On January 16, 2008, I posted a note on Hansen et al 1988 containing the following graphic comparing the three Hansen scenarios to the most recent GISS and RSS temperature versions.
.
Although there had been much furore in the past about the differences between Hansen Scenarios A and B in previous controversy, I observed that Scenarios A and B were both at very elevated levels by 2010 and that noticeable increases in RSS temperature would be required to keep pace even with Hansen Scenario B. While I expressed this in terms of the RSS data, the same thing is true for the GISS surface data.
This graphic used the following file downloaded from RSS on January 16, 2008
http://www.remss.com/pub/msu/monthly_time_series/RSS_Monthly_MSU_AMSU_Channel_TLT_Anomalies_Land_and_Ocean_v03_0.txt
Eli Rabett today accused me of using “an older version” of the RSS temperature reconstruction as follows:
But lo, our auditors had also used an older version of the RSS microwave tropospheric temperature reconstruction. It had a serious error. What does the corrected version look like?
An older version?
Two days after my post, on January 18, 2008, RSS issued an amended version of their TLT data http://www.remss.com/pub/msu/monthly_time_series/RSS_Monthly_MSU_AMSU_Channel_TLT_Anomalies_Land_and_Ocean_v03_1.txt
Unfortunately, there is no announcement of the error at the RSS homepage http://www.remss.com/msu/msu_data_description.html. As far as I know at present, the only notice of the error is through a readme (http://www.remss.com/pub/msu/data/readme_jan_2008.txt ) dated January 16, 2008, but only posted up on January 18, 2008 (see http://www.remss.com/pub/msu/data/ ) . Rabett did not state how he happened to become aware of the readme mentioning the error as to my knowledge, it is not linked through the home page or through an explicit announcement. Rabett linked to the readme, carefully including the reported date of January 16, 2008, but failing to mention that the readme was posted on Janueary 18, 2008 (also after my post).
Last January, I made a small change in the way TLT is calculated that reduced the absolute
Temperatures by 0.1K. But I only used the new method for 2007 (the error).
When the data are merged with MSU, MSU and AMSU are forced to be as close as possible to each
other over the 1999-2004 period of overlap. This caused the error to show up as a downward
jump in JAnuary 2007. To fix the problem, I reprocessed the 1998-2006 AMSU data using the new
code (like I should have done in the first place), and merged it with the MSU data.We would like to thank John Christy and Roy Spencer, who were very helpful during the diagnosis
process.Carl Mears, RSS, January 16 2008
Thus, on January 16, 2008 when I did my post, I was using the current RSS version. Yes, two days later, RSS changed their 2007 numbers. Here are two graphics showing the impact of the RSS changes – Jan 3, 2008 (red) and Jan 18, 2008 (blue):
and a second one showing the difference between the two series, which works out to about 0.12-0.14 deg C during 2007.
Also here is a revised version of my graphic, implementing the Jan 18, 2008 RSS changes. The main point of my earlier post clearly stands — in 2010, the difference between Scenarios A and B is not particularly large and some of the past furore over scenario versions becomes less material.
In his coverage of the RSS error, Rabett disparaged my analysis, implying that, by using an “older version” of the RSS data, I had done something improper. Rabett misleadingly failed to cite the dates of the RSS versions, which would have shown that I had used the then current RSS version (Jan 3, 2008, then current as at Jan 16, 2008) and that that the newer version (Jan 18, 2008) was not available at the time of my post. Rabett shows the readme dated January 16, 2008 implying that I should have been aware of it (even though it is nowhere linked on the RSS webpage) without stating that it was posted on January 18, 2008 subsequent to my post.
In the same post that Rabett criticized here, as originally written, I had incorrectly missed a comment in Hansen et al 1988 saying that Scenario B was the “most plausible”, an error which I picked up about 8 hours after the original posting (about 9 am EST) and immediately corrected it when I noticed it. So there was an actual incorrect statement at CA for about 8 hours. Imagine that. I didn’t post up notice of the change until about 9 hours later (I was playing in a squash tournament and had to do some chores and went out after making the correction and posted the notice when I returned.) Meanwhile, a few hours after I made the correction, Lambert wrote a post on this error without mentioning that the error had already been corrected as at the time of his post. This caused a tizzy of excitement over at Deltoid over the possibility that I might actually have made a mistake on something (outside my core area). Gavin Schmidt interrupted his day at NASA to weigh on the matter at Deltoid.
I don’t claim to be infallible. I corrected this particular error promptly. Compare the treatment of an error in an incidental blog post that I corrected within about 8 hours to the treatment of the RSS error. Do Rabett or Lambert excoriate RSS for their goof? A goof that occurred in a high-profile data set? Of course not. To the extent that they place any blame, Rabett blames me for not using an RSS version that didn’t even exist at the time of the post.
As to RSS, they made a mistake and corrected it. Good for them. Errors happen and RSS corrected their error. Rabett observed of the RSS error:
PPS: Lots of folk are falling into the RSS error trap.
Rabett proceeds to blame users of RSS data. The fact that users are “falling into the RSS error trap” is one more good reason why RSS should have issued a clear error notice, rather than the obscure readme. They should issue a proper notice of the error in their public webpages and wherever else appropriate. They were pretty quick to publicize errors by Spencer and Christy and should accord equal publicity to their own error.
Update Jan 24, 2008 12.40 pm:
Here is a version showing including the MSU version centered to synchronize with GISS for the period up to 1987. As you can see, the MSU version was bracketed by the GISS and old RSS versions. These spaghetti graphcs get pretty messy pretty fast and since it was bracketed, I did not include it in my prior version. Eli Rabett has made an issue of this and I have accordingly shown the MSU version here on the same basis as the prior graphic. As you see, it is no longer bracketed by GISS and the new RSS version, but is slightly lower than either, the point of my post remaining unaffected.
