A year and half ago Steve recalled some encounters with NASA GISS. One could imagine that after all that embarrassment the quality control in GISS temperature products was by now highly improved. One could be also wrong.
In the March 2010 GISS temperature anomaly map Finland appears as a “hot spot” surrounded by cold temperatures:
GISS station values are even more spectacular, the warmest March on record is set in every Finnish station GISS is following. For instance, according to GISS, the mean March temperature in Sodankylä (61402836000) was a remarkable +1.5 °C beating the old record (-2.2 °C) from 1920 by 3.7 °C!
Well, according to the Finnish Meteorological Institute, March 2010 was colder than usual all over Finland, especially in the northern part. For instance, the mean temperature in Sodankylä was -10.3 °C, which is almost three degrees below the base period 1971-2000 average (-7.5 °C). So the GISS March value for Sodankylä is off by amazing 11.8 °C!
Some quality control, please!
h/t: Leone
Climate Audit 
91 Comments
When there are blocks on the grid that are 4 to 9.7 deg above the baseline and they are immediately adjacent to blocks that are -.5 to -1 deg or even -1 to -2 deg relative the baseline, that should be a telltale sign something is probably amiss.
This is a real “WTF?” moment for GISS. One wonders how often this kind of thing happens and doesn’t get noticed.
One also wonders if they tend to go in one direction or are random. This is why these data sets need to be open and the underlying data and code publicly available. If it isn’t, we may never find out how frequent these errors are or what effect they may have on the records.
Re: timetochooseagain (Apr 15 10:22),
The datasets are open. You can see the original CLIMAT form material for March in Finland at OGIMET.
and the standard response:
1. Finland represents less than 1% of the worlds surface, therefore
any mistakes make no difference. this excuse can be made
virtually any time. In fact, there is no requirement to get any
country correct.
2. Whatever warm bias errors there are , they will be cancelled
by the magic of the LLN. Cool biases are sure to exist and perfectly cancel the warm biases
3. Do your own damn science.
4. Oil shill!
The response to that should be “until we have a full blown audit, we won’t know if it’s just Finland, we won’t know if it makes no difference. But we MUST have a complete audit, otherwise there will be nothing but doubt over the reliability of the results.”
Mosh, Yep!
Why can’t “oil shill” be number 1?
Then the other three wouldn’t be needed.
I think they simply added a few degrees. No error or problem with methodology. Just simple fudging of the data.
In cold periods the Finns are noted for their habit of sitting in roasting hot saunas until near death by heat exhaustion then rolling about in the snow until cool and flagellating themselves with twigs. The cold March would have activated more of this so GISS were probably didn’t account for the Sauna Heat Island Temperature effect.
A ex colleague of mine worked in Finland for a couple of years. He told me it was so damn dark and cold all winter that they just drank loads of vodka so they didn’t notice. That’s maybe what GISS did as well.
It isn’t just ‘some quality control’ which is needed – one wonders if anybody at the GISS computers is actually capable of using their common sense.
Does it not ring a bell when there is such a divergence between their shiny model and the actual data by the Finnish Meteorological Institute, which is most certainly not an amateur observer somewhere in the sticks?
This shows nicely that for certain people computer models beat reality.
To be fair, the quality has been quite well controlled, and is now nearly absent from the product. Gavin should be able to extinguish any remaining quality in the coming months.
Excellent!!
What makes anyone think that GISS staff even look at the output maps?
Craig:
You better believe that they would look it over if the output maps were all BLUE!
If they were all blue, it would most likely have caused an adjustment.
It seems to me that there are many terra incognita (grey) on that temperature map, and we are in 2010.
How would it have looked like in 1900 or even 1950 ? And how can “homogeneisation” be done with so little data ?
Re: Benjamin (Apr 15 11:00), The map represents data that has come in by 12 April. The grey will diminish as new data gets posted.
Nick:
They have not yet increased the size of the grids to eliminate the grey!
“And how can “homogeneisation” be done with so little data ?”
You put the blender on “frappe”.
Jim Hansen and his fellow court jesters have recently posted a preprint of their latest paper.
Pages 25-27 on “data flaws” are particularly interesting in the light of Jean’s revelation. It talks of the Y2k error and the Russian 2008 error – with no mention of CA of course. They even get the description of the error wrong (they say the Nov data was a duplication of Oct, but it was Oct/Sept). Can they get anything right?
Then there is this paragraph:
I wonder who these ‘several scientists’ are? Clearly the process is not working, they have not learned any lessons, and they can look forward to further ridicule.
if GISS gets it this wrong in a modern country with a competent national weather service, why would anyone have confidence in all the temperatures that are “estimated or extrapolated” for areas like the artic?
