small populations, and completely predictable for large ones.
A seasonal forecast is a very small population of one.
”””””Mattias, Sweden (19:31:11) : Seasonal forcasting is built on statistics . . . . Still I think it isfun that seasonal forecasts exist. I know that they are uncertain but I would miss them if they were not to be produced anymore. . . . . Sorry for the long post. . . . .”””””’
Mattias,
I think the MET just didn’t do their statistical homework honestly/objectively on the seasonal forecasts. The MET was telling an AGW story that they wanted us to hear. Someone else will now do the seasonal forecasts, hopefully in a more honest and statistically proper way.
Goodnight.
John
Before they can forecast the weather or the climate by extension, they have to understand how the weather really works…
http://research.aerology.com/aerology-analog-weather-forecasting-method/
Ideas and supporting research on how the “chaotic system” is just a compound signal that can be demodulated, to extract the intelligence signals out of the composite “Looks like noise” signal. By solving for more of the Natural variables, it will make the remaining “unknown CO2 influence” in the equation easier to solve for.
Based on the repeating patterns found in the research, in 2007 I generated Maps for the following 6 years of “daily forecast for the USA”. These are still viewable for the next 4 years, (till January of 2014.)
http://www.aerology.com/national.aspx
I’d just like to add that Joe D’Aleo’s winter forecast for the US was spot on in case some have forgotten.
“The Big Question” is should the BBC consider hiring him instead.
A good way to test your random number generator is this:
static const unsigned int COUNT = 10000;
std::vector count_vector(COUNT );
for (int i = 0; i < COUNT; i++)
{
count_vector[i] = 0;
}
for (int i = 0; i < COUNT; i++)
{
unsigned int sum = 0;
for (int j = 0; j < COUNT; j++)
{
sum+= rand() % COUNT;
}
unsigned int average = sum / COUNT;
count_vector[average]++;
}
When you are done executing this, you should have an perfect gaussian distribution in count_vector centered around COUNT / 2. Of course, chances are your random number generator is not very good and the distribution is not perfect.
Don’t ever try to post C++ code on web discussion forum. It thinks angle brackets are html tags.
If the Met office record is comparable to the level of accuracy of NOAA, then they ought to be forgetting about bonuses — even for showing up for work on time.
For days, all we heard was how wet Sunday would be here in Buffalo. The rain was to depart on Monday, with the sun returning in the late afternoon. Right. Most of Sunday turned out sunny and mild with high clouds. Trace of rain late night, with solid rainy periods on Monday all the way through a chiily Tuesday drizzle. Sun didn’t reappear till Wednesday. Good thing I don’t count on them for anything other than a SWAG. I just followed the progress of the storm through the kinks and curls of the jet stream, as its progress was slowed to a crawl. Supposed to get some rain tomorrow late afternoon; guess I’d better get my grocery run completed first thing in the morning to miss the bulk of the rain; who knows what Friday will bring – they’re guessing sunny and cold.
As for the long range forecasts — does anyone ever really pay any attention to them?
What would Magnus Pyke say all about this?
Another thing that has disappeared…..
Anthony…any chance to make this a feature post and invite the experts like Morner….I am sure he would love to chime in…
http://news.yahoo.com/s/ap/20100324/ap_on_sc/as_india_disappearing_island
Damn ASSoc. Press….this is their clincher?
How much is due to sinking in one of the world’s largest, lowest-lying deltas….not to mention other things at work here (not to mention the 700-mile 40-foot jump of the tectonic plate that caused the Boxing Day Tsunami in 2004)??
Let the truth be discovered….
Caution, though….it won’t be found in the annals [or should I say anals] of the ASSociatied Press.
Chris
Norfolk, VA, USA
John Whitman (20:20:56) :
“I think the MET just didn’t do their statistical homework honestly/objectively on the seasonal forecasts. The MET was telling an AGW story that they wanted us to hear. Someone else will now do the seasonal forecasts, hopefully in a more honest and statistically proper way.”
