HadCRUT global numbers are out, and is at 0.43°C, still lower than the GISS number of 0.67°C.
Click for a larger image
Once again Jim Hansen’s NASA GISS is the highest global anomaly:
RSS (satellite)
2008 1 -0.070
2008 2 -0.002
2008 3 0.079
UAH (satellite)
2008 1 -0.046
2008 2 0.020
2008 3 0.094
HadCRUT (surface, land-ocean)
2008/01 0.056
2008/02 0.187
2008/03 0.430
GISS (surface, land-ocean, polar estimates)
Year Jan Feb Mar
2008 .12 .26 .67
from what i can discern from the April RSS graphs to date
http://discover.itsc.uah.edu/amsutemps/
600 mb and up, they GISS and HADCRUT) may have a problem “forcing” April Temps up
Having looked at the data from all five “important” measurement groups, and compared monthly data, generally speaking, a large portion of the month to moth variation appears to be due to actual temperature changes at the surface of the earth. So, those are weather.
However, each month, there are also differences between measurements from each group. These are what I would call “measurement uncertainty”, or “measurement noise”. These differences are of the magnitude one would expected based on the measurement uncertainty estimates claimed by the various groups.
The true weather variability in month to month measurements appears larger than the measurement uncertainty.
Mike and others, the other point that no-one else seems to have mentioned is that while the jump from the absolute February anomaly to March is being discussed, surely the real story is with the difference from Feb to March, the two satellites steps being .074 and .077, while the HADCRU number is .253 and GISS a whopping .41. If we cant have consistent numbers, who do we believe? Or should the error bars be about +/- 2 degrees for the land numbers?
JK
It goes back to around 1920. The temperature trends correspond to ocean oscillation (up 1920s-1940s, down 1940s-1970s, up 1979-1998, flat 1998-2008) better than with CO2 increase, which took off c. 1950 right when temps got lower.
Not only that, but I distrust the CO2 record because it shows no increase whatever during WWII with all the full war production and burning cities.
The world temp measurements are somewhat higher in 1998 than in 1940 and I suspect the difference is roughly equal to the amount of bias added by the microsite violations.
It all has to add up. And the site violations must be accounted for.
What I think we are seeing is a real warming period (1978-1998) caused by ocean oscillation and exaggerated by Heat Sink Effect from severe microsite violation (that occurred after 1980). HSE will exaggerate a real increase, exaggerate a decrease, and not affect a flat trend much.
Henry said:
And, there is still the debate between GISS and HadCRU as to who does a better job of charting the anomaly. GISS supporters like to use the “GISS tracks Arctic temps, HadCRU doesn’t” line. If both systems used the same reference period, GISS’s “extra” warming goes away.”
However as I understand things, there really is no actual measurement of Arctic temps, but rather only estimates. And if that’s the case, GISS figures are meaningless because Hansen & Company can fudge their estimates until they reach their predetermined figures.
What a crock!
Jack Koenig, Editor
The Mysterious Climate project
http://www.climateclinic.com
The way I see it, the whole concept of an anomoly is largely useless because we don’t know and can’t agree on what is “normal.”
Darts anyone?
As far as I know, GISS set their baseline period before Hadley did, so castigating them for it is a tad off the mark. There is no reason to objectively prefer 1951-1980 to 1961-1990 or 1979-1998. And converting the temperature series to a common baseline is rather trivial; all you have to do is add the mean value of the new chosen baseline period to every point in the respective series. You end up with a pretty graph like this: http://www.yaleclimatemediaforum.org/pics/0408_gtr.jpg
If anyone wants the raw monthly anomalies for all four major series normalized to the same baseline period, I’d be happy to send you a copy of my excel sheet. Just email me at zeke@yaleclimatemediaforum.org
[snip – you are welcome to rephrase and resubmit that]
REPLY: It may be trivial for you and participants of this forum, but show me a general news reporter that can take the GISS data or anomaly graph, with its warmer component due to the base period they use, as presented on their website, and perform that normalization prior to print, and I’ll believe it’s not an issue.
