Can I also take this opportunity to point out the very dodgy choice of scales on his first graph. The temperature data is limited to 1/7th of the height of the graph, while the C02 changes are stretched to full height.
OK. Downvote me then, but if you read the original blog post he's talking about (http://motls.blogspot.com/2010/01/warming-trends-in-england-...) and you look at the '30 year bands' he's chosen they are cherry picked. And he's saying 'Look here in 1661 or whatever it warmed faster than it's warming now'. Seeing fast warming in central English in the past doesn't disprove man-made global warming now.
If you dig into his Mathematica code and take a look at it you'll see a plot he's made (which he doesn't blog about) that looks just like the global warming trend plots everyone is putting out.
Simply because you don't like him or the paper doesn't make them wrong.
Agreed. But the underlying analysis for his article isn't a paper, it's a random blog posting. The guys at the Met Office publish the actual data set and they have put out papers based on them which are linked from their web site. The HadCET dataset is referenced constantly in their other work.
The 30-year thing is significant; the AGW advocates quote trends since 1979 usually, which is significant as during the 1970s scientists were most worried about global cooling and pointed out a cooling trend since the 1950s.
The side with the least-arbitrary data set wins the credibility debate; "since records began" is a good one.
Scientists were not most worried about global cooling in the 1970s. What you had were scientists who published papers positing the idea, and then having their work scooped-up, dumbed-down, by journalists and crammed into into middle-brow publications like Time and Newsweek. That's not a scientific consensus: That's the press stirring the pot.
Considering that the 17th century was just after the start of the little ice age (wiki it), it would be shocking if England wasn't warmer today. Is England warmer today than it was in the 11th century though? Indications are no, it's not. Take from that what you will.
> you look at the '30 year bands' he's chosen they are cherry picked.
Even if it's true, that's the standard that the mainstream climate scientists have set for this debate - look at hundred of tree ring records, pick just those that support the narrative; use thermometer readings up until the 1990s, but when they start to show a temperature decline switch to satellite data and drop the thermometers; use data from weather stations that are surrounded by black-top and waste heat exchangers (building air conditioners), and - when challenged - deal with the GIGO problem by merely AVERAGING the bogus heat increase into surrounding weather stations.
Why are you linking to the Hadley dataset? This is the adjusted(and since "climmategate", discredited) dataset where the original data has been "lost",and they havn't released the source for creating it, you even say so in the comments of your own blog.
"You need to read my next post which is coming soon. I don't think this is the source code actually used to generate CRUTEM3. Looks to me like they wrote it specially for this release and it appears to contain a smallish bug to do with the use of suspect data."
You're right, seeing global warming in the past doesn't disprove man-made global warming. However, it does disprove the claim that the temperature changes we're seeing now are unprecedented.
But that's exactly what makes me skeptical about man-made global warming. This persistent attempt to make it appear more dramatic and less uncertain than it really is. It's exactly what we saw with weapons of mass destruction not long ago.
I'm not saying man-made global warming isn't real, it may well be, but even if it were real there's still the question of costs and benefits. I don't buy the economics as laid out in the Stern review (http://en.wikipedia.org/wiki/Stern_Review). We just don't have the scientific methods to make long term predictions for complex systems.
He included that plot in the article (albeit with a different y-range). It's the red and purple line in "Central England Temperatures & Global CO2 Emissions" at the very top of the article.
(At least, it looks like the same plot. The shape/yrange is the same, but your pic on imgur has no axis labels so I can't be sure.)
Just remember: you only measure what you measure. This data suggests that Central England temperatures have only been slowly rising. It doesn't say anything about average world temperatures.
(The Gulf Stream might be one of the reasons North-West Europe is a relative stable point.)
I'd be interested to see similar plots from the Arctic.
If you hand pick your data set, you can probably show anything you want. I remember reading about the stock market, that if you just shift the starting and end points of "measurements" (for example about your fund's value), you could always claim an x% increase over the last y years.
So this here is an interesting data point - but it really is just that.
