Buy

Books
Click images for more details

Support

 

Twitter
Recent comments
Recent posts
Currently discussing
Links

A few sites I've stumbled across recently....

Powered by Squarespace
« GWPF calls for inquiry into Met Office | Main | Quote of the day »
Tuesday
Dec212010

Winter resilience

With the whole of the UK apparently grinding to a halt with the cold and snow, it was interesting to be pointed to an official review of the UK's winter resilience capabilities (H/T John B).

A small team was set up under the leadership of Dr David Quarmby, a member of the "great and good" with background in transport. The team published its terms of reference here; an interim report was published last summer, and the final report appeared just a couple of months ago.

For our purposes the interim report is more interesting since it has a whole section entitled "Weather forecasting and climate change". All emphasis below is added by me.

12.7 The science of forecasting up to 30 days ahead and beyond has made great progress in recent years and will continue to develop; comparison of outturns against probabilistic predictions out to 30 days suggests that the information is of increasing value for winter service resourcing and planning.

"Increasing value" eh? I wonder what the absolute value of these forecasts is though?

12.8 The Met Office has ceased publishing seasonal forecasts through the PWS, because – again – the nature of the weather and climate means that at these timescales it only makes sense to give probabilistic predictions rather than definitive forecasts, and this has proved difficult to communicate.

"Difficult to communicate"? Don't they mean that they were wrong? I'm struggling with the idea of a "barbeque summer" that turned into a rain-drenched washout being a communication problem.

12.9 Yet, as this Review makes clear, critical policy and strategic decisions would be enormously enhanced by even a probabilistic prediction about next winter’s weather. Forecasting over this timescale and beyond takes us into the area of climate forecasting and the impact of climate change.

12.10 We have explored these issues in some depth with the climate research team at the Met Office Hadley Centre. The starting point is the slow but steady rise in average global temperatures. The consensus on the UK is that on average summers will become warmer, and winters will become warmer and wetter, though the next 10–15 years may be dominated by natural variability. When severe weather events happen they may be more extreme in terms of heat and rainfall.

Aha! So the Met Office were involved, and told the review that winters were going to become warmer and wetter. What else did they say?

12.11 Although the probability of severely cold winters in the UK is gradually declining, there is currently no evidence to suggest similar changes in extremes of snow, winds and storms in the UK.

12.12 We have also explored whether or not the occurrence of two successive severe winters influences the probability of a third in succession – in other words, is there any evidence of clustering? There is some small influence from year to year but these matters are still very uncertain and it would be safer to assume that there is statistical independence between one winter and the next.

12.13 In other words, we are advised to assume that the chance of a severe winter in 2010–11 is no greater (or less) than the current general probability of 1 in 20.

Now didn't the Met Office tell us just yesterday that they didn't make any predictions on the weather for this winter? I would have thought many people might have mistaken the words highlighted above as discussing a forecast of some kind. Perhaps it's another of those communication difficulties.

12.14 For the purpose of this report, the following summarises what we understand:

  • The probability of the next winter being severe is virtually unrelated to the fact of just having experienced two severe winters, and is still about 1 in 20.
  • The effect of climate change is to gradually but steadily reduce the probability of severe winters in the UK.
  • However, when severe winters come, they could still be extreme – in terms of snowfall, wind and storms, though not necessarily in relation to temperature.

12.17 But we need to understand and accept that the chance of a severe winter is still relatively small and that there will be many years when some will question the degree of resources committed to winter resilience.

12.15 An important consequence of the declining occurrence of severe winters is the loss of knowledge and experience among planning and technical staff in local highway authorities and their contractors, especially if the severe winters which do occur have more extreme snow events.

12.16 All this, in our view, reinforces the need for comprehensive resilience planning, and for ensuring that the salt supply chain is resilient.

After the publication of the interim review, the team recommended that the UK import 250,000 tonnes of salt to cope with a possible shortfall. I wonder if this was (a) enough and (b) actually done in practice.

Either way, this looks like more trouble for the Met Office.

 

PrintView Printer Friendly Version

Reader Comments (70)

macsporan is another zeds like tights gift incubated and unleashed by cacc?

Dec 22, 2010 at 2:40 AM | Unregistered CommenterShub

Hey wasn't it cold by November.
Predicting that it will be cold when it's already IS cold is quite skillful.