Well, I am not going to defend Met Off. 😉 I wanted to tell that long time range forecast of parameters like temperature and precipitation is at the current knowledge level are difficult to make. You can for example try to use parameters like the current ocean temperature, that will likely affect the coming month temperature, in ocean near states, because of the large water heat storing capacity. But there are so many other things that will mess up the forecast. Like weatherpatterns corresponding to positive or negative AO or NAO index.
Can anyone here forecast volcanoes, sand storms, forest fires, sunspots, asteroids, magnetic storms, clouds, ENSO or ice twelve months out?
If you can’t, then you can’t forecast the climate twelve months out.
Steve Goddard (20:15:46) :
Mattias, Sweden (20:05:34) :
“There is no reason to believe that the models will ever be accurate more than a few days out, due to chaos. Weather modelers understand this and it is surprising that climate modelers don’t. “
Weather forecasters are, at least to some extent, market tested. Climatologists are not, ever. They are all on the government tit and are also the dumb men of science. They are the guys who couldn’t do chemistry and physics.
“Steve Goddard (20:19:26) :
A good random number generator is one which generates a nearly perfect gaussian, i.e. it is completely unpredictable for small populations, and completely predictable for large ones.
A seasonal forecast is a very small population of one.”
Well that is the point isn’t it? A RNG works, perfectly, every time (say for a gaming/wagering/betting organisation in my experience) predictable or not.
Modelled seasonal “weather” forecasting doesn’t (For The Met, BoM, NIWA etc etc in my expience).
NickB. (19:57:17) :
The Way Back Machine didn’t archive it? Is that thing still around?
——-
I checked the way back machine and they only had one page newer than 2007 from the uk metoffice.
http://web.archive.org/web/*/http://www.metoffice.gov.uk/
Patrick Davis (21:48:36) :,
Met Office forecasts are binary Hot/Cold Wet/Dry . A random number generator would have a 50/50 chance of getting it right.
Met Office does worse than that because they always predict warming, while the climate has been cooling the last few years.
Anthony, I regret to inform you that the US government does give bonuses. It gave bonuses to SEC employees who failed to detect the hollow condition of several big investments banks a couple years ago.
The DOD even gives bonus progress payments to defense contractors who have not yet begun work. Or it used to.
Makes me wonder every time the Met Office pull a stunt! They cannot do a “prediction” for 6 months ahead and (I love this comment of theirs) because “the UK weather is very hard to predict”!
Ask anyone who has live a few years in the UK and they would agree but if they cannot manage 6 months how the hell can they “predict” years ahead?
Yep, its total nonsense (Nonscience?) but then again, it always has been.
Me, I will stick with Piers!
@ur momisugly John Whitman (20:20:56) :
“The MET was telling an AGW story that they wanted us to hear. Someone else will now do the seasonal forecasts, hopefully in a more honest and statistically proper way.”
– – – – – – –
As soon as MET figured out that 51% of the public didn’t want to hear their claptrap …
How can you be wrong for 10 years and still be in business? Do lives depend on these fools? Does anyone ever get canned?
well here’s one company thinking they might not need to stock as much gas in the future based on the first report:
“At the end of September the Met Office issued its early indications for winter 2009/101. They report that ‘preliminary indications continue to suggest that winter temperatures are likely to be near or above average over much of Europe including the UK. Winter 2009/10 is likely to be milder than last year for the UK, but there is still a 1 in 7 chance of a cold winter’.”
http://www.nationalgrid.com/NR/rdonlyres/C3A81245-D988-48A4-80F2-5082F601E06D/37301/Winter_Outlook_Report_200910_01102009.pdf
and this:
“With the recent trend of warmer winters there is the possibility that our 1 in 50 basis for severe conditions based on the last 80 years of weather data is now less relevant than for previous security analyses. We have been working with the Met Office and other energy companies to explore this.”
Wonder if they are still relying on the met office for ‘expert’ guidance…
crossopter (17:55:20) :
“This of course begs the question: why is an experimental, unproven forecasting method being used as the basis for a national press release?”
Simple: it’s the garbage they shove in, so that they can be sure of getting the required garbage out.