It’s all about the presentation, GISS gets used more than the others, and it’s baseline choice presents the data with a greater positive offset than UAH, RSS, and HadCRUT.
Tony Edwards — that is precisely why I raised the question — all the discrepancies between the surface and atmospheric measurements. The error bars must be quite large. I wish that the various sources would include their error bars, otherwise, their graphs can be quite misleading at least short term. I have seen a hadcru graph with the error bars and that is quite revealing.
In fact, you inspired me to do the analysis myself. Here are the residuals between GISS and the mean of the other three temperature series, with everything normalized to the 1978-1998 baseline. Make of it what you will, but its clear that GISS is my no means an outlier.
http://i81.photobucket.com/albums/j237/hausfath/GISSResiduals.jpg
REPLY: Show your work, please.
The work that generated the above chart that Anthony Watts asked for:
Take the monthly anomalies for all four temperature series. Subtract the mean 1979-1998 value of the GISS series from every point in the series to normalize it to the same baseline used by RSS and UAH (e.g. subtract 0.238 from every point). Do the same for HadCRU (the value subtracted this time should be 0.146). Now, find the mean value of HadCRU, RSS, and UAH for each year, and find the difference between the GISS value and the mean of the other three for each year.
In retrospect, it might be more meaningful to compare GISS to HadCRU, since the land-based temperature series and the satellite based ones tend to differ from eachother in the same direction (e.g. when GISS is colder than RSS and UAH, HadCRU will likely be colder as well, and vise versa), so comparing GISS to the mean of one land-based and two satellite-based records might be showing some of the general differences between land-based and satellite-based measurements rather than anything unique to GISS.
REPLY: Thanks, I see in your most recent YCMF article that you agree with my issue about how reporters use graphs to present the points they try to make and must take some caution in doing so. I doubt any general media reporter has ever considered the issue of baseline differences when using one of these graphs.
This is why I think it would be useful to have them all on an agreed upon baseline.
Anthony, puhleeese get off your hobby-horse about GISS using a different normals base. All that does is modify the offset. The real scandal is that Hansen is passing off a GISStimate as real data. Go to http://data.giss.nasa.gov/gistemp/maps/ and select
Land: GISS analysis
Ocean: none
Map Type: Anomalies
Mean Period: Mar
Time Interval: Begin 2008 End 2008
Base Period: Begin 1951 End 1980
Smoothing Radius: 250 km
Projection Type: regular
and click on the “Make Map” button. This gives an idea of what the GISS land coverage (or lack thereof) is like. He’s missing lots of Antarctica, chunks of China and Eurasia and Brasil, most of the Arabian peninsula, and almost all of Canada Greenland and Siberia, and almost all of Africa other than the northwestern bulge.
I don’t know about his other missings, but there’s quite a bit of Canadian data to be had. Go to the Canadian government website http://www.climate.weatheroffice.ec.gc.ca/prods_servs/cdn_climate_summary_e.html and select “March”, “2008”, and “Plain Text” and click on “Submit”. A few seconds later, a text output comes up. There’s a code key at the bottom. Save the webpage as a text file (e.g. as x.txt).
One of the columns is labelled “Tm” for monthly mean temp. Another one is labelled “D” for deviation from the latest long-term mean, which happens to be 1971…2000. Also, watch the column labelled “DwTm” for days without mean temp, i.e. number of missing days. Let’s say we’re willing to accept up to 2 days of missing data. Anyone with unix/linux (or even Cygwin on Windows) can use the following command to extract those lines from x.txt to y.txt…
grep “^.\{57\} [012]…[0-9]” x.txt > y.txt
The result for March 2008, is 220 data points. Obviously, someone has to go to the trouble of digging up 1951..1980 normals for these sites. Some newer sites will not not have normals for the 1951..1980 period, but there should still be well over 100 sites that can be used, along with their mean temperatures as listed here.
What are the height thicknesses used for the sat measurements. Anyone know?
