The Met Office has released all of its stored temperature readings "confident that they will prove its prediction of global warming". Useful.
Purely at random I chose a Midlands city - Oxford, which has recorded data going back to 1853.
The last year for which there is recorded data in full is 2008 so I looked at every 10th year from 1868 onwards. I thought it fair, since we are talking about proving global warming, to note the highest recorded temperature in June and in December. Anyone with a week to spare can look at other months and years.
The highest June days were recorded in 1858 and 1868 at 23.6C. The highest December day was 1868 with 10.4C [1898 and 1988 being 10.1C]. The lowest recorded noon temperature in June was 1888 at 16.3C.
The relative records for 1998 [the hottest year on record, we are told] are almost identical to those of 1938.
Returning to 2008 we read the June high to be 20.1C which is beaten by 6 years. December of that year had a high of 6.9C and higher December records occurred in 1858, 1868, 1888, 1898, 1918, 1938, 1948, 1958. 1978, 1988 and 1998. I could have shortened that by telling you that only on 4 occasions has the December high been lower than in 2008.
You may choose any other town or city and any other range of years. The few that I have looked at show no evidence of any warming trend at all unless you compare 2008 with 1928 or 1968, the only years in the sample of 16 which were cooler than 2008.
I am perfectly sure that the Met Office has all of these trends computerised and graphed. If they had proven the contention then those graphs would have been released, not just raw data.
Of course, I am perfectly prepared to concede that Oxford is not necessarily typical of all world spots. Neither is any other place.
I encourage readers to pick other cities and do the same as Victor has done. The Met Office data is HERE.
UPDATE Monday 5pm: Watching the Church of Climateology react to this post has been a joy to behold. Clearly the left, stung by the post about Kerry McCarthy seem to think they can get back at me through this post. Well, hey, good luck to them. The insults flying around on Twitter have to be seen to be believed.
I posted this because I thought it was interesting. It was not my work, as I made clear in the opening line. I then at the end encouraged readers to test the data themselves. And yet Will Straw, for it is he, seems to think I have something to apologise for. In his world maybe, but not in mine. I regard the internet as a place for debate - where you can throw something out there and let people debate the rights and wrongs.
My mind is not closed ont he issue of climate change. I like conventional orthodoxies to be challenged, to be tested. The reason I posted this was because I found what Victor had done interesting. I wanted to get a reaction, to test what Victor had done. To see if it stood up to scrutiny or not. Readers can make their own judgements as to whether it has or not.
Most people on the left on Twitter haven't really bothered engaging in an argument - all they have done is hurl insults. That's fine. I expect it from them. They're the internet equivalent of pond life. At least Will Straw attempts to argue the toss without descending into the gutter. Unlike his friends at Liberal Conspiracy who I won't even grace with a link. Until they grow up.
75 comments:
The link is here:
http://www.metoffice.gov.uk/climate/uk/stationdata/
I did a little more work on these sets which are bedevilled by entire years being missing and in many cases the records only go back 40 years or so.
However Bradford shows the hottest June day as being in 1968 and the two years with virtually identical peaks are 1908 and 2008. 1928 was the coldest year and 1988 the warmest.
Canterbury has a lot of records missing but we see the warmest June day in 1938 and the warmest December day in 1988. 2008 was on both counts cooler than 1988.
So far I can conclude that 1988 may have been quite benign weather although 1918 showed the warmest December day at many stations.
The extracts I made were, I stress quite random, dictated only by the necessity to truncate the mass of data. I shall not pursue this as a hobby - I only wished to look at the new data released to see if it indicates any warming trend. If it does I cannot see it although a subjective analysis might reveal one.
Info on it is here (http://www.guardian.co.uk/environment/datablog/2009/dec/08/uk-us-temperature-change-global-met)
No Ian, Oxford is most definitely green and yellow - being ruled by the Greens and LibDems. And a greater collection of idiots you would search for years to find!
That's a rather odd methodology don't you think? Why would you do that rather than calculate the mean temperature over the whole year?
Iain,
For the counter example, just take a look at this website - all in French, I'm afraid, but the pictures of how glaciers have disappeared in the same time scale is quite eloquent. If it's not getting warmer, why is all the ice melting?
http://www.glaciers-climat.fr/Mer_de_glace/La_Mer_de_glace.html
I don't claim to know, but I don't think that temperature records for Oxford at 10 year intervals analysed by an amateur are necessarily very meaningful.