Nothing to see here. Move along.
Could it just be a data handling error? Does anyone have temp datasets from FMI to compare with the GISS data?
-Sale
Interestingly, the error only appears in the “combined” version of the GISS data linked above by Jean.
If you ask for the “raw” version or the “homogenised” version, March is 999.9 = not available.
Roman:
That explains it! Eliminate the first two 9s and you have the GISS temperature 9.9C 😉
Looks like GISS fixed it: They have posted a News bulletin on the GISTEMP homepage:
2010-04-15: The data shown between 4/13 and 4/15 were based on data downloaded on 4/12 and included some station reports from Finland in which the minus sign may have been dropped. NOAA updated GHCN on 4/13 by removing those data and we updated our displays today. The March 2010 global mean temperature was affected by about 2/100 of a degree Celsius, well below the margin of error (about 15/100 of a degree for monthly global means).
Re: Fred Nieuwenhuis (Apr 15 12:11),
Ahh, that explains it, of course, the difference between +1.5 and -10.3 is just a minus sign!
Thanks for the explanatory update, Fred. But let’s notice the obscurantist way the correction was noted. The dropped minus signs were said to affect the “global mean temperature“ by a mere 0.02 C.
They’ve minimized the apparent error by smearing it all over the globe. Nothing is said about how the errors affected the mean Finnish March 2010 temperature. We can suppose that if the error had not been caught, the falsely warm Finnish Spring would have been yet more evidence of AGW.
Actually, if we scale the 0.02 C error by the ratio (Earth surface area)/(Finland surface area), we can approximate the error in the March 2010 Finland temperature. So, that’s 0.02 C*(5.1*10^8/3.38*10^5)=30 C (areas in km^2). Quite an error.
Credit to climateaudit, of course.
Oh wait…must have been Gavin’s “mystery man.”
Hey all, new guy here although I’ve been reading CA for the past couple years.
I have to agree with Sauli that this looks like some sort of handling error, where perhaps data was missing the “minus” character when entered into the map. Not to deride any current/former US Government employees here, but is it really any surprise that a government agency could be staffed with people who really don’t care to get it right? In my personal experience, they tend to outnumber the ones who do in fact make an effort.
It’s not unprecedented for NASA to make such mistakes:
Source: Wikipedia
It is almost certainly due to gridding and the stations used, or not used as the case may be. I can’t see the point of gridding except for the sake of pictures. The process takes precisely measured point data and then averages it in some fashion with other data from farther away, coming up with effectively imaginary values (estimates) for grid cells that are only vaguely related to the original, precise data. The gridded maps then are presented and give an illusion of “completeness” of information that can be illusory.
Duster, some kind of gridding is necessary. Suppose we have ten temperature stations in a small warm area, and one station some distance away in a cooler area. A plain average doesn’t give the correct picture, because the one station will not affect the average much.
I’m in the process of doing a monte carlo study of using cluster analysis rather than gridding as the basis for the averaging. So far I have shown that it performs as well as gridding when the data points are randomly distributed. My next move will be to see how it performs when they are clustered. If I have any success I’ll post about it here.
Make the grids smaller. Don’t over sample a region while under-sampling adjacent regions.
This is simple stuff.
I don’t think these people know the meaning of quality control. Maybe they learned what they know from the auto industry. When you have a flawed methodology in the first place it is very hard to pick up these little glitches. Perhaps it is simply a cost saving measure. Let volunteers from find your problems for you. Then you can pick and choose which ones to accept or reject.
Hey, it’s only settled science! Who cares about those silly and inconsequential divergences?
Even more, it’s Post-Normal Science ™ and so a few degrees on or off should not affect the extended facts. Right?
It’s settled science. “Science” that’s been allowed to settle after being cooked each month.
I don’t know, in the past few days there has really being a heat wave here. All the snow diseappeared under few days last week here in the south.
The data is from March, not April. 😉
Oh, okay.
Then it is weird!
So Uredrra, you mean to say that GISS is now predicting temps?
Awesome!
Whether the correction is/was made to april/march data, the .txt-file still shows 1.5 for March 2010, not -1.5, etc. When to they update the datasets in the .txt files?
This is the one I’m looking at: http://data.giss.nasa.gov/work/gistemp/STATIONS//tmp.614028360003.1.1/station.txt
-Sale
…which brings to mind the question whether the minus sign has been dropped from other stations, in other countries around the world. Is there any way to check? Comparing each station’s records to GISS records?
-Sale
Re: Sauli Särkkä (Apr 15 13:58),
Yes, there is a way to check. The actual CLIMAT form data which GHCN uses can be seen at OGIMET.