Dec 22, 2010 at 7:00 AM | Unregistered CommenterAusieDan

It is interesting to note that this 1-in-8000 probability occurrence of three severe winters in a row also happened in 1939/40-1941/42 and that there is good historical evidence of similar (or longer) series of severe winters e. g. in the 1430's, 1690's and 1810's.
What it all suggests is that the Met Office dogma that the probability of a severe winter is independent of the character of the previous winter(s) is quite simply wrong.

It is interesting to note that in October 1941, Franz Baur, the leading german expert on long-term weather prediction advised the Wehrmacht that the coming winter was unlikely to be severe, since three cold winters in a row would be unprecedented (details in Bulletin of the American Meteorological Society 68:620-629) .

"When will they ever learn?"

Dec 22, 2010 at 9:44 AM | Unregistered Commentertty

CET data
2005 10.44 degrees C
2010 8.858 degrees C (to Dec 20th)

Unless there is a serious turn around in weather between now and the New Year I don't think that the two years are going to compare at all. The fact that London temperatures remain static calls into account the lack of knowledge of UHI by the experts.

Dec 22, 2010 at 9:59 AM | Unregistered CommenterLord Beaverbrook

Many years ago when I was reading meteorology I was taught that the way to discern whether your forecasts were any good was to compare them with the ”persistence forecast”, i. e. “the weather will be the same as now”. If you couldn’t beat that for accuracy your forecast was worthless, since the persistence forecast requires zero knowledge to make.

Now let us have a look at the Met Office annual forecast (I have deleted 2009, sinmce the Met Offic apparently didn’t commit itself to a specific figure that year):

Year……….Actual (CRU)………Met Office……..”Same as last year”
2000………….0.24………………..0.41………………..0.26
2001………….0.40………………..0.47………………..0.24
2002………….0.46………………..0.47………………..0.40
2003………….0.46………………..0.55………………...0.46
2004………….0.43……………….0.50…………………0.46
2005………….0.47………………..0.51………………...0.43
2006………….0.43………………..0.45……………..…0.47
2007………….0.40……………….0.54……….………..0.43
2008………….0.31……………….0.37…………………0.40
Average………0.400………………0.474………………0.394.
Average error………………………+0.074…………….-0.006

Not only has the Met Office forecasts shown zero skill, the error is actually more than an order of magnitude worse than the “persistence forecast”!

Dec 22, 2010 at 10:49 AM | Unregistered Commentertty

@ ThinkingScientist

This is so close to the uniform distribution of 33.3% for each outcome as to be completely meaningless.

This reminds me of the bank analyst who, in 2008, predicted that the price of oil would be $200 by year end, but also predicted that it would be $25. He gave each prediction a lowish probability. If $200 oil had happened he could then have said "See, told you that would happen", and if it didn't he could then instead say "See, told you that probably wouldn't happen".

An analysis I have seen of bank analysts' oil price forecast in 2006 to 2009 showed a 0.0% correlation between their forecasts and the subsequent price outcomes, but an 81% correlation with the oil price at the time the forecast was made. In other words, their view of the future price was essentially based on today's price.

Something of the same mentality seems to have infected climate mugs generally. They see, or think they see, a small unremarkable temperature uptick, and they parlay this into a crisis. Meanwhile, they also hedge their position such that any day to day outcome can be claimed to be in line with a previous prediction.

As with the banker above, it's all about making a sufficient number of sufficiently widely-spread predictions.

Dec 22, 2010 at 11:43 AM | Unregistered CommenterJustice4Rinka

@ tty

I broke off from drafting my post and didn't see yours until just now, but broadly your description of the "persistence forecast" appears to be the same sort of thing as my example of bankers. Essentially they concoct elaborate models and qualitative rationalisations for oil price forecasts that, quite often, simply argue for the persistence of today's price.

Dec 22, 2010 at 11:47 AM | Unregistered CommenterJustice4Rinka

Why all the fuss about David King's interview on the Today programme in which he said that we had had a "hot summer" in the UK? I'm sure that everyone will agree that last summer was better than the "barbeque summer" that we had the year before. If a "barbeque summer" is hot then, by definition, one that is even better than that is bound to have been very hot.

Next summer will be even hotter still. Global warming statistics are like statistics for the output of tractors in the old USSR. Everybody knows which way the trend is.