Adrian,
No one denys that climate change is happening. It always has done.
As far as I can see Victor has calculated the figures very well. I have now been on the site to check myself. Not sure how a "professional" could have done it differently to him. But feel free to try!
Adrian,
There can be other reasons for melting ice. Increase in pollution would make the surfaces of the glaciers darker and therefore absorb more heat thereby melting faster.
"As far as I can see Victor has calculated the figures very well. I have now been on the site to check myself. Not sure how a "professional" could have done it differently to him"
Someone wanting to carry out a meaningful analysis would look at the mean temperatures. It is that which counts. No, I don't know what that reveals but I would certainly want that done before trying to draw any conclusions, and before using a political blog to propagate the idea that man-made global warming is a myth and that those at Copenhagen were all wasting their time.
Adrian @ 9.38pm the reason for the retreat of the glaciers in the Alps is the end of the 'little ice age'(Après la fin du Petit âge glaciaire, vers 1900) which is mentioned in the section on the Bionassay Glacier.
Regarding temperatures in Oxford or anywhere else you need to know where the temperatures were actually taken over the period and if there are any factors that would cause warming such as urbanisation. This is the whole problem with the IPCC. They have made the temperatures fit the theory by selection.
Assuming the Met data isn't as corrupt as the CRU data, what exactly is the point of using the one location as some kind of defining trait?
Individual local climates alone don't really offer an insight into any global changes.
The UK is especially difficult, given our climate is strongly controlled by the North Atlantic Conveyor, we shouldn't even be regarded as a valid data point without accommodating for that - although many other regions suffer from similar things (forest burning around Indonesia creates particulates that reduce warming there for example.)
It's certainly interesting to see local climate trends, but it neither supports nor opposes climate change on its own.
It seems obvious to me that neither you (Iain) or Victor from Kent have any formal knowledge of statistics. The points you are both making are actually irrelevant for a vast array of mathematical reasons if one is trying to measure a general trend across year.
Iain, as previous comments have pointed out, this is an odd methodology for one location done by an amateur. The statement that 1998 was the hottest year on record is for *global* temperatures, not for Oxford. Your implication is that the Met Office is just wrong because somebody can have a quick perusal through random bits of data to "disprove" a straw man statement. How is that adding meaningfully to the debate?
Te other problem with these datasets is the fact that the instument readings are often not cross-correlated with with other or calibrated to a standard reference.
Now, bear in mind that the current increase in climate (if you accept it) is less than 1 celcius. Now, nip out and look at the nearest thermometer and tell me how accurate a reading you can take.
Now suppose that I use the same thermometer for 40 years and then someone drops it and I replace it with a new one. Does it give the same readings as the old one?
A lot of the data released is of very poor quality in terms of verification and cross referencing (where two thermometers are used together to determine any inbuilt difference) and as such the reliability of the temperature record is highly questionable.
I've been skeptical but I'm now 60/40 in favour of global warming, based on comparison of northern hemisphere data to southern hemisphere.
I haven't regressed on the back of this analysis. I have doubts about the method which has already been raised but mainly on the premise.
Global warming does not mean it will get hotter everywhere. Its not called local warming.
Knowing the met office they probably rounded up their coversions from Fahrenheit to Centigrade, to make it look even warmer.They cannot get tomorrows weather right so how can they say what will happen in 10 years.
Adrian
In 1903, Amundsen led the first expedition to successfully traverse Canada's Northwest Passage. He said he found no ice.
So why was that then.
It comes and goes and no one is sure why.
The critical point is what have they released.
They haven't release the data for the world, because they still claim confidentiality.
They haven't release the raw data, it's manipulated data.
What will people assume? That its the raw.
I don't wish to be a grumping climate change denialist denialist (sic), but really, please, stop with this kind of rubbish. You are none of you statisticians, none of you know anything about handling large and messy datasets. Pulling random crap out of this data proves absolutely nothing and makes people who do know a little about stats and data analysis think that you are clearly on the wrong side. Thanks.
For example, where is Greenwich as a station?
Not on the list.
What a surprise. Remove data that you don't like.
The Met Office has made its bet and its GW.
It won't release the raw data because that blows its bet,
For the counter example, just take a look at this website - all in French, I'm afraid, but the pictures of how glaciers have disappeared in the same time scale is quite eloquent. If it's not getting warmer, why is all the ice melting?
Who's said it is melting?
If snowfall is reduced, the glaciers will retreat, and that doesn't depend on any warming.