The Satellite data suggests that there may be other issues with the GISS data. See http://pielkeclimatesci.files.wordpress.com/2010/04/0310.jpg
Vodoo Science?
I just has to be a simple error. I quickly went over monthly average temps 1951-2010 and compared these two:
http://www.fmi.fi/saa/tilastot_57.html#7 (the lower image pertains to Sodankylä) and
http://data.giss.nasa.gov/work/gistemp/STATIONS//tmp.614028360003.1.1/station.txt
The last line of the GISS dataset is the only one that has an aberrant figure for march 2010.
Looking at that map you can almost see the path of the jet streams, in cooler/nuetral shades.
Finland was the only place where Crimson (+++) was adjacent to a blue tone (–). Would be surprising to find such a huge change in such a small region.
Looks like the current map is much different:
http://data.giss.nasa.gov/cgi-bin/gistemp/do_nmap.py?year_last=2010&month_last=3&sat=4&sst=3&type=anoms&mean_gen=03&year1=2010&year2=2010&base1=1971&base2=2000&radius=250&pol=reg
Maybe Finland found an effective and inexpensive way to warm up the country and just haven’t shared the new technology with the rest of the world.
That could explain it.
Comical. If this stuff wasn’t so important, I would be laughing…
GISS has apparently corrected the Finland anomaly. See:
http://data.giss.nasa.gov/cgi-bin/gistemp/do_nmap.py?year_last=2010&month_last=3&sat=4&sst=1&type=anoms&mean_gen=03&year1=2010&year2=2010&base1=1971&base2=2000&radius=250&pol=reg
Presumably doing so makes March even a little warmer than before?
Nope. The correction means that the measured temperature anomaly for Finland is a shift to colder.
With no h/t to CA or Jean, let alone an indication that a correction has been made, that I can see.
Re: Hu McCulloch (Apr 16 14:21),
GISS simply works automatically off the GHCN data. If GHCN corrects, GISS follows.
I’m intrigued that for March nearly ALL of the inland area of Antarctica was 2 – 4 degrees centigrade
warmer compared to the 1971 – 2000 baseline period.
That nearly doubles the global surface with a 2 – 4 degree anomaly for that month.
Yeah. And aren’t the polar regions the places that “really matter”? Inhabited areas have lots of seasonal temperature variation and long term fluctuation. 2 – 4 degree centigrade increases at the poles would seem to be worthy of attention.
Note: I have no detailed knowledge of any of the scientific work in this field. This is only opinion.
“Some quality control, please!”
Uh, why should they start now – they’ve been doing quite well (in the media and public’s eyes, anyway) without it so far.
I see that they appear to have “corrected” this error as shown by Ken Hatfield’s link.
May have been a glitch in their program that required a manual correction?
Good to see that actual observations are being used to correct possible programming errors, isn’t it?
The chart linked to for Sodankylä in the main post shows mean annnual temperature, not mean March temperature. Following the link to the underlying data, the mean annual temperature for 2009 is 1.05. The data as yet contains no figure for March 2010 but records Feb 2010 as -17.5, and March 2009 as -7.0. I can’t conjecture where you got the figure of 1.5 for March 2010.
Andrew, the reason you can’t see the 1.5 figure is that GISS have deleted it and replaced it by a 999 since Jean pointed out their mistake.
Oh, fair enough. But if the latest data has values of 999 for March, what are they using to generate the new map without a hot Finland?
It seems that the correction for this problem was a simple deletion of the March value. If you want the original .txt-datafile with 1.5 listed for March, drop me a line via email and I’ll send it over. This applies to anyone that wants the datafile that was downloaded yesterday, on April 15th 2010.
You won’t even have to file a FOIA request for it 😉
I live in the middle of Finland and I can assure you that
March was extremly cold.
Hasse
OT
I just wanted to post this idea somewhere visible before the AGW circus kicks in.
Icelandic eruption leads to cooler summer in N. Europe. Sceptics remark yet another year of “absence of warming” since they are paid by big oil conspirators and DIRTY coal.
Mann-made warmists say this does not disprove AGW, it is just a freak event temporarily hiding the undeniable , unprecedented , unrelenting global warming which is due to man burning fossil fuels and we need loads of nuclear NOW.
In fact it proves AGW since more frequent, more extreme events like this volcano are EXACTLY is predicted by global warming computer models.
The marked lack of volcanic activity over the last 20 years was irrelevant. That’s not the same , it’s just a denier myth.
IPCC reports that it is very probably (95% confidence level) “tails we win, heads you loose.”