Update: Jan 27, 2008
John Christy observed below that “one must either multiply the surface by 1.2 or divide the troposphere by 1.2” to gave apples and apples. I will re-plot the graphic once I clarify a point on this.
Climate Audit 
149 Comments
Maybe somebody can develop a model that will predict ex post facto dataset changes that will be made in the future so you won’t go making such silly errors. 😉
rssht3.gif seems to have a swapped legend – Jan 18 should be blue as in the accompanying text and as graphed, and red etc for Jan 3rd
Steve: Fixed.
RSS is kind enough for versionning its data pre & post correction (for the moment). Don’t even dream about it for GISS and CRU.
Well, now we shall see what kind of gentleman Josh Halpern is.
Steve,
Actually, I interpret this as a compliment to you and your work- that they would get so worked up, and pat each other on the backs over errors that were not your fault and/or were quickly corrected by you in an open manner, shows how little actual error they have been able to identify in your work.
That’s a pretty systematic change of the 2007 records. Why exactly is the raw data adjusted so steeply?
Does this error “matter”? IOW, if we take the same tack as RC would, does this small error change the main point of Steve’s post – that there needs to be a large change in temperature by 2010 to match Hansen A/B/C scenarios? That we are currently below even Hansen’s scenario C? Doesn’t look to me like it does.
So there you go – another entry for Bender’s database of double standards. Always assuming that he doesn’t already have such an entry given the response to the Loehle paper corrections.
It’s obvious you are upset by this Steve – and quite rightly,IMO. Know that:
1) you are not alone
2) As usual, you display great restraint and keep your cool – not sure I could.
Actually there are two in there:
#23. skeptics are to be blamed for all errors in their analyses; warmers are allowed to shift some blame onto their sources
#24. appeal to sensitivity analysis to show errors are inconsequential is ok for warmers, but not skeptics
The MSU people need to do more. The old version (link) of the data is still up at the same address, without comment that it has been superseded.
Still, it’s a relief because the RSS data for 2007 now aligns better with the UAH and HADCRU3 series. The 10-year trend on RSS has gone from -0.06 degrees C per decade to + 0.01 degrees per decade, so the world is warming after all. If this goes on we’ll be ruined within 5,000 years.
It all shows the uncertainty of determining global temperature trends when tenths of a degree are important.
4, when you’re a bunny trying to save the earth, you can’t be a gentleman:
Apparently the GCM calculators do not have a method in place to alert output-data users to errors/problems in the calculated numbers. The users must check the errata Web page at PCMDI. A proper Software Quality Assurance program would have a method in place for rapid communication of such information to all users.
This page at the Program for Climate Model Diagnosis and Intercomparison (PCMDI) says:
The errata page is here.
Am I the only one who finds this admission of apparently unilateral and unexplained “changes” a bit unsettling? Moreover, it sounds like this took a while to be (a) identified and (b) corrected. Perhaps John Christy and Roy Spenser can provide a little more insight into the reasons for the change and why it took so long to get it right.
Looks like UAH now. Steve Milloy needs to update this, I guess:
http://www.junkscience.com/MSU_Temps/RSSglobe.html
If by the time you see this, it looks fine, then he already did so.
I kind of figured that UAH was right on this one. That big negative anomaly seemed weird.
Still doesn’t square with “2nd warmest on record” though.
#10 While the new data aligns a little better, there is still a pretty significant difference, looking at just the past five years. I have some trend analysis I’m doing, using (global, land and sea) NCDC, MSU-RSS, and MSU-UAH, breaking the 1979-2007 period into 14, 10, and 5 years. The trends (stated as C/decade) break out like this:
1979-1992
NCDC 0.104C/decade ***
MSU-RSS 0.074C/decade **
MSU-UAH 0.021C/decade
1993-2002
NCDC 0.289C/decade ***
MSU-RSS 0.339C/decade ***
MSU-UAH 0.360C/decade ***
2003-2007
NCDC 0.017C/decade
MSU-RSS -0.110C/decade
MSU-UAH 0.027C/decade
*** monthly trend significant at 99% level
** monthly trend significant at 95% level
All three source show no “significant” trend since 2002, though the sign of the MSU-RSS trend is sharply negative, and this after using the corrected data. Statistically, there is pretty good agreement for all three sources for 1993-2002, and 2003-2007. The low trend value for MSU-UAH for 1979-1992 — low and not significantly different than zero — stands out from the other two data sources, however. FWIW, I’ve computed specific constants for each series as well, but do not report them above. I’ll post visual plots at the CA forum if there is any interest in seeing them.
In my first sentence to #15, I probably should have said “dramatic” rather than “significant” (so as not to infer that the difference is statistically significant).
About a month or two ago I saw a statement (can’t remember where) by Christy that he figured that there was a error in the RSS data. I’ve been waiting to see if there would be a correction. I find it interesting that Christy and Spencer helped with the problem and that fact makes me trust them more that the “warmers”. Also the RC types loved the RSS data when it was higher than the UAH. Kind of interesting that the GISS and satellite tmperatures are so close.
This is not a criticism of Steve in any way Kudos Steve, for quickly making the changes and for this Blog and phpBB.
I could almost believe Hansen C. Not sure what it does out past 2010. But as it stands, it’s not all that bad.
Hi,
#15 I wonder what that would look like if you removed the 1997-98 El Nino. Can we be sure the the current RSS number is correct. Why do they do all of this in secret. I though we the tax payers are footing this bill. What ever happened to the Freedom of Information Act. They should post the raw data for all to see.