I did not say that my survey was exhaustive nor scientific. I explained exactly what I had done.
The Met Office has not published the mean annual or even monthly temperatures over the period which might make a scientific comparison possible.
The reason for choosing June and December should be obvious even to the most ardent worshippers at the Church of Climatology.
I am very skilled at interpreting temperature and humidity trends in microclimates. It was how I earned a living. I concede this does not make me a weather expert but I would have expected to have seen some significant warming trend over a fairly large number of readings. In fact I looked through all of those records, year by year, month by month. You may do the same and see if you can interpret the raw data in any other way.
The locations do not contain any description of location. I do know that many of those locations have been in use for a great many years.
And lastly, for Adrian, I did not "interpret" them - I stated them. You can verify them for yourself.
I have to say, that wasn't the clearest explanation I've ever read. However, I think I get the gist of it.
I'm not going to accuse Victor of making the data up, and I'm sure we both acknowledge that the data isn't anywhere near thorough enough to draw any great conclusions. However, I would question the use of individual figures. That always skews data. Individual figures can be used to illustrate points, but show little in themselves. Trends carry more weight. Mean figures are better, and I'd say it would be better to analyse those. I've done that very quickly (Click) and it shows different results. It also appears to suggest the 1950s were cooler, interestingly. It doesn't prove much, but it does show the 1900s were a fair bit cooler than the 2000s.
There are problems with the quantity of data available to use though.
"If it's not getting warmer, why is all the ice melting?"
Because that's what ice does
I see the Russians are now quoted as complaining that the data on Siberia has been used selectively in the various models in that they have only taken up data from city areas. As those have industrialised over the last 80 years guess what? The temperature has gone up. Cities are always hotter anyway. But the figures for the vast expanses of Sibera show no sign of warming....and they were left out
Victor: "I am very skilled at interpreting temperature and humidity trends in microclimates. It was how I earned a living. I concede this does not make me a weather expert but I would have expected to have seen some significant warming trend over a fairly large number of readings."
How could you possibly expect that? This is just statistical nonsense. Leaving aside whether they are wrong or not, the IPCC have estimated that because of man-made carbon dioxide deposition global temperatures have risen since pre-industrial times by 0.7C. To take maximum temperatures for two months of each year is pointless since it is mean temperatures that are important. Even if you had presented mean temperatures, when dealing with a global rise of only 0.7C random local variation will overwhelm small changes of that kind taken from one location only (Oxford). You need a global mean of mean temperatures to see a trend.
The problem is the rises that are expected over the next 50 years to be significantly greater than 0.7C. The IPCC estimate with current carbon pledges from Copenhagen they will rise by about 3C. Again, I am not asking you to either accept or deny that (the IPCC might be wrong), but your sampling method as a means of rubbishing global warning is nonsense.
Withering Vine:
You're rather missing the point. I don't think anybody is suggesting that a random sampling of a few months in one place can disprove global warming. That is presumably why we are now encouraged to pick other weather stations.
It does however call into question the weird modern paganism that seems to predominate these days, evident in the popular assertion that global warming must be happening because it's "all around us"; summers are obviously hotter now, winters were obviously colder than in years gone by; weather is more extreme... constant floods... polar bears... ice on Kilimanjaro... Hurricane Katrina... All nonsense because the possible variations are much greater than the slight trend, perceptible only on a global scale.
I chose Ringway on account of the fact that it's ten minutes' walk from my house. I've plotted a graph (it includes all the data not only two months'): http://img686.yfrog.com/img686/1529/ringway.jpg Seems the other stations are the same.
The problem is, the evidence for global warming is only really apparent from a detailed statistical analysis. As the Oxford example shows, it's not intuitive. We rely on the honesty of statisticians, particularly those at CRU and others implicated in Climategate, who were largely responsible for piecing together the long term records that form part of the "consensus" view that emerged from the IPCC.
If you'll forgive a clumsy metaphor, it's like a rail journey from Exeter to London. I know the train doesn't have to be constantly travelling East but if I don't see the sun through my right hand window for most of my journey, I start to wonder if I'm on the wrong train. There need to be actual weather stations (a lot of them) with actual warming for an average to appear. Now it seems likely there's no evidence of an actual warming trend in the UK. We also know that the hottest year in North America is officially 1934. It's all very well to say we shouldn't read too much into individual results. But the fact that so much of the world shows no trend, together with doubts about the honesty of the scientists, surely raises some questions.