/OT
Satire.
re. unfounded claims about storm frequency.
Well, go look at http://kenskingdom.wordpress.com/ to see how GISS has “modified” the temp data in Oz according to the site. Lowering 1930’s temps relative to the more recent for example.
Looking at that gridded temp data – the east coast of Australia is cool but Tasmania is a “hot” spot. Nonsense, don’t believe it.
[On topic]
The probable cause of the Finland “unprecidented warming” will almost certainly be homogenisation similar to Darwin Zero.
Every temperature drop is an anomaly that has been adjusted out. Every temperature rise is doubled for good measure.
In order to maintain the charade in the face of the physical evidence, the “adjustments” will have to become more and more aberrant.
Expect more of the same in sparse areas where a few stations may influence the average over a large area.
The GISS hot-side polar anomalies are in areas with little/no independent validation/verification. It’s the GISS T-base differentiator; GISS measures where no one else is looking.
With the Y2K bug, GISS published incorrect data for seven (7) years and no one noticed but Steve McIntyre. No one in NASA or the broader climate science field noticed.
So the obvious questions are: “Are there reports, papers etc in the literature based on this incorrect data? Or does nobody acctually use this data?”
I think we deserve the complete traceback of how this happened.
Did someone copied Fiji for Finland data?
Why was the manual action needed? Was there a camera in the room when the fictitious data were overcopied. Wy was the camera switched off? This kind of stuff..
Quality entertainment , please!
Becoz we’re worth it!!
I wonder if Gavin has already run over to report this as an independent find of his own, owing nothing to Climate Audit.
Here is how CRU handled missing data – this method was peer reviewed:
http://www.eastangliaemails.com/emails.php?eid=596&filename=1134497252.txt
Am I doing something wrong?
The link given in Jean’s article appears to point to a graph of ANNUAL mean temperatures at Sodankylä, and the underlying data at the time of writing at:
http://data.giss.nasa.gov/work/gistemp/STATIONS//tmp.614028360003.1.1/station.txt
doesn’t appear to have any data for March 2010.
I can’t reproduce the global image in the article either. All I get is:
http://xrl.us/bhhcb3 (Link to data.giss.nasa.gov).
Help appreciated!
Ah, forget all that. I should have read the rest of the thread first ..
It appears to have been a sign error. I speculate it could be a reporting error from the finnish side. It is common in Finnish spoken language to just say “11 degC”, while actually meaning “-11 degC”. It can be a bit confusing when temperatures are close to zero.
Re: Aslak Grinsted (Apr 16 02:56),
If you change the sign of -10.5 you don’t get +1.5.
I think the error is for a day or days, not the whole month. There’s a post on WUWT which may explain it (assumign this is the error here).
http://wattsupwiththat.com/2010/04/17/giss-metar-dial-m-for-missing-minus-signs-its-worse-than-we-thought/
So losing the minus sign means -10 becomes +10 for a day, which changes the average for the month from -10 to +1.5 (N.B. example not to scale!)
Re: Aslak Grinsted (Apr 16 02:56),
C’mon, first, the Finnish language thing you are referring to does not apply to mean temperatures, only for reporting a particular cold temperature (“kymmenen astetta pakkasta”). Second, that type of speaking is hard to even (mis)translate to English; I doubt “Finnish side” is reporting their measurements in Finnish to NOAA/GISS. Third, I seriously hope that the measurements are not reported with speech communication.
Fourth, and the most importantly, a simple sign error explanation does not seem to hold water. I checked all values (as I said in the post, all of them set a new record) reported by GISS, and the range was from +1.5 (Sodankylä) to something like +3.6 (Helsinki). The true values are in the range from -1.8 (Helsinki) to -10.3 (Sodankylä), i.e., the dynamical range of the actual values is much larger.
Also the News item at GISS web page is IMO very worry some. It indicates that these absolutely crazy values passed both GISS and NOAA checks. A simple comparison with, e.g., GHCN-daily (which caries Helsinki, Jyväskylä, Jokioinen, and Sodankylä) should have revealed this problem.
I was wondering all this time why the Finns weren’t coming out of their saunas this winter. This clears everything up.
Great work Jean. If they missed a big one, how would they find the smaller ones?
OMG!!!! They took three whole days to correct this error!?!? Jim Hansen should be replaced by Anthony Watts immediately!!!
It seems though they didn’t correct it due to their own internal review process (which missed the error).
Are they relying on the public to review their work in lieu of their own quality control, or do they believe errors of this sort are insignificant?
Somehow I just find this sad.
