Y2K Bug in NASA Climate Data

[AtlanticFlamethrower]

NASA has now silently released corrected figures, and the changes are truly astounding. The warmest year on record is now 1934. 1998 (long trumpeted by the media as record-breaking) moves to second place. 1921 takes third. In fact, 5 of the 10 warmest years on record now all occur before World War II. Anthony Watts has put the new data in chart form, along with a more detailed summary of the events.

The effect of the correction on global temperatures is minor (some 1-2% less warming than originally thought)

Link

[Optimus Prime]

Interesting but it still doesn't distract us from the current record breaking warm period we are in now albeit apparently not as warm as 1934.

For the UK though 1934 wasn't nearly as warm as present. With a yearly CET 9.99c, yes it's a warm year but not excessively so (+0.57c) easily the warmest since 1921 (which had an average 10.47c) but December 1934 was extraordinarily warm widely the anomoly was +3.5c - 4.0c above average.

[AtlanticFlamethrower]

Furthermore, only 3 (not 9) took place since 1995 (1998, 1999, and 2006). The years 2000, 2002, 2003, 2004 are now below the year 1900 and no longer even in the top 20.

[Optimus Prime]

I find that hard to believe AF.

[West is Best]

I find that hard to believe AF.

The link is to a blog linking to a blogger ... it's not exactly hard science by the look of it. It would be good to see something a little more solid published properly on the subject. If you find that in due course AF please be sure to let us know!

[VillagePlank]

the joys of modulo arithmetic ....

[parmenides3]

Sorry, AF, but this is a deception put out by some denialist blogs. The temperature record was adjusted slightly as a result of some outstanding digging by Steve McIntyre, which revealed a potential discrepancy in the measurement systems in the 1930s. NASA looked into it quickly, and realised that McIntyre had found something important. They then recalibrated and adjusted the current US land anomaly figures downwards by about 0.16C. The effect of this was to make 1934 return to its position as the hottest year on record in the USA, a tiny bit warmer than last year.

They also didn't do this 'secretly'; this was another attempt by denialists to discredit NASA's surface temperature record, by implying some kind of dishonesty. Surely, if they had wanted to be dishonest, they'd have ignored the problem, wouldn't they? It also ignores the fact that they mention the adjustment openly on their own website.

This should be reassuring; it shows that scientists are constantly looking to make their material as accurate as possible, and are willing to recognise both downward and upward adjustment, which kind of puts paid to the notion that all science is exaggerating GW.

For a detalied discussion, see RealClimate's post yesterday : http://www.realclimate.org/

:)P

[VillagePlank]

Hi all,

I have to say that I am almost entirely certain that this is not a Y2k bug issue for the following reasons:

(i) Scientists often use Julian days for dates, not Gregorian so the Y2k issue is irrelevant. Indeed even if you are a programmer and you use the an internal date type you are using a Julian system underneath anyway. In Microsoft products a modified Julian system is used where the world began in 1899.

(ii) Why would the data be linearly correlated with the date? A Y2k issue implies that the date forms part of the magnitude of the Y axis.

(iii) If a Y2k bug was at fault then normally systems either crash (so don't work at all) or some hitherto previous date is used affecting the magnitude of some other date and not the temperature around the Y2k period.

I don't know why the error occured, but I am fairly certain the Y2k was not at fault - primarily for reason (i)

[Bobby]

Sounds very dodgy to me. A blogger? I'll wait until someone with a bit more stature clears things up. I fail to see what it has to do with Y2K too.

[Thundery wintry showers]

Note that the findings, if accurate, show vastly less warming only over the USA. Globally, the warming is shown to be only 1-2% less.

1934 might have been the US's warmest year on record, but averaged globally it doesn't change the assessment that 1998 (or 2005, depending on which source you use) was the warmest year globally. In the meantime, the 'climate change' issue that most of us are concerned about is global warming, not US-warming.

[AtlanticFlamethrower]

Ignore the "blogger" - sorry I even linked to it.

Here is NASA's updated figures.

Yes, everything I quoted is true of the statistics, check them out. Incredible.

[parmenides3]

Ignore the "blogger" - sorry I even linked to it.

Here is NASA's updated figures.

Yes, everything I quoted is true of the statistics, check them out. Incredible.

What's incredible, AF?

:)P

[AtlanticFlamethrower]

It's brand new data, corrected. What we we were looking at before was false.

http://data.giss.nasa.gov/gistemp/graphs/Fig.D.txt

Nasa's new "top 10" chart shows 6 out of 10 warmest on record occured before 1952. 4 out of the 10 occured in the 1930s.

 Year	   Old	   New
1. 1934 	1.23 	1.25
2. 1998 	1.24 	1.23
3. 1921 	1.12 	1.15
4. 2006 	1.23 	1.13
5. 1931 	1.08 	1.08
6. 1999 	0.94 	0.93
7. 1953 	0.91 	0.90
8. 1990 	0.88 	0.87
9. 1938 	0.85 	0.86
10. 1939 	0.84 	0.85

Edited August 11, 2007 by AtlanticFlamethrower

[parmenides3]

It looks as if most of the figures have been adjusted by hundredths of a degree. This doesn't seem that amazing. Data correction is not that unusual; if scientists find that the data they have been using has a flaw in it, they change the data. If it disproves a hypothesis, they change the hypothesis. What this proves, if anything, is that the scientific process is open and transparent, and we should be pleased by that. Another example might the the argo float data from last year; first analysis suggested a cooling of the oceans between 2003-05. Subsequent analysis showed a fault in some of the floats; net result; an adjustment which showed a small net warming.

