Ever wonder why NASA’s Jim Hansen (and many others) see red at high northern latitudes?
Above 2011 Temperature Anomaly. Source: NASA GISS interactive plotter
With all that red up north, you’d think Jimbo, Gore, and Trenberth would want to get a look at that firsthand, instead of making a fossil fueled boat trip to Antarctica during peak of the southern summer melt season so they could give us grand proclamations about the melting there.
All the “hot action” is up north according the the latitude plot that accompanies the GISS anomaly map:
Funny how in the anomaly map above, with the great Texas Heat Wave this year, Texas is not red. WUWT? (The way it was portrayed in media, you’d think it was a permanent condition).
It seems to be all in the adjustments. Cooling the past helps the slope of the trend:
How GISS Has Totally Corrupted Reykjavik’s Temperatures
Guest post By Paul Homewood
GISS Surface Temperature Analysis
Now that GHCN have created a false warming trend in Iceland and Greenland , and GISS have amended every single temperature record on their database for Reykjavik going back to 1901 (except for 2010 and 2011), we should have a look at the overall effect.
The red line reflects the actual temperature records provided by the Iceland Met Office and shows quite clearly a period around 1940, followed by another 20 years later, which were much warmer than the 1970’s. GISS, as the blue line shows, have magically made this warm period disappear, by reducing the real temperatures by up to nearly 2 degrees.
Meanwhile the Iceland Met Office say that “The GHCN “corrections” are grossly in error in the case of Reykjavik”.
=================================================================
Just for completeness, here is the GISS trend map and latitude plot for the start of the GISS baseline (1951) to 2011.
UPDATE: 1/26/2012 10:30AM
I added (The way it was portrayed in media, you’d think it was a permanent condition) to the body of this post. since my intent with that statement about Texas wasn’t clear. I got distracted by phone calls and other business in the middle of writing this post and lost my train of thought (and I haven’t been following comments on it either). It is one of the pitfalls of trying to run a business and family while trying to keep up with the demands of this venue. Apologies to anyone who thought I was suggesting Texas summer temp data would show up in December data. Such transient events are just one more indication of the synoptic scale blocking high which caused that event, not any long term climate issue.
Paul Homewood sends his email correspondence and supporting data from the Icelandic Met Office. Here is a PDF file containing the data (referenced in the emails): Reykjavik-1871_Akureyri-1881_Stykkisholmur-1845
From: Trausti Jónsson
To: paul homewood
Cc: Halldór Björnsson
Sent: Monday, 23 January 2012, 17:40
Subject: Re: monthly temperatures
We have sent a questions to the GHCN database regarding this and they will look into the problem. Regarding your questions:
No we were not aware of this.
No, but we are asking for the reasons
GHCN are both valid and of the correct value? If so, why?
The GHCN “corrections” are grossly in error in the case of Reykjavik but not quite as bad for the other stations. But we will have a better look. We do not accept these “corrections”.
No.
No changes have been made in the Stykkisholmur series since about 1970, the Reykjavík and Akureyri series that I sent you have been slightly adjusted for major relocations and changes in observing hours. Because of the observing hour changes, values that where published before 1924 in Reykjavík and before 1928 in Akureyri are not compatible with the later calculation practices. For other stations in Iceland values published before 1956 are incompatible with later values except at stations that observed 8 times per day (but the differences are usually small). The linked paper outlines these problems (in English):
Click to access Climatological1960.pdf
The monthly publication Vedrattan 1924 to 1997 (in Icelandic) is available at:
http://timarit.is/view_page_init.jsp?pubId=278&lang=is&navsel=666
and earlier data (in Icelandic and Danish – with a summary in French) at:
http://timarit.is/view_page_init.jsp?pubId=240&lang=is&navsel=666
http://timarit.is/view_page_init.jsp?pubId=241&lang=is&navsel=666
Monthly data from all stations from 1961 onwards :
http://www.vedur.is/Medaltalstoflur-txt/Manadargildi.html
Best wishes,
Trausti J.
Til: “Trausti Jónsson”
Sent: Mánudagur, 23. Janúar, 2012 17:09:30
Efni: Re: monthly temperatures
From: Trausti Jónsson
To: phomewooduk
Cc: Guðrún Þórunn Gísladóttir
Sent: Tuesday, 17 January 2012, 11:19
Subject: monthly temperatures
I attach a table including the monthly temperature averages for Reykjavik (1871), Akureyri (1881) and Stykkisholmur (1845).