No one has yet focused on the suppressed technologies that could get rid of carbon, are cheaper and cleaner.
The $200 trillion fossil fuel industry is too powerful to tolerate innovation. And the carbon credit industry puts up costs even higher...effectively a fossil fuel energy tax, which means even more incentive to block innovation.
I've assembled the key technologies in six video presentations from youtube.
HERE. It's fascinating stuff. And the potential beyond belief.
Lets hope the Copenhagan fisaco is the turning point and some reality starts to come in.
Brown wants to rush things just in case by some tiny chance he loses the election.
Do you note that out of the blue he is promising all sorts of money gifts for years to come. he knows the Tories will cancell it, so he has set things up so he can attack the Tories for it.
How people continue to support Labour is very hard to understand.
Good discussion on temperature measurements in Darwin here...
http://wattsupwiththat.com/2009/12/20/darwin-zero-before-and-after/#more-14358
Isnt it hilarious how the excuses for manipulating the raw data to show an artificial rising trend arrive just in time?
Its OK to manipulate the raw because..er..you are not experts...you know nothing about statistics...er..shut up and leave it to the experts.
The whole problem of the historic temperature series has come to light because the raw data says one thing and the 'adjusted' data says something else. The act of manipulating the raw data can lead to the temptation to dress up the figures to show a desired trend, its called the 'I want to find an effect so I do mentality'. if its possible to introduce bias, if the conditions are suitable then its reasonable to suggest that tampering will take place.
There is the raw data and there is the method of adjusting that data, if the method introduces a recurring rising trend then its reasonable to suggest that the method may be wrong.
I reproduce a letter from Dr David Bellamy printed in todays D.T.
We are not alone.
SIR – The only good news to come out of Copenhagen is that, in the words of Greenpeace: "There are no targets for carbon cuts and no agreement on a legally binding treaty."
Hooray! Along with tens of thousands of global-warming sceptics, the world can now breathe a sigh of relief and return to the sanity of real science, which counsels that carbon dioxide is not a poison, let alone likely to cause a heat-driven Armageddon.
We can now burn non-sulphurous coal again to ameliorate the effects of the colder climate that has already been with us for the past decade and is likely to stay for the next 30 years.
Dr David Bellamy
Bedburn, Co Durham
@ Calum
Massive and sweeping generalisation. Who is this 'you' as in "you are none of you"? How do you have the slightest idea of the professions of those who read and/or contribute here?
Like so many, you're clearly letting emotion over-rule intellect. And that has been the problem all along.
I agree there is a legitimate debate about climate change to be had, however I don’t think that is helped by publishing Sunday afternoon musings from “Victor of NW Kent” as if it were scientific evidence.
Whilst I believe you are normally fairly open minded, Iain, on the topic of climate change I think that you portray a worrying trait – just like the hard core True Believers you seek to admonish – of holding up any grains of evidence for your own argument as gospel yet dismissing all evidence to the contrary out of hand.
Yes, as others have noted, this is statistically unsound and it's quite possible that a wider-ranging study would come to a different conclusion. But it is an interesting sanity check.
We're continously being told that we're now feeling the effects of global warming and that we're all going to burn - a strangely Catholic turn of phrase there - and yet that's not what's happening. Temperatures at weather stations aren't actually matching the dogma about the Warmest Decade and how hot recent years have been, they have the temerity to show that years in the past have been just as hot or even hotter.
When we're left with something that can only be detected by "advanced" statistical methods - misused according to Prof Edward Wegman's report - and one must therefore question how real it is. A trend that appears only in gridded global reconstructions but not in raw station data is curious to say the least.
Why does the god only reveal himself to the priesthood?
The stupidness of chosing to look at urban areas cannot be overstated. The heat resulting from, and retained by, urban areas will drastically alter the figures.
But then I'm guessing you knew that, Iain.
Hooray for David Belamy! Hip hip hooray!
plotting the same for Heathrow over the period 1948 - 2008 shows a 1.6 degreee increase in both maximum and minimum temperature in June, with a shallow trend, but a fall in December maximum and minimum temperatures of closer to 2 degrees.
The odd thing is that 1968 had both the second warmest June and by far the coldest December.
I can determine nothing whatsoever from these figures. They certainly don't convince me that LHR has become a warmer place over the last 60 years.
Iain Dale unwittingly starts the campaign for "Prospective MPs to be given compulsory statistical training in case they cause monumental damage by not getting figures".