Not being a mathematician, scientist, or statistician, it would seem to me that the sensible thing to do would be to for scientific reasons to adjust the baseline to the newest historical dataset with enough length to be reasonably accurate as a beginning trend line. It is unreasonable to expect the newer method of measurement to try to extrapolate temperature measurements prior to the beginning of their taking of active measurements. If comparisons must be made to reach any sort of accuracy then the base lines must be normalized.
It just appears that it is almost a contest of I am better than you.
are not
am so
are not
am so
I think they should just fix it they are all supposed to be seekers of the truth and scientists.
Just my 2 cents
Bill Derryberry
Using the same base period as HadCRUT(1961-1990)
GISS
Jan 08 .04
Feb 08 .18
Mar 08 .55
Go to GISS homepage, click on Global Maps and make your custom map with whatever baseline you want.
http://data.giss.nasa.gov/gistemp/
GISS has an unusual amount of missing data for March. Most of which looks to be negative anomaly. If it weren’t for that, GISS would be much closer to HadCRUT like the previous 2 months.
Zeke,
I am speaking from ignorance on this subject as in my work assembling data for statistical analysis is very straight forward. Adjusting raw data is verboten.
It would be useful to have one standard baseline. That said, what is the justification for Hansen’s endless “adjustments” of data? This has been addressed on numerous occasions at ClimateAudit, including recently:
http://www.climateaudit.org/?p=2964
Has anyone figured out exactly how and why Hansen makes these multiple adjustments throughout the record? There is no question these adjustments affect year to year (and month-month) record, but the trend as well.
Now, if GISS is not an outlier, how is it then Hansen keeps trotting out these “record” temperature statements (via the Press) when the other three products do not, particularly the satellite data which clearly show a decline since 2001? 1998, 2002, 2005 (the highest) and 2007 in the GISS record are very close to the same.
Or is the GISS data a result of error in the measurement itself?
So my question is, since anomalies rather than actual temperature are the metric by which temperature changes are reported, how do we actually know if temperatures are rising or falling? In other words, UAH/RSS show a ~.08 global rise, GISS .41 and you can bet the GISS record is being milked for all it’s got.
It would seem the further out from the current decade the baseline is, the less relevant the anomaly is if the baseline is derived from a particularly cold/hot or departing (up or down) period.
Jeezuz CHRIST. You repeated a comparison with different base periods, so that the anamolies have a built in bias. Didn’t you learn that from the last kerfuffle. you need to be castigated. You are dumb! You need to hold an M-1 rifle for long periods at extend arms. You need to be whipped into shape. You are so, so fucking slack. And don’t give me any crap about sugar and vineger. Shape up.
REPLY: I originally blocked this comment, but I thought I should leave it up as a demonstration. Now we’ll see him and the usual suspects run over to Tamino, post something like the above, and there will be yet another round of screaming angry posts.
Also you’ll no longer posting on this blog. Bye TCO, add WUWT to the list of blogs your juvenile language has banned you from.
See, here’s the thing the angry phantom people like TCO don’t get. I know how the base period works, my point is in the presentation of premade graphics and data for public consumption and use by the general press.
Be angry, yell, scream, call names, do what ever you like. It won’t change the fact that this GISS anomaly graph,
http://data.giss.nasa.gov/gistemp/graphs/Fig.A2.gif
which is probably one of the most widely distributed in the world, would look a lot different if plotted using a different base period.
Again, IMHO, all four metrics should have a common base period for the wide PR and public data/graphs they make available so that anomaly graphs and data presented in the general press is shown and compared on an even presentation field.
That’s it. I don’t think it is an unreasonable request.
“The true weather variability in month to month measurements appears larger than the measurement uncertainty”
Lucia I respect the work you do but I do not agree with the above statement. From what I can tell the measurement uncertainty could be several degrees C. The fact that the resolution is perhaps less the tenths of a degree speaks nothing of measurement uncertainty and absolute accuracy.