Dec 22, 2010 at 2:29 PM | Unregistered CommenterRoy

hmc:

So, if you are right, the Met had two forecasting systems operating in parallel in October, the old system (OS) and the new system (NS). OS predicts a warm winter. NS predicts a cold winter. The winter is cold. The Met tells us (now) that the leading edge super accurate NS predicted this all along. Suppose the winter had been warm. Then the Met would have told us that their trusty reliable OS had predicted this (and that NS was not fully operational, "still being debugged", etc.). Despite the apparent contradiction, there is an inner consistency here: what Rick Bradford refers to above as a "consistent lack of personal integrity".

Dec 22, 2010 at 2:34 PM | Unregistered CommenterJane Coles

I really can't be bothered to discuss with the trolls like hmc but may I point you to the Met off site where you can use the drop down menus to get their forecast made in October. It show the whole of the NAtalantic and NW Europe to Iceland at 60% or higher for an above average december onwards. Now it is possible that the rest of winter may be 3°C above normal but their own recards show that the Jan-feb which have followed a record Dec have all been average or below. Now if we assume that each period calculated is independent then it is possible that the winter will end well above normal.

AQlso, when the Met Off announced their winter forecasts to the professionals there was complete amazement worldwide at how 'different' it was to all the independent (read private, earn a living from weather forecast that are correct) forecasts. It was not until after the uproar worldwide that they changed their forecast to 'early start'. Hardly the 100 year winter forecast much earlier by Joe B and Piers Corbyn.

It is time to clear out the Met off of all NGO members and bring in the professionals to run it. It is potentially the best service in the world but at the moment it is well, not the worst but certainly the most embarrasing. However, I am French now and I have to say that the Meteo here is absolutely appaulling ( affreux ). No one here takes any note at all of their forecasts although everyone watches them.

Interestingly, all the old farmers and gardeners here laugh at AGW. They just say c'est les cycles.

Dec 22, 2010 at 9:07 PM | Unregistered Commenterstephen richards

HaroldW writes:

"I think we've got to take David Quarmby's word for it that a warning was issued in October."

The reading that is most favorable to the Met is found in November but says only that the likelihood of below normal temperatures is 40-60%. That hardly qualifies as a warning. More of a coin toss, wouldn't you say? Bizarrely, the December "non-prediction" is roughly the same.

Dec 22, 2010 at 9:21 PM | Unregistered CommenterTheo Goodwin

From the "Resilience of England’s Transport Systems in Winter" report:

"8.41. Clear Service Level Agreements have been established with airport authorities at Heathrow and Gatwick on snow and ice clearance (BA)"

It will be interesting to see how that plays out, once the lawyers get their hands on it...

Dec 23, 2010 at 3:38 AM | Unregistered CommenterIan

"...it would be safer to assume that there is statistical independence between one winter and the next. ..we are advised to assume that the chance of a severe winter in 2010–11 is no greater (or less) than the current general probability of 1 in 20."

Yep classical probability - the toss of a coin. But here's the thing about probability from the toss of a coin. The chances of getting a head is 1/2 or 0.5. The chances of getting 3 heads in a row is 1/2x1/2x1/2 = 1/8 or 1 in 8

The chances of getting 3 severe winters in a row, each having a 1 in 20 chance of happening, is 1/20x1/20x1/20 = 1 in 8,000 or 0.000125 a very low probability.

In the normal course of things this is a one in a 8 thousand year occurrence, ever since Global Warming took over recently in the last 50 years, and we who are alive are really lucky (?) to witness it.

Kind of like breaking the bank at the Casino on opening night.

Dec 23, 2010 at 1:07 PM | Unregistered CommenterRichard

Given solar, ocean and other cycles it seems incredible to me that the Met Office works from an assumption that cold winters are statistically independent from year to year. As someone said earlier in this post, Roy Meadows was spectacularly wrong for making the same statistical error and was initially struck off by the General Medical Council.

The report on his error by the Royal Statistical Society makes for interesting reading. See the Statistical Contoversy section on wikipedia:

http://en.wikipedia.org/wiki/Roy_Meadow

Especially the part about the difference between "cause given effect" as opposed to "effect given cause". The closing line stating the perils of allowing non-statisticians to present unsound statistical arguments is also rather pertinent.

