By Steven Goddard
GISS has explained their steeper temperature slope since 1998 vs. Had-Crut, as being due to the fact that they are willing to extrapolate 1200 km across the Arctic into regions where they may have no data – whereas Had-Crut prefers to work with regions of the Arctic where they actually have thermometers. WUWT reader “Wren” suggested that I compare GISS Arctic trends vs other sources to see how they compare. GISS has been showing Arctic temperatures rising very fast, as seen below.
However, GISS Arctic temperatures have been rising much faster than other data sources. The graph below shows the difference between GISS and RSS (GISS minus RSS) Arctic temperatures.
And the same graph for UAH.
Conclusions: GISS explains their increases vs. Had Crut as being due to their Arctic coverage. Their Arctic coverage is poor, and they rely on extrapolations across large distances with no data. Comparisons with other data sources show that GISS extrapolations across the Arctic are likely too high. In short, GISS trends over the last decade are most likely based on faulty extrapolations in the Arctic, and are probably not reliable indicators of global or Arctic temperature trends during that time period.
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
RACookPE1978 says:
May 21, 2010 at 2:29 pm
Where do you get your information from? How do you know what has been done to the data processing, and that things like ‘making the earlier data cooler’ were done?
Tim Clark says:
May 21, 2010 at 11:55 am
Phil: I don’t get to read here as often as I use to, but you really like to give novices erroneous information.
I answered the following question:
“So of what “independent satellite IR measurements” does he speak? Is there something other than the MSU lower troposphere data? ”
So I replied that: “Yes, there is data other than MSU, it’s in the NSIDC database (the one that Steve doesn’t think they have despite the DC standing for data center).
Here’s an example from April: http://nsidc.org/images/arcticseaicenews/20100504_Figure4.png”
What is “erroneous” about that? It’s absolutely factual, I even gave an example of what the poster asked for! Your objection appears to be that it’s NOAA data, well we knew that because NSIDC annotated the graph with NOAA/ESRL! It’s a database, convenient and searchable, who cares if the data wasn’t generated by NSIDC? In fact I’d expect something called ‘National Snow and Ice Data Center’ to be a repository of data rather than the organization that generates it.
RACookPE1978 says:
May 21, 2010 at 2:29 pm
I not surprised that GISS “research” and so-called data is so often quoted by AGW alarmists – that is their intended purpose. But it is scarcely “scientific” …
As far as I can tell none of your statements are true, where on earth do you get them from.
Hey, Steve. Aren’t you glad I brought this to your attention? Getting quite a bit of mileage from it.
You’re welcome. Anything to improve the science.
Krishna Gans,
My German’s a little rusty.
Clue: Steve Goddard claims to have been a volcano researcher in his early career.
Clue: Based on skiing at Taos claim is in his 50s or older
Clue: Not American, if spelling is anything to go on.
1. How could ANYBODY get temperature data from a moving ice sheet that gets dragged around by currents or blown across the ocean by wind, half of which melts and turns into open ocean every year? It seems some estimation of temperature is required.
2. The unmeasured arctic ocean area is about 2% of Earth’s surface. Is it possible for GISS extrapolations into this region to account for the 0.2C over 10 years difference that Steve’s GISS vs HadCrut plot shows? That would require a 10C change in the Arctic over 10 years, while Steve’s GISS arctic trend plot shows less than a 2C peak to peak over 10 years and his GISS – RSS plot is less than 1C.
Conclusion: There is much more to the GISS – Hadcrut difference than just the extrapolated Arctic Ocean data. (But what)
Gail Coombs, May 20, 11:53AM
So we can call somebody stupid as long as we use it in a quotation from someone else. OK.
There is no Nobel prize for name-calling.
(That’s a nice quote though!)
Tim Clark says:May 21, 2010 at 11:55 am
Phil: I don’t get to read here as often as I use to, but you really like to give novices erroneous information.
Phil. says:May 21, 2010 at 5:06 pm
I answered the following question:
“So of what “independent satellite IR measurements” does he speak? Is there something other than the MSU lower troposphere data? ”
So I replied that: “Yes, there is data other than MSU, it’s in the NSIDC database (the one that Steve doesn’t think they have despite the DC standing for data center).
Here’s an example from April: http://nsidc.org/images/arcticseaicenews/20100504_Figure4.png”
What is “erroneous” about that? It’s absolutely factual…
Bull, there is no “independent satellite IR measurements” in the NOAA ESRL database.
Mooloo said: “A correlation of 0.5 is pathetic. Seriously, in what other branch of science would a correlation of that size be regarded as useful for extrapolation?”
Aw carp! Now you made me go learn how to post pix. Thanks Bob Tisdale for the “how to”.
How “pathetic” is a correlation coefficient of 50%? Let me show you a 63% correlation.
http://i49.tinypic.com/sq40b5.gif
This is a “scatterplot” of CO2 vs. HadCrut temperature from 1850 – 2009. As it happens, the temperature record goes from 1850 on the left to 2009 on the right, because both CO2 and temperature are rising. There are more dots on the left because CO2 initially rose slowly so it took a longer time to get from 280 to 330 ppm than it did from 330 to 380 ppm, thus more dots on the left. All it would take to drop that correlation down to 50% would be to have the dots be just a little more scattered above and below the line. Like this:
http://i45.tinypic.com/2vvnrr4.gif
All I did in the above was add random noise to the temperature in the range of -0.25C to +0.25C. The gist of the correlation is still the same as before, even though the correlation coefficient is down to a “pathetic” 50%.