The dataset - thanks for the link - is quite easy to use. Rather than one month from one region, you can take at least 7 regions that have enjoyed consistent measurings for 120 years, and average their whole years. This is 84 data points per year, not just one like the amiable but misguided Victor has used.
The results are then (a) actually slightly meaningful rather than arbitrarily data-mined in order to make a tabloid-level point and (b) clearly showing some sort of rise, after an hour's work. Go to my blog to see.
Iain, I think you and Victor have unintentionally demonstrated the thin data you sceptics rest on in order to believe something you desperately WANT to believe. Keep it coming, it strengthens the AGW side.
Withering Vine, you criticise others for their statistica skills and then make the same mistake yourself.
Your argument that Victor's sample isn't large enough to disprove global warming is true. Unfortunately for you it appears the it is also true the whole of the warmists' data set. The data is so noisy that even using all the available global data the 0.7 degree warming cannot be stated with satisfactory confidence. That is, statistically, the increase does not exist. That is not to say it doesn't exist in reality, just that the liklihood of it existing is similar to the liklihood of it not existing.
All in all, the case for global warming is unravelling and the more the data is probed by genuinely independent people the more bias is exposed, often introduced accidentally by people genuinely misunderstanding statistics (a mistake the best of us can make, e.g. Withering Vine) but also for personal gain, as the CRU 'trick' emails and - more dammingly - their programmers' code demonstrate.
................................. said... :
No. The increasing density of urban surroundings would, if anything, lead to an exaggerated heating. This makes the lack of heating even more surprising, insofar as it is surprising at all. As has been pointed out, this is of course very unrepresentative. That doesn't mean the Met Office's figures are "wrong", and presumably the Met Office wouldn't, despite your belief otherwise, site a weather station in a location that is so heavily affected by human activity as to render any measurements useless.
Will 883: You may not the like the implications but Victor's 'Sunday afternoon musings' ARE science.
As a good scientist he has made his method clear and anybody so inclined can verify his results. The day of the week is irrelevant. As is also the social standing of the person making the analysis.
What matters is not,say, his skin colour or his professional qualifications but the quality of his argument. This is the breakthrough of science and what distinguished modernity from less enlightened times.
To defeat his argument you cannot attack him personally. Neither can you cannot comment on his social standing. This approach demonstrates backwardness and a fundamental misuderstanding of modern enlightenment. Instead you must engage with his logic and defeat him as a fellow scientist. To do this you need no special permission but merely a willingness to let go of your emotional attachment to your own prejudice. Try it.
Scary Biscuits, let's give you some basic stats. This morning is the first time I have seen the data.
Taking 7 places with continuous records from 1900 to 2008, the average all year round temperature has gone up from 8.5 to 9.5.
The annual standard deviation is 0.43. The annual change on average is 1/43rd of this i.e. 0.01. So any single year statements are bound to be swamped by noise, yes.
Better to take 5 year averages. The standard deviation over 5 year averages is 0.29. Assuming lack of correlation across years, we can test the hypothesis of the last few years being just randomly warmer. In brief, using this sort of thinking:
http://en.wikipedia.org/wiki/Standard_error_%28statistics%29
With 20 5 year averages to think about, you get a standard error of about 1.29. So the change we have seen is just less than 1 standard error away. There is therefore a 18% chance, perhaps, that it is explained by the noise, and the rest ought to indicate some sort of trend. In my basic view, given the risks of it being real and the costs if it is, this gives us enough reason to worry.
Otherwise, what a pity we did not have Victor in Copenhagen, shouting "look don't you realise? In June 1868 in Oxford it was really quite warm?" That would have convinced them, for sure. Bet they never thought of that.
Freethinkecon: you make exactly the same mistake as I highlight above for Will. You call Iain Dale 'unwitting' and then unwittingly yourself make a statistical error.
The 10,080 data points your cite, although much more than Victor, are still nowhere near enough to make the conclusion you apparently draw with any reasonable certainty, let alone to extrapolate to the AGW argument, you cite with such clear emotional attachment. That would be true even if the data you used contained only natural variation, was noiseless and free of any bias, which of course it isn't.
Anyone displaying Mr Dale's capacity for deciphering scientific papers should keep a low profile rather than advertise their ignorance as if it is suddenly a virtue.
There is little point refuting the ridiculously stupid information in the main post. the internet is a useful reservoir for this kind of stuff. At least it will never get within 100 miles of a serious journal.