I wouldn't be surprised if there were places on earth where temperatures have been falling steadily, or have been rising or falling by extreme amounts.
It seems to be one data set from one location. How can this be "the best data set" out there? I don't think looking at one location is sufficient at all.
If you are going to measure a dataset by its completeness and its length you are going to find it's the best based on that scale. However, if you want to define some other scale which cares less about those factors and more about geographic coverage feel free, but don't try to mix there definition with yours because your "best" will be less compleat or cover a shorter period.
PS: If you feel some other dataset would be a better proxy feel free to argue the point on its own merits. But be careful when you start to redefine the terms someone else is using, it has little to do with their argument and is not useful.
I have nothing against the data set as such - I think it is great if such a complete data set exists. All I am saying is that a data set from one location is not sufficient to calculate the climate of the whole planet. It is one data set among many, which happens to be especially complete.
Let me use your reply as an opportunity to make points I've made in a heated fashion elsewhere on this thread more calmly:
1. It's not clear that CET is a proxy for global temperatures.
2. If CET is a good proxy (and that could be easily checked from the Met Office data) then fast warming in the past needs to be explained to be sure that we are seeing something anomalous now.
3. This Daily Telegraph claims that man-made global warming is refuted by the CET. I don't believe that to be true. It may be questioned by the data, but questioning is within the realms of science.
4. The original blog posting makes the claim "since we've seen 30 years of warming in the past that are faster than the last 30 years of warming there's nothing to worry about". The last part is not correct. If this is correct (modulo #1) then we need to determine what was causing warming in the past compared to today.
If you shift the end point you cannot make any claim about the last y years, because the last y years always end at the beginning of the current year. What he did is to take the entire dataset, not some selected part of it.
And there aren't many places on earth for which long term measurements even exist. So selecting this particular dataset seems plausible enough to me.
On the other hand I would like to hear the counter arguments as well, because if this is such an important dataset, proponents of AGW will have an opinion on it. There may be a good counter argument.
Shifting the end point was an example from the stock market (I think I remember better now: to make your fund look good, you can almost always find a point in the past from where you made earnings relative to today). It was just an example of hand picking data to prove what you want.
I have nothing against this particular data set. I just doubt that it is all that relevant in the bigger picture. It is just another data point that can be fed into models and whatnot.
It may be an example from the stock market but it's not an example for what the author of the article did. He did not select an arbitrary period.
I don't know anything about the relevance of the dataset, so maybe you're right, but I suspect that the dataset is important because no other country I know of has reliable records going back that far.
He did not select an arbitrary period, but an arbitrary place.
For example, recently there was a post on HN about the current cold wave. Turns out while it was very cold in Europe, at the same time it was very warm in the arctic.
It would be interesting to look at the data from central England and see what kind of results one could get by cherry picking individual weather stations.
The place is not arbitrary. It's one of the very few places for which we have good "long term" data. Climate scientists on all sides of this debate rely on datasets like this one because they have no other choice. That may be a statistical problem, but cherry picking it is not.
Your arctic example makes no sense because climate is not the same as weather.
The linked blog lists the top ten tri-decades, ordered by temp rise. Here they are in chronological order:
1687 - 1716, 4.333 °C/century
1688 - 1717, 4.7 °C/century
1689 - 1718, 4.446 °C/century
1690 - 1719, 4.754 °C/century
1691 - 1720, 5.039 °C/century
1692 - 1721, 4.642 °C/century
1694 - 1723, 4.524 °C/century
1977 - 2006, 4.95 °C/century
1978 - 2007, 5.038 °C/century
1979 - 2008, 4.705 °C/century
Can you see a pattern there?
So the first 7 are consecutive apart from skipping one year/tri-decade (1693-1722) and the most recent three are 3 out of the last 4 years. And indeed the 2nd, 3rd and 5th highest rise.