So you see, 50% correlation is not so bad. It only means that there is a lot of other noise in there making the data bounce around. When it comes to climate, that noise is “natural variation”, mainly El Nino, decadal and multidecadal (ocean-driven) oscillations, sunspots (don’t get excited, just +/-0.05deg C over the length of a sunspot cycle), and volcanos, to name a few.
That 63% correlation at the top looks like THIS when you use the equation to calculate temperature vs. CO2:
http://i45.tinypic.com/id664g.gif
That is real Hadcrut data (dark blue) and the calculated effect of CO2 on temperature (light blue). 63% correlation. In this case, 63% only tracks the long term trend, with noisier influences added. Obviously there are lots of short-term jiggles (a lot caused by El Nino) and some long term sways (caused by the PDO, AMO, or whatever you prefer), but the noise is almost always within +/- 0.4 deg C of the CO2-temp correlation. You could say that this 0.8C wide band is the range of natural variation. (Some attribute the WWII bump to a temporary change in ocean temperature measurement. I wouldn’t know about the 1910 dip.)
So most scientists would jump for joy for a “mere” 50% correlation!
Jbar says:May 22, 2010 at 9:26 am
So most scientists would jump for joy for a “mere” 50% correlation!
I don’t think so, at least none that I know. See (h/t Leif)
http://leif.org/research/CETandCO2.pdf
RACook wrote:
1. They (the GISS “corrections” to the original data) are undocumented, unaudited, and untraceable. That is – exactly WHAT is being changed to the original data, and what theoretical reasons are justified to CHANGE the original data has not been published, is deliberately being hidden from FOIA requests, and is NOT independently reviewed on a theoretical basis.
Has anyone tried a FOIA to get at the GISS data like they did with Jones et al at Hadley?
What was the FOIA output if any?
-Jay
Jay Cech says:
May 22, 2010 at 1:07 pm
RACook wrote:
“1. They (the GISS “corrections” to the original data) are undocumented, unaudited, and untraceable. That is – exactly WHAT is being changed to the original data, and what theoretical reasons are justified to CHANGE the original data has not been published, is deliberately being hidden from FOIA requests, and is NOT independently reviewed on a theoretical basis.”
Has anyone tried a FOIA to get at the GISS data like they did with Jones et al at Hadley?
What was the FOIA output if any?
-Jay
Well that approach hasn’t worked very well for McIntyre. You’d be better off just asking nicely, it’s worked for others who’ve been able to duplicate GISS results, clearclimatecode.org for example.
Phil, May 21, 2010 at 5:52 am
Thanks for the reply. Interesting data source. However, as far as I can tell, it’s not satellite IR measurements, as Hansen stated. When I said that his comment can be “addressed and summarily refuted”, it means that the satellite IR measurements either exist, or they don’t. If they don’t exist, his point can be summarily refuted.
I wasn’t assuming a conclusion. I was pointing out that this was a start. To reach a defensible conclusion, Goddard has quite a bit more work to do.
Oakden Wolf says:
May 23, 2010 at 4:41 am
Phil, May 21, 2010 at 5:52 am
Thanks for the reply. Interesting data source. However, as far as I can tell, it’s not satellite IR measurements, as Hansen stated.
The AVHRR (from 1978->) Brightness Temperature data is all acquired in the IR band (see below) and is what I assumed Hansen was referring to, there is also more recent data from ASMR-E etc. which is at a longer wavelength (microwave) and has advantages over IR. In any case distinguishing between IR and Microwave is more of a semantic argument than a real difference. The dividing line between them is usually for convenience 1GHz (30cm wavelength).
AVHRR/3 Channel Characteristics
Channel Number Wavelength (μm)
1 0.58 – 0.68
2 0.725 – 1.00
3A 1.58 – 1.64
3B 3.55 – 3.93
4 10.30 – 11.30
5 11.50 – 12.50
Tim Clark
Jbar said: “So most scientists would jump for joy for a “mere” 50% correlation!”
Tim said: “I don’t think so, at least none that I know. See (h/t Leif)”
Don’t see what in your link supports your statement. I will stay on point and not address Leif’s content.
There are more things than CO2 driving climate change. You will never get a high % correlation for any individual one of those things. Therefore a 50% correlation between temperature and any other individual factor is VERY STRONG.
100% correlation is an exact perfect match, and this almost never happens in real life.
The point that’s being missed is that the correlation referred to in the Hansen paper has nothing to do with a fit to a straight line which seems to be misleading posters here. What Hansen did was to construct a spatial correlogram from which he derived a length scale, he chose the 50% point as an upper limit. In turbulence studies the integral length scale is determined by integrating the area under the curve, from the shape of the correlograms the integral length scale is about 1200km.