Well, sad and fairly frustrating. If we were to seriously depend on these well-known, top professionals that supply a seriously large part of the world with temperature data upon which, if not in whole, but at least in majort part, the global average temperatures are calculated, it brings to mind too many serious and disturbing questions that I’m not going to ask a single one.
Oh wait, it’s not an “if”.
-Sale, Finland
Dutch climatesite http://www.klimaatgek.nl did some good work on the matter. Only disadvantage is the language…you need to translate it.
This may be bigger than Climategate, a basic mistake…..
http://wattsupwiththat.com/2010/04/17/giss-metar-dial-m-for-missing-minus-signs-its-worse-than-we-thought/
Its related to the story otherwise snip
Looking at the map it seems the hottest areas all seem to abut or lie within gray areas. I assume this means the remotest sensors are recording the highest temps.
Are they reliable?
It was asked early on how someone could create a 10 degree F discrepancy. Well, as I remember, GISS and NOAA decided to drop the ARGO buoy net as a source of raw data. As well as temp stations in northern latitudes, and picked ones near major metro areas in more southern latitudes. And then, to fill in the gaps of temps from deleted data sets, they use averaging algorithms that take data from as far as 1,000 miles away to project what the temperature “should be.” This altered and fabricated data record is then used as the actual data record, which allows them to say that such huge temps increases are happening. Which of course gets them more funding from the political orgs and governments that need them to prove such a thing. It’s akin to selling the customer what the customer wants. “You want fries with that?”
Of course, we already know about the data debacle at the CRU. The recent revelations of the dropping of entire data sets by both NASA/GISS and NOAA. So, it’s not that surprising that they would report Finland as warming drastically. This is what happens when we let politics control science.
It is very scary that the temp. is rising so fast, and for me this has a significant importance as i live in Scandinavia. They must come to an understanding soon and stop thinking about money and loans.
17 Trackbacks
[…] many of you know, Steve McIntyre reported suspicious readings are reported for […]
[…] el mapa de GISS observarán cómo Finlandia ha sido un horno este Marzo pasado. Curioso, verdad? En Climate Audit se han tomado la molestia de verificar la mancha roja finlandesa, con los resultados que eran de […]
[…] GISS: Warmest March ever in Finland A year and half ago Steve recalled some encounters with NASA GISS. One could imagine that after all that embarrassment […] […]
[…] NASA data to believe? 17 04 2010 Over on Climate Audit, Jean S points out a curious anomaly in the March GISS Temperature data for […]
[…] my last post, we talked about the a curious temperature anomaly that Jean S. found in the March GISS data and posted at Climate […]
[…] Bron: Climateaudit.org Meer gesjoemel vind je bij Wattsupwiththat […]
[…] GISS: Warmest March ever in Finland […]
[…] it demonstrates a serious weakness in surface data reporting. In my last post, we talked about the a curious temperature anomaly that Jean S. found in the March GISS data and posted at Climate Audit: The anomaly over Finland […]
[…] no thats not proof. NASA doesnt know what they are talking about Do you really want us to? "NASA doesn't know what they are talking about". No as a matter of fact they don't. __________________ Well, the trouble with our […]
[…] 17. Finnland-Gate […]
[…] climate change. 27. Five-star WWF-gate Swanky, carbon-spewing environmental cruise by the WWF. 28. Finland-gate NASA GISS shows record high temps for Finland in March 2010 when it was actually colder than […]
[…] influence-gate 14. Drought-gate 15. EPA-gate h/t Climate Depot 16. Five-star WWF-gate 17. Finland-gate 18. Flooded house-gate 19. FOI-gate 20. Fungus-gate 21. Gatekeeping-gate 22. GISS Metar-gate 23. […]
[…] climate change. 27. Five-star WWF-gate Swanky, carbon-spewing environmental cruise by the WWF. 28. Finland-gate NASA GISS shows record high temps for Finland in March 2010 when it was actually colder than […]
[…] requests. Steve McIntyre discovered the Y2K bug in GISS data. A few months later they found more slipshod work from the house of Hansen. Data from Russian stations was cherry picked. This is particularly […]
[…] 42. Five-star WWF-gate (Fox News) Swanky, carbon-spewing environmental cruise by the WWF. 43. Finland-gate (Climate Audit) NASA GISS shows record high temps for Finland in March 2010 when it was actually colder than […]
[…] Finlandia-gate (Climate Audit): GISS de la NASA muestra registros de altas temperaturas de Finlandia en marzo de 2010, cuando en […]
[…] climate change. 27. Five-star WWF-gate Swanky, carbon-spewing environmental cruise by the WWF. 28. Finland-gate NASA GISS shows record high temps for Finland in March 2010 when it was actually colder than […]