Dec 23, 2010 at 7:34 PM | Unregistered CommenterThinkingScientist

@ThinkingScientist methinks what you are trying to say when you come across an extraordinary coincidence series look for explanations other than coincidence. In other words suspect the occurrences are not independent of each other.

In the case you have cited, Roy Meadow rightly surmised that the deaths were not a coincidence, but wrongly concluded they were unnatural. There was an alternate explanation to murder.

In this case, the warmists say, this is just an unikely coincidence. An alternative view is, maybe not.

Dec 23, 2010 at 8:32 PM | Unregistered CommenterRichard

Richard, what Roy Meadows asserted was that it must be murder because the probability of 3 consecutive cot deaths was 1:73,000,000. He was wrong because his calculation assumes the events are independent when there are many natural factors which may give rise to a dependency and thus a high chance of multiple cot deaths within a family.

I believe the Met Office is guilty of the same error because they appear to consider cold winters as independent events. If this model is correct, we are witnessing an event which is likely to occur just once in 8000 years. Given the climate history of the last 8,000 years, this suggests to me the Met Office really does not have a clue about the causes of extreme winters, particularly as their long term trend is for winters becoming progressively milder which makes the odds of three bad winters in a row even longer.

I am a great believer in testing the validity of models with very simple "back of the envelope" calculations and on this basis I think the Met Office cold winter model fails. Their model is also inconsistent with a projected long term warming trend model resulting in progressively milder winters

Dec 23, 2010 at 10:37 PM | Unregistered CommenterThinkingScientist

Richard

I think you are being too soft on Roy Meadow. First he made an unfounded statement to the police that the probability of two cot deaths in one familiy was one in a million. Then at the trial he took a probability from a confidential report into perinatal deaths and squared it but DID NOT MENTION that the same report gave evidence of clustering in cot deaths.

If only defence counsel had asked him for the probabilities of:
a) finding Haemophilia in a male child (1 in 5-10,00)
b) having found it, then finding it in the boy's brother (1 in 2)

It is not only climatologists who recklessly abuse statistics in support of their doctrine.

http://www.bmj.com/content/331/7510/177.1.full.pdf

Dec 23, 2010 at 11:09 PM | Unregistered CommenterDreadnought

Dreadnought you are precisely correct with your example. There is a scientific explanation for a cot death, assuming the baby has not been deliberately murdered. The second (and subsequent) cot death(s) are thus not statistically independent from the first.

As a paediatrician, his ignorance of the scientific explanation for cot deaths was inexcusable. And using this ignorance to falsely send an innocent mother, no doubt already traumatised by the death of her infant, to jail for its murder, a crime.

No less a crime than willfully ignoring any scientific basis for colder winters in the midst of rising CO2 levels and "Global Warming".

Dec 24, 2010 at 12:52 AM | Unregistered CommenterRichard

@tty "It is interesting to note that this 1-in-8000 probability occurrence of three severe winters in a row also happened in 1939/40-1941/42 and that there is good historical evidence of similar (or longer) series of severe winters e. g. in the 1430's, 1690's and 1810's."

But in the previous years it was not a 1 in 8000 year occurrence. Remember according to the IPCC, courtesy of industrialisation, Global warming has only kicked in since 1950 (with increasing severity).

So since then, and specially this century, the chances of a severe winter, or indeed spotting any snow whatsoever, has diminished to 1/20. Thats whats makes this run of 3 so special.

Dec 25, 2010 at 8:39 PM | Unregistered CommenterRichard

@Karmakaze --

We know the last time this happened the Thames froze over and it hasn't yet. So clearly it's warmer now than then. What is the difference? Six centuries or so of global warming, perhaps?

The Thames didn't freeze instantly -- it was several years into the LIA. Another difference is that the Atlantic Multidecadal Oscillation is still in its warm phase, though headed downward.

The Met Office apparently has never heard of ocean cycles; if they had, they would know that if this year is unusually cold, the odds of next year being the same are much higher than otherwise. Haven't these computer bunnies ever heard of autocorrelation?

Jan 12, 2011 at 5:20 PM | Unregistered CommenterCraig Goodrich

PostPost a New Comment

Enter your information below to add a new comment.

My response is on my own website »
Author Email (optional):
Author URL (optional):
Post:
 
Some HTML allowed: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <code> <em> <i> <strike> <strong>