And yet the point he chooses to highlight directly after this list, is that the most recent tri-decade (1980-2009) isn't in the (totally arbitrary cut off point) top ten at all. It is however the 13th biggest rise. In other words 4 of the last 4 tri-decades are in the top 13 and at least 7 of the remaining 9 all are from consecutive periods starting 1687-1694.
Does this really show what he's suggesting it does? I'm not seeing anything particularly useful drawn out by his analysis apart from records from around the turn of the 17th century show a large drop followed by a large rise then another massive, but short lived drop.
There's a lot of back-and-forth in this thread about whether this disproves AGW or not.
Before people get too invested in these arguments, it might make sense to step back for a second or two and ask yourself the question:
"What is the null hypothesis?"
As far as my understanding goes, in order to be doing good science, the null hypothesis must be that there is no phenomenon to explain. So the qustion we should be asking is: "is this evidence consistent with, or does it refute, the null hypothesis?"
As far as I can tell, this evidence is consistent with the null hypothesis. While that is interesting, it's not super-important, so we should all save our passion and debating time for evidence that potentially refutes the null hypothesis. That kind of evidence is much more important to find and pore over...
33 comments
[ 4.0 ms ] story [ 90.0 ms ] threadhttp://motls.blogspot.com/2010/01/warming-trends-in-england-...
What am I missing?
Can I also take this opportunity to point out the very dodgy choice of scales on his first graph. The temperature data is limited to 1/7th of the height of the graph, while the C02 changes are stretched to full height.
Graph here: http://c3headlines.typepad.com/.a/6a010536b58035970c0120a7c8...
The obvious impact of this is to downplay any shift in temperature in response to changes in CO2 levels and any deviations from the linear trend line.
OK. Downvote me then, but if you read the original blog post he's talking about (http://motls.blogspot.com/2010/01/warming-trends-in-england-...) and you look at the '30 year bands' he's chosen they are cherry picked. And he's saying 'Look here in 1661 or whatever it warmed faster than it's warming now'. Seeing fast warming in central English in the past doesn't disprove man-made global warming now.
If you dig into his Mathematica code and take a look at it you'll see a plot he's made (which he doesn't blog about) that looks just like the global warming trend plots everyone is putting out.
See: http://imgur.com/K8Pcn.png
Wonder why that doesn't get a mention?
And could you get a mirror for that post on blogspot for me, I live in China and can't access it, cheers.
Agreed. But the underlying analysis for his article isn't a paper, it's a random blog posting. The guys at the Met Office publish the actual data set and they have put out papers based on them which are linked from their web site. The HadCET dataset is referenced constantly in their other work.
The side with the least-arbitrary data set wins the credibility debate; "since records began" is a good one.
Even if it's true, that's the standard that the mainstream climate scientists have set for this debate - look at hundred of tree ring records, pick just those that support the narrative; use thermometer readings up until the 1990s, but when they start to show a temperature decline switch to satellite data and drop the thermometers; use data from weather stations that are surrounded by black-top and waste heat exchangers (building air conditioners), and - when challenged - deal with the GIGO problem by merely AVERAGING the bogus heat increase into surrounding weather stations.
"You need to read my next post which is coming soon. I don't think this is the source code actually used to generate CRUTEM3. Looks to me like they wrote it specially for this release and it appears to contain a smallish bug to do with the use of suspect data."
His Mathematica code downloads http://hadobs.metoffice.com/hadcet/cetml1659on.dat.
In what sense is "This is the adjusted(and since "climmategate", discredited)" true of HadCET?
But that's exactly what makes me skeptical about man-made global warming. This persistent attempt to make it appear more dramatic and less uncertain than it really is. It's exactly what we saw with weapons of mass destruction not long ago.
I'm not saying man-made global warming isn't real, it may well be, but even if it were real there's still the question of costs and benefits. I don't buy the economics as laid out in the Stern review (http://en.wikipedia.org/wiki/Stern_Review). We just don't have the scientific methods to make long term predictions for complex systems.
(At least, it looks like the same plot. The shape/yrange is the same, but your pic on imgur has no axis labels so I can't be sure.)