Best wishes,
Trausti J.
Lýsing: Could you please send me, or let me know where I can access, annual mean temperatures for Reykjavik and Akureyri, back to 1900,(or when records are available from).. Many thanks Paul Homewood –
Paul Homewood says January 27, 2012 at 3:50 am
“I have just received an email from Trausti Jonsson at the Iceland Met. He makes this very telling personal observation…”
Paul, Trausti and others – thanks for presenting the local data and historical evidence. I find the graph you link to in your 3:50am comment very worrying. Given Trausti’s comments the Quality Adjusted Data is more akin to corrupted data. I am however open to explanations how this QAC data is “better” because I cannot fathom it myself.
ChrisM says:
January 27, 2012 at 1:04 am
Agreeing with Caleb, I thought Nome had record ice that prevent the fuel oil arriving but it is shown as having record highs. Not a missing “-” sign again is it?
Luther implies in his reply to Caleb that there isn’t a link between ice formation and temperature. That may need further explanation.
The other interesting point is GISS baseline warming map shows no warming for the Antarcic Peninsula. What happened to Larsen B?
_________________________________
Hello Chris,
I apologize for the confusion and meant no such implication. Temperature obviously plays a part in the annual ice/melt cycle.
Since Caleb was referring to the early onset of icing which caused problems for Alaskans, I shouldn’t have confused the issue by referring to total Arctic ice loss.
Why is the ‘adjusted’ data not showing the colder Bering Straits?
It appears that comments and hard data links in this thread have uncovered the reason.
If past is prologue, perhaps we should stand by for more Antarctic data modeling.
Luther
Thank you for your reply.
I agree that the thread and links have shown something ishorribly wrong with the anomaly calculations. It isn’t something that “modelling” would fix. I suspect that the code doing the calculations has had numerous rejigs and “fixes” – the type of stuff Harry complained about. Another bodge isn’t what is needed. It calls for a good old fashioned data audit. You know the type of stuff retired mathmeticians up in Canada are really good at.
Bill Illis says: January 27, 2012 at 4:41 am
‘The point is that all the negative adjustments are done on the earlier periods and all the positive adjustments are done to the recent periods.
They do not balance out to Zero over “time”. The histograms shown above do not show time in them.’
The histograms show the effect on trend. If what you say is true, the trend changes would be positive. The histogram shows that trend changes are equally likely to be negative.
DaveS says: January 27, 2012 at 5:20 am
“It was also shown that the size distribution of the Darwin adjustments was a red herring. The issue was the order in which the adjustments were made, which, very conveniently for the alarmists, increased the warming trend.”
Likewise. Yes, the adjustments increased the trend for Darwin. The histogram shows that this was unusually large (among GHCN stations), and that large negative changes were about equally likely.
Paul Matthews says: January 27, 2012 at 5:58 am
“Looking at the unadjusted data you can see the sharp cooling around 1965 in all 8 of the data sets, so there can be no doubt that it is real.”
Exactly. The reference point quoted is GHCN’s currently published data. Yet GHCN is accused of altering history.
It does seem likely that this was a false positive.
Agust Bjarnason says:
January 27, 2012 at 6:26 am
Agust:
We both know that Iceland has the highest literacy rate in the world. What you don’t know about me is I am of Icelandic heritage and a US citizen. We both know that the warm period in Iceland prior to the present one was about equal in intensity, just as we both know there was a really cold period in the very recent past. My knowledge prior to this article was based on emperical evidence. At least that is what my cousins tell me.
I consider us lucky. It is quite apparant than an error in the methodology has presented itself concerning the revisions of temperature. I am 100% confident that these are done by computer, as I can’t see someone pouring line by line over the temp data.
The obvious answer to the delima presented is for GHCN to examine their programs and potentially find the raw data from the world and start over.
I am sure the Icelandic Met has adjusted errors as required. I take pride in doing good work, having a semblence of knowledge etc. I can’t see that ethic having changed dramatically in a few years in Iceland….can you?
I am glad that this error has been uncovered. There is nothing wrong with robust examination of “revisions”. And it is readily evident that this revision needs examing.
I do not believe that there is malfeseance happening at GISS. They are doing the best they can with what they have. I do believe that honest men make honest mistakes.
As far as relying on BEST data, has it even made it to publication yet? Are folks relying on a pre-published report of temperatures that has not stood the robust examination required of methodogy etc?