The anomaly dance often trips up people. It would be good practice to
Label the charts with the base period and ALSO, the average for the base period.
It really doesnt matter what base period you use, since adjusting from one to the other is simple.
I will give a simple example: DATA: 1,2,2,3,3,4
AVERAGE of the 2nd and 3rd numbers: 2.
ANOMALY with regard to that period. -1,0,0,1,1,2
Now do the anomaly dance to base period of the 4th and 5th year: average 3.
ANOMALY -2,-1,-1,0,0,1
Question: how do you shift a series based on year 2 and 3, to one based on
on year 3 and 4?
easy. just do the anomaly dance again.
Zeke.
I took a little different cut at the problem. Frst I averaged the satillite data
Two instruments looking at the same thing. makes sense.
Then I went to look at how GISS differs from this record, and how hadcru differs
from this record. To see if there was a systematic difference. I just started that
You can steal the idea if you think it has merit.
Normalize to a common base period ( 79-98 ) first.
The obvious solution for the baseline issue is to make the entire dataset the baseline. This avoids the cherry-picking of a baseline issue.
I’ll suggest they don’t do this, because it will over time make the warming (from a straight line trend) look less and less impressive. It, of course, won’t effect the trend or month to month comparisons. But will effect the anomaly value of a month or other period (BTW an essentially meaningless number) as the baseline changes.
Steve Mosher,
My previous post was not clear and in error now that I reread it. I understand it “shouldn’t” make a difference with respect to the baseline used and they “should” reconcile if adjusted for different baseline periods. However, the anomalies are clearly not the same.
The GISS data it is said has a ripple effect that can reach through the entire record after it is “adjusted”. Adjusted how?
There should be no reason to constantly adjust the data once it is initially done. I recall after the so-called “Y2K” error was found, it was barely a month before 1998 magically was even with 1934.
http://www.climateaudit.org/?p=1880
Looking at the Leaderboard, 2005 wasn’t even in the list in August 2007, but after over 90 “adjustments” 2005 somehow now is top dog, and Hansen makes sure everyone is aware of it. 2002 was not in the top 10 either.
Zeke,
As far as adjustments go, I believe NASA applies the same TOB and homogeniety adjustments that NOAA uses. At this point you must also realize that all organizations grid thier adjusted data in 1200KMx1200km gridcells. At that point, there are various extrapolations used, especially in empty grids. Climate Audit has found so many problems (as has Anthony) with not only raw data, but also miscatogorized stations (rural vs urban), UHI questions (esp concerning China data used by Phil Jones), that the statisical analysis performed by all organizations is quite useless from a scientific perspective. In any other field, the error bars would be so great as to render any analysis useless. At least we can be thankful that the automotive engineers who design our brakes have higher quality standards.
I’ve been looking at the data at
http://www.cpc.noaa.gov/products/global_monitoring/temperature/global_temp_accum.shtml
And one question I have about Hadcrut and GISS is how do they weight a particular data point? For example, you may have 1,000 data points in the continental US, but only 50 data points in Alaska. Yet Alaska has a surface area roughly 20 percent of the continental US. So do the 50 data points in Alaska get a weighted average to compensate for the lack of data points?
If this is not done on a consistent basis, how can anyone speak reliably of an average global temperature when it is biased towards nations and areas with a greater density of temperature measuring sites?
Evan Jones; is the HSE effect related to Roy Spencer’s thesis that cloud flux causes surface temp vaiations and not vice-versa? A twist to Spencer’s work and the solar issue generally is at:
http://www.jennifermarohasy.com/blog/archives/002914.html
On the issue of baseline discrepancies, are they generally corrupted by such events as the Pacific Climate shift described in this paper?
http://mclean.ch/climate/Aust_temps_alt_view.pdf
VG
Thanks for this link: http://discover.itsc.uah.edu/amsutemps/
I hadn’t been aware of it. It’s great. Comparing 2008 with 2007, one sees the atmosphere has really cooled off.
It’d be great if such data were available for the oceans.