(The Gulf Stream might be one of the reasons North-West Europe is a relative stable point.)
I'd be interested to see similar plots from the Arctic.
So this here is an interesting data point - but it really is just that.
I wouldn't be surprised if there were places on earth where temperatures have been falling steadily, or have been rising or falling by extreme amounts.
The only problem with this article is that there is no mention of how the gulfstream could have had a stabilising effect.
The 30 year periods aren't cherry-picked either, they are picked because they were the ones with the greatest increase in temperature.
Lol. If that's not it, would you mind explaining to me exactly what you think cherry-picking is?
PS: If you feel some other dataset would be a better proxy feel free to argue the point on its own merits. But be careful when you start to redefine the terms someone else is using, it has little to do with their argument and is not useful.
1. It's not clear that CET is a proxy for global temperatures.
2. If CET is a good proxy (and that could be easily checked from the Met Office data) then fast warming in the past needs to be explained to be sure that we are seeing something anomalous now.
3. This Daily Telegraph claims that man-made global warming is refuted by the CET. I don't believe that to be true. It may be questioned by the data, but questioning is within the realms of science.
4. The original blog posting makes the claim "since we've seen 30 years of warming in the past that are faster than the last 30 years of warming there's nothing to worry about". The last part is not correct. If this is correct (modulo #1) then we need to determine what was causing warming in the past compared to today.
And there aren't many places on earth for which long term measurements even exist. So selecting this particular dataset seems plausible enough to me.
On the other hand I would like to hear the counter arguments as well, because if this is such an important dataset, proponents of AGW will have an opinion on it. There may be a good counter argument.
I have nothing against this particular data set. I just doubt that it is all that relevant in the bigger picture. It is just another data point that can be fed into models and whatnot.
I don't know anything about the relevance of the dataset, so maybe you're right, but I suspect that the dataset is important because no other country I know of has reliable records going back that far.
For example, recently there was a post on HN about the current cold wave. Turns out while it was very cold in Europe, at the same time it was very warm in the arctic.
It would be interesting to look at the data from central England and see what kind of results one could get by cherry picking individual weather stations.
Your arctic example makes no sense because climate is not the same as weather.
1687 - 1716, 4.333 °C/century
1688 - 1717, 4.7 °C/century
1689 - 1718, 4.446 °C/century
1690 - 1719, 4.754 °C/century
1691 - 1720, 5.039 °C/century
1692 - 1721, 4.642 °C/century
1694 - 1723, 4.524 °C/century
1977 - 2006, 4.95 °C/century
1978 - 2007, 5.038 °C/century
1979 - 2008, 4.705 °C/century
Can you see a pattern there?
So the first 7 are consecutive apart from skipping one year/tri-decade (1693-1722) and the most recent three are 3 out of the last 4 years. And indeed the 2nd, 3rd and 5th highest rise.
And yet the point he chooses to highlight directly after this list, is that the most recent tri-decade (1980-2009) isn't in the (totally arbitrary cut off point) top ten at all. It is however the 13th biggest rise. In other words 4 of the last 4 tri-decades are in the top 13 and at least 7 of the remaining 9 all are from consecutive periods starting 1687-1694.
Does this really show what he's suggesting it does? I'm not seeing anything particularly useful drawn out by his analysis apart from records from around the turn of the 17th century show a large drop followed by a large rise then another massive, but short lived drop.
Before people get too invested in these arguments, it might make sense to step back for a second or two and ask yourself the question:
"What is the null hypothesis?"
As far as my understanding goes, in order to be doing good science, the null hypothesis must be that there is no phenomenon to explain. So the qustion we should be asking is: "is this evidence consistent with, or does it refute, the null hypothesis?"
As far as I can tell, this evidence is consistent with the null hypothesis. While that is interesting, it's not super-important, so we should all save our passion and debating time for evidence that potentially refutes the null hypothesis. That kind of evidence is much more important to find and pore over...
* [edit] fixed a few typos