I am amused and flattered by the interest that my "idle musings" have generated. I am also amused by the arguments of the Church of Climatology that I must believe it because they say so. Have we heard that before somewhere, some time?
I hear from Malcolm Turnbull that old people [I am one] are more sceptical than the young on this topic. You should expect that Mr. Turnbull - we have lived long enough to have seen through the constant stream of propaganda which every government puts out.
I have seen a great many of those hot summers and cold winters. Last summer was not hot. Last winter was not cold. This one is.
I have to believe that which I can see and touch and feel. What temperature trends might exist in Patagonia or Japan are not in my personal experience.
Glaciers have been receding since I was at school and have been certainly over the past 250 years that we can measure.
I have witnessed the desertification of the margins of the Great Karoo - it is due to over grazing by sheep, not by global warming.
Just some more musings.
A graph for Heathrow temperatures:
http://img85.yfrog.com/img85/2370/heathrow.jpg
Here there is a slight increase year on year, though well withing a standard deviation: i.e. statistically of little significance.
Scary Biscuit
The standard deviation of any one month in Oxford is about 2 degrees.
The standard deviation for 7 weather stations and taking 5 year averages is about 0.29.
Who knows how much worse it is if you just take one year in 10.
The statistical statements made of the basis of the two methods should be given credibility according to such comparisons of natural variability. A view made by Iain or Victor without even mentioning the size of the variability gives me a fair clue of how seriously to take the claims justified therefrom.
I am not saying anything can prove 100%. I think my 1 hour glance could only prove 80%. But I think Victor's taking of 15 data points can prove about 1%.
I am showing the negative of a negative. Given the variability, Victor's position is approximately irrelevant. Given the variability when you take wider samples, the conclusions are not definitive - but they are many times more relevant.
Compare
Scary Biscuits, let's give you some basic stats. This morning is the first time I have seen the data.
Taking 7 places with continuous records from 1900 to 2008, the average all year round temperature has gone up from 8.5 to 9.5.
But the data is manipulated. It is not raw data. You can't take manipluated data, do some analysis and say, there is a trend in temperature, because you can't tell the difference between a temperature trend and a manipulation trend.
It all goes back to what is called homogeneous data. ie. That the data is produced using a consistent measurement approach.
So.
Observations at the same time of day
Same method for calculating the days figure
Same instrumentation
Same location
Same environment
etc.
If they are all the same, then you can assume the that the series is homogeneous.
If not you have more than time series in sequence.
The problem with the CRU is that they have stitched together the time series in ways that are completely unacceptable. eg standard environmental lapse rate for changes in altitude and not measured lapse rate.
That means they have ended up with large adjustments to current temperatures, when the instrumentation will be accurate. Old instrumentation less so.
Most of the adjustments I have seen have been increases, but urban heat islands should be negative changes.
The have selectively choosen sites and ignored others. Which have they ignored? The ones that cool. The world is warming so any site that shows a cooling must be suspect being the logic.
It's bad bad science.
I once tossed a coin and it came up heads 9 times out of ten.
From this I conclude that the coin had more heads than tails.
@freethinkecon:
"The standard deviation of any one month in Oxford is about 2 degrees. The standard deviation for 7 weather stations and taking 5 year averages is about 0.29."
That assumes that you're repeatedly measuring the same thing. You're not.
It's a cute modelling assumption to imagine that there is some standard global temperature and that each station is recording some randomly-skewed version of it... but that's far from the only possible model.
Here's Victor's mistake:
I thought it fair, since we are talking about proving global warming, to note the highest recorded temperature in June and in December
The highest (and lowest, for that matter) temperatures in any given month will be prone to a lot more variation than the mean or median temperatures for that month.
The correct way to do the calculation is to generate mean temperatures for each month (or week) and graph these. This will give you a fairly noisy looking graph, so you can superimpose on this a moving average of 12 months or 53 weeks.
If you do this, you will see a clear trend.
Oops - in my previous post I'd assumed that the data posted was daily temnperature, not max and min on a monthly basis....
So, the thing to do is to take, for each month, the average of tmax and tmin, and graph that, along with the the 12 month moving average.
The good news is that it's a far faster calculation, and it still shows the same upward trend.
Steve:
See my two graphs which do exactly that.
The problem is this data can't be trusted. Urban areas are far warmer now than they were in the past due to the increase in housing vehicles and heating systems.
None of this data should be used in trying to calculate any increase in temperature.
A poster stated
"...Individual local climates alone don't really offer an insight into any global changes...."