Camburn says: January 27, 2012 at 2:08 pm
“The obvious answer to the delima presented is for GHCN to examine their programs and potentially find the raw data from the world and start over.
I am sure the Icelandic Met has adjusted errors as required. I take pride in doing good work, having a semblence of knowledge etc. I can’t see that ethic having changed dramatically in a few years in Iceland….can you?
I am glad that this error has been uncovered.”
This gets very frustrating. But again:
“find the raw data from the world and start over”
They don’t need to find the raw data. They have it, and currently publish it in their qcu file. It hasn’t changed.
“I am sure the Icelandic Met has adjusted errors as required.”
So am I. This isn’t about “adjusting errors”. If GHCN thought there were errors, they would have removed the data from their qcu file. What they are doing is homogenizing, for the purposes of a global index.
“I am glad that this error has been uncovered.”
An error has not been uncovered. All it shows is that homogenized data is different from unhomogenized. In the process of homogenization, some true causes of change that are unrepresentative of the region are detected, and there will be some false positives. In a global index, there is a lot of averaging, and what matters is that bias is reduced. It’s the statistics thta matter.
@Nick Stokes – All it shows is that homogenized data is different from unhomogenized. In the process of homogenization, some true causes of change that are unrepresentative of the region are detected,
Seven out of eight Icelandic stations all show similar adjustments, so clearly the changes are not representative of the region.
e.g.
ftp://ftp.ncdc.noaa.gov/pub/data/ghcn/v3/products/stnplots/6/62004018000.gif
Same in Greenland
ftp://ftp.ncdc.noaa.gov/pub/data/ghcn/v3/products/stnplots/4/43104250000.gif
Nick Stokes says:
January 27, 2012 at 2:40 pm
“So am I. This isn’t about “adjusting errors”. If GHCN thought there were errors, they would have removed the data from their qcu file. What they are doing is homogenizing, for the purposes of a global index.”
My point entirely Nick. They are homoginizeing the data. And the homoginization has an error.
I will be the first to admit that I dont know the homoginization process, nor do I have the time to break down this process to try and understand it.
What I want is a reliable temperature metric that bears scrutiny well.
And off topic, but how do you italicize someone elses quote???
It makes it so much easier to follow something the very nice way that you have done this.
Camburn says:
January 27, 2012 at 3:04 pm
“And off topic, but how do you italicize someone elses quote???”
====================================================
Camburn, scroll up towards the top, on the right side of your screen, you’ll find a button. It has the words “Rick Werme’s guide to WUWT” If you click on it, it will take you here…… http://home.comcast.net/~ewerme/wuwt/index.html
Scroll down until you see a section titled Formatting in comments …. while not complete, it gives you some common ways to employ HTML formatting.
Camburn,
Wordpress allows a few html tags. If you write < i > italics < /i > it comes out as italic.
< b >bold < /b > makes bold. But remember to close the tags with </i>
Paul Homewood says: January 27, 2012 at 3:01 pm
“Seven out of eight Icelandic stations all show similar adjustments, so clearly the changes are not representative of the region.”
Paul, what I said was:
In the process of homogenization, some true causes of change that are unrepresentative of the region are detected, and there will be some false positives. In a global index, there is a lot of averaging, and what matters is that bias is reduced. It’s the statistics that matter.
The homogenized (adjusted) file is not the place to look for the true record of individual stations. It is designed to statistically improve the averaging in global index preparation. There are hits and misses, and if the hits reduce bias more than the misses increase it, it is worth doing. It’s likely that in this case, the sudden (real) temp drop did cause a false positive, and that gives a wrong picture of the history of part of Iceland.
But that’s not what they are after – they want a picture of global temperature. So it’s that balance again. If net bias is reduced, they are ahead.
Maybe you think it won’t be. You’d need a statistical argument there, not just one station.
GHCN should definitely not correct this “error”. The algorithm is based on the tendency of the unavoidable false positives to have cancelling effect. If you intervene manually to fix some and not others, the bias will increase, not decrease.
Thank you James and Nick.
Nick Stokes says:
January 27, 2012 at 3:54 pm
GHCN should definitely not correct this “error”. The algorithm is based on the tendency of the unavoidable false positives to have cancelling effect. If you intervene manually to fix some and not others, the bias will increase, not decrease.
Nick:
What is the present bias of the GHCN at this time?