You need to change your bottom line; 1934 was the warmest year on record so far in the USA, not globally. The warmest year on record globally was (just) 1998. Until next January, that is.

:)P

[Stratos Ferric]

It's brand new data, corrected. What we we were looking at before was false.

http://data.giss.nasa.gov/gistemp/graphs/Fig.D.txt

Nasa's new "top 10" chart shows 6 out of 10 warmest on record occured before 1952. 4 out of the 10 occured in the 1930s.

 Year	   Old	   New
1. 1934	 1.23	 1.25
2. 1998	 1.24	 1.23
3. 1921	 1.12	 1.15
4. 2006	 1.23	 1.13
5. 1931	 1.08	 1.08
6. 1999	 0.94	 0.93
7. 1953	 0.91	 0.90
8. 1990	 0.88	 0.87
9. 1938	 0.85	 0.86
10. 1939	 0.84	 0.85

AFF, let's be absolutely clear les you inadvertently go down the same road that I think the originator went down somewhat mischieviously: these data are for the lower 48 states in the US only. If anyone cares to read some of the responses in the blog the data may be cast in a subtly different light. As TWS says, the impacts globally are way short ofbeing material.

It looks as if most of the figures have been adjusted by hundredths of a degree. This doesn't seem that amazing. Data correction is not that unusual; if scientists find that the data they have been using has a flaw in it, they change the data. If it disproves a hypothesis, they change the hypothesis. What this proves, if anything, is that the scientific process is open and transparent, and we should be pleased by that. Another example might the the argo float data from last year; first analysis suggested a cooling of the oceans between 2003-05. Subsequent analysis showed a fault in some of the floats; net result; an adjustment which showed a small net warming.

You need to change your bottom line; 1934 was the warmest year on record so far in the USA, not globally. The warmest year on record globally was (just) 1998. Until next January, that is.

P

Quite. I ave to say, casting an eye over the CET, it would appear that the millenium bug struck here about ten years early.

[AtlanticFlamethrower]

US figures have been used as a proxy for global figures.

The "minor" change resulted in - 1-2% less global warming and reminds us climate science is not complete.

Here's the graph - the black lines are the actual yearly numbers

Considering Northern Hemisphere sea ice must have been greater in the early 20c, Earth coming out of a little ice age, that's quite a sight to behold.

The global graph therefore can be seen as two steps rather than a continuous warming curve ... assuming the figures for the rest of the globe are accurate.

[Devonian]

US figures have been used as a proxy for global figures.

News to me, so, by who and when?

[Iceberg]

Quite agree what the two graphs show us is that US temps should NOT be used as a proxy for global.

The second graph is very misleading in it's start date,scale and 0 mean, even so the warming is very noticable.

[AtlanticFlamethrower]

The irony is the US temps have been seen as particularly reliable out of all global records - huge area, continental landmass, most modern measuring techniques in the world going back many years

[Devonian]

The irony is the US temps have been seen as particularly reliable out of all global records - huge area, continental landmass, most modern measuring techniques in the world going back many years

Nice try, but they still are. The change is mostly to records post 2000.

[AtlanticFlamethrower]

Nice try, but they still are. The change is mostly to records post 2000.

They are reliable but we will hear no more from the enviros about global warming in the US dataset. It simply doesn't exist in this dataset any more as you can see from the graph (previously posted).

BTW changes have occured to nearly all of the data. The largest change however may be to the post-2000 data, which has become cooler.

Year	   Old	   New
1. 1934	 1.23	 1.25
2. 1998	 1.24	 1.23
3. 1921	 1.12	 1.15
4. 2006	 1.23	 1.13
5. 1931	 1.08	 1.08
6. 1999	 0.94	 0.93
7. 1953	 0.91	 0.90
8. 1990	 0.88	 0.87
9. 1938	 0.85	 0.86
10. 1939	 0.84	 0.85

Edited August 11, 2007 by AtlanticFlamethrower

[Bobby]

The temperature of the US definitely doesn't indicate global temperatures. Also as the adjustments are only very small, if the changes were plotted on a chart the warming trend would still be clearly visible and almost unchanged.

Edited August 11, 2007 by Magpie

[AtlanticFlamethrower]

The temperature of the US definitely doesn't indicate global temperatures. Also as the adjustments are only very small, if the changes were plotted on a chart the warming trend would still be clearly visible and almost unchanged.

[Bobby]

Well, that's the US chart, not a global one. It also shows sharp warming since 1970, a good 37 years of it.

[AtlanticFlamethrower]

Well, that's the US chart, not a global one.

That's right. Previously US temperature record was the "jewel in the crown" of the global warming dataset. No longer.