Quite right. So please explain why the BBC insists that every local even is down to man made climate change?
Notice how quiet Roger Harrabin has been recently since the snow and cold? Can you imagine if we'd had a really warm period over the last few days, Harrabin and the rest of the leftie BBC would be stating it was "climate change"
Hmm. So amateurs shouldn't comment on climate change data. So what qualifications does the BBC's "environment analyst" have in science then?
Would anyone like to tell me what scientific qualifications Mr Roger Harrabin has? To the best of my knowledge he has a useless degree in English not normally deemed of much use to study the Chemistry or Physics of climate change.
Steve,
The graph you suggest:
http://img69.yfrog.com/img69/4189/oxford.jpg
Max/min 12 month averages for Oxford.
You speak of the same upward trend but it's very slight. In fact the variation within each decade is far greater than the rise. In other words, all that can really be said is that a given year is more likely to be a hot year today than (say) fifty years ago.
@Quodcumque solveris: so they do. It's a bit hard to see a trend on that scale, though: would it invalidate the data to subtract (say) the average for the period 1900-1950 and whack the Y axis magnification up a bit?
Here you are Steve:
http://img684.yfrog.com/img684/6100/oxford2.jpg
This shows deviation in tenths of a degree from a 1950s baseline.
Can we get a trend line on that then?
r^2 nearest 1 obviously.
Linear regression might be better than nothing.
1. Estimate average monthly temperature as mean of specified high and low
2. Compute 12-month running average
3. Linear regression (line of best fit) to the running average
For heathrow data:
http://img706.imageshack.us/img706/8966/heathrowdalegraph.jpg
Estimated temperature change is +0.029 +/- 0.001 deg C / year (since 1948).
Hence, an approximate 1.74 +/- 0.06 deg C rise in last sixty years.
For Oxford data:
http://img689.imageshack.us/img689/1605/oxforddalegraph.jpg
Estimated temperature change is +0.0075 +/- 0.0004 deg C / year (since 1853).
This would imply a lower temperature rise of 0.45 +/- 0.02 deg in the last sixty years; however, the oxford data considers temperatures back to 1853 (whereas Heathrow is since 1948) and our linear regression assumes a constant linear rise over the entire ~150 year period; if warming is a more recent trend then this, then the best fit will be adversely affected by this early data.
Comparing, instead, Oxford temperatures since 1948 (so, comparable time period to Heathrow data):
http://img706.imageshack.us/img706/4211/oxfordsince1948.jpg
Shows a temperature change of + 0.020 +/- 0.001 deg C / year, which implies an estimated average temperature increase of +1.20 +/- 0.06 deg over the last sixty years....
Why should averaging, then computing another running average, and then least squares be a good method.
Statistics on statistics on statistics isn't a good idea.
So long as you use a whole number of years, you can just go direct to the least squares, so long as the data is homogeneous.
This statistics on statistics on more statistics, with adjustments at every stage, into a model is exactly the sort of thing that the CRU has been up to with disasterous results.
@Nick:
Using the monthly (high+low)/2 and doing a least-squares fit gives an R^2 of 0.0000000244.
Or, in other words, the trend's predictive power is less than one part in forty million.
@Matthew.a.brown an interesting approach however you make one crucial assumption. You say If warming is a more recent trend... but the climate gurus all claim that warming was noticed back in the 1850s!!!
One very interesting thing about the blue raw data on those graphs is that they suggest we have the best of both worlds. Summers are not getting any hotter but the slight upward trend in some stations is caused by milder winters. This explains the original observations. Does anyone know if this is observed elsewhere?
I tend to think a trend line for the mean data is a bit dodgy. R^2 on the raw data shows a gentle rise from, almost precisely, 10 to 11 (if I've done this right - and I may not have done - which is why I'm reluctant to post the graph). Anyway the standard deviation (however you calculate it) will be in that sort of region so it's not a very significant rise in comparison with the year on year fluctuation, is it?
Iain, I'm not sure that this statement...
"The reason I posted this was because I found what Victor had done interesting. I wanted to get a reaction, to test what Victor had done. To see if it stood up to scrutiny or not."
... can be reconciled with your comment...
"As far as I can see Victor has calculated the figures very well. I have now been on the site to check myself. Not sure how a "professional" could have done it differently to him."
Your update gives a far more open impression than your comment.
Norfolk Blogger said
"I once tossed a coin and it came up heads 9 times out of ten.
From this I conclude that the coin had more heads than tails."