Camburn says: January 27, 2012 at 5:03 pm
“Nick: What is the present bias of the GHCN at this time?”
The issue is the bias that station changes etc might impart to the global index. The answer is, I don’t know. What I do know is that whether you homogenize or not makes little difference. I suspect that there wasn’t much bias, and homogenization didn’t add any. But maybe there was and homogenization created an identical bias.
What I also know is that NOAA has a statistical argument for believing the homogenized data works better for indices. If you’re thinking of using it, you need to read and understand their argument. Otherwise I’d suggest leaving it alone and sticking to their unadjusted data.
Nick Stokes says:
January 27, 2012 at 6:08 pm
“What I also know is that NOAA has a statistical argument for believing the homogenized data works better for indices. If you’re thinking of using it, you need to read and understand their argument. Otherwise I’d suggest leaving it alone and sticking to their unadjusted data.
I wish this had come up earlier as limitations on my time won’t allow the study and confirmation required. I would think this would be great thesis material. To go grid by grid to see what the actual confirmation bias is.
One thing I can and will say is the histograms presented in this thread, which also includes Greenland, does not bode well for the methodology used to homoginize.
For now I will have to go with the adjusted data. Altho, some presentations of temperature pictures based on adjusted data have not accurately portrayed the area of which I live which is also a bit north.
This bears further investigation and I hope someone does the work.
Moderator:
Please pull this as I am trying to learn the art of italics:
Please pull this as I am trying to learn the art of italics:
Did the italics stop now?
[REPLY: Yes they did. We have a test page for just this purpose here. -REP]
Nick Stokes says:
January 27, 2012 at 6:08 pm
The issue is the bias that station changes etc might impart to the global index. The answer is, I don’t know. What I do know is that whether you homogenize or not makes little difference. I suspect that there wasn’t much bias, and homogenization didn’t add any. But maybe there was and homogenization created an identical bias.
When you say “you do know is that whether you homogenize or not makes little difference”, on what bases do you know this?
Can you point me to reference materials that can confirm this?
Smokey:
In your link:
http://zapruder.nl/images/uploads/screenhunter3qk7.gif
Is the anomoly base period identical?
Camburn,
Sorry, it’s not my graph. I found it on WUWT, but I don’t recall the thread.
Camburn says: January 27, 2012 at 6:32 pm
“Can you point me to reference materials that can confirm this?”
Yes. I’ve done it on this thread, here,
here, and
here. There’s even a thread at WUWT Many indices from GHCN, some with adjustment, some without. There are still more at the Blackboard. Zeke has done both with his code.
Nick,
There are hits and misses, and if the hits reduce bias more than the misses increase it, it is worth doing. It’s likely that in this case, the sudden (real) temp drop did cause a false positive, and that gives a wrong picture of the history of part of Iceland.
Sounds a bit like two wrongs making a right to me! I am still curious why the algorithm suddenly created this adjustment, when the previous GHCN V2 did not.
While you might be right from a statisticians’ point of view, my experience tells me that if you find a mistake, you correct it rather than worry if there is a compensating error somewhere else.( I am an Accountant by the way!!)
According to the GHCN Technical Report
The process of removing the impact of non-climatic changes in climate series is called homogenization (e.g station moves, land use changes etc)
If the adjustments to Reykjavik and other Icelandic stations are correct, then they must be justified by comparison with other stations that are not affected in this way. I see no evidence for this.
BTW Nick
I think you said you still had Version 3.0 from last July. Do these adjustments appear then or only in Version 3.1 released in November.
According to Reto Ruedy at GISS, they switched straight from Version 2 to Version 3.1 in December, so I cannot tell from their dataset.
Thanks
Paul
Nick:
I have what some might consider to be a rather simplistic question.
IF there is no difference between unadjusted trends and adjusted trends, why waste time adjusting? Seems like a lot of busy work created with no discernable change.
Why not just use the raw data and then there are no questions/doubt created.
Nick, you are disregarding the fact that our friends in Iceland have already made corrections for location changes…… “‘No changes have been made in the Stykkisholmur series since about 1970, the Reykjavík and Akureyri series that I sent you have been slightly adjusted for major relocations and changes in observing hours.”
As I stated earlier to some others, the people in Iceland are aware of these issues and already have made the proper corrections. Then GHCN corrects the corrections and then GISS will apply theirs….. always, the corrections/homogenization result in decreasing historical temps and increasing the ones near the present.