--------------------------------
Well only you would come up with that, but then you are a Lib/Dem
The King of Wrong:
Your value of 1 in forty million is probably meaningless. R^2 doesn't recognise the fact that every twelfth data point has a much stronger correlation: it's useless even for wave patterns with obvious trends. It's also pointless to try R^2 on the 12 month averages for each month which are not independent data points. I plotted a rather crude graph of fifty eight data points, one for the average of each year. The result was a fairly strong upward trend R^2 = .4142, from which I'm tempted to conclude statistics can be made to show whatever you want.
The King of Wrong:
Your value of 1 in forty million is probably meaningless. R^2 doesn't recognise the fact that every twelfth data point has a much stronger correlation: it's useless even for wave patterns with obvious trends. It's also pointless to try R^2 on the 12 month averages for each month which are not independent data points. I plotted a rather crude graph of fifty eight data points, one for the average of each year. The result was a fairly strong upward trend R^2 = .4142, from which I'm tempted to conclude statistics can be made to show whatever you want.
@quodcumque solveris
"Anyway the standard deviation (however you calculate it) will be in that sort of region so it's not a very significant rise in comparison with the year on year fluctuation, is it?"
Presumably Global Warming implies that the total (average?) thermal energy (density?) of the atmosphere is increasing... The relative sizes of the underlying trend, and the transient fluctuations on top of it, may not be relevant, as long as those fluctuations are short enough ('sufficiently transient'). Because, on average, the fluctuations will cancel out and it is only the average bulk change that matters (whatever that is, plus, minus, zero).
@quodcumque solveris:
Oh, indeed. Given the same-month correlation it's probably far more reasonable to just pick the same one from each year (as Victor was doing originally).
Averaging across the year is interesting and produces nice values in the middle, but makes the assumption that January temperatures are somehow related to July ones. I'm not convinced on that one, personally...
In the end, this data release is completely useless. Assuming that the temperature changes every 10 minutes, we're getting only the most extreme two samples from 4400 or so in a month - 99.95% of the information has already been lost and we're just sampling the noise.
What part of 'global' and 'average' do people not understand?
The importance of 'global average temperature' is that it is dependent on the radiation balance of the earth, on which CO2 has a major effect.
Why not find out something about the science, instead of engaging in these silly games and thinking they mean something?
eveningperson:
What is your point? We all know what "global average temperature" means. The point of these "silly games" is firstly to dispel some of the less scientific myths put about by fanatics outside realms of peer-reviewed science, particularly the tendency to see extreme weather and the results of global warming everywhere. Secondly, if a large number of weather stations do not show any warming, and I don't claim this is the case, but if they do, it helps us understand the scale of "correction" that has been performed on the data. This data isn't so complicated that it requires any specialist understanding, indeed it has been manipulated by people at Hadley without any statistical background.
What an excellent discussion, so much more civil and mature than I have seen anywhere on this subject and populated by some folks who know something. Says a lot about the quality of Iain's readership.
My own view is that there is enough in the data to cause concern, along with other evidence such as ice melting, changes in animal and plant behaviors etc.
The complementary work, as eveningperson said, is to understand the physics. I am convinced of the ability of the C-O bond to absorb IR radiation and re-radiate it so that it is dispersed and that this phenomenon occurs at very small concentrations on CO2. The question then is whether any balancing mechanisms occur.
I think the possibility that Oxford has not warmed in line with the “global average” is hugely significant. Firstly, an exception to the rule needs an expalnation and if scientists can't explain it then they do not yet understand weather and climate. Secondly, the GCM predictive models are supposed to break the world down into grids. Is the grid containing Oxford showing static temperature from 1850? Or forward to 2100? A good test for verification of the GCM.
Then of course there is the question of climate debt. If Oxford has not contributed to GW and is now to be prevented from developing, how much money are they owed by developed nations to compensate for the lost opportunity – and who is to get this money, the mayor?
Oxford is merely a microcosm of the wider debate and if you cannot answer the questions raised, what chance have you to answer the global questions?
And there is a related aspect.
If GW turns out to be a load of horse manure, where do I get my money back?
Will the greens pay me back the money?
Will the politicians dip into their own pockets to refund me?
Why can't I offset any CO2 and not pay the tax? Offsetting a ton cost less than a fiver, but I'm billed vast amounts if I fly to New York, far in excess of the offseting cost.
Unless the government answers those questions, its taxation for taxes sake, not CO2
Nick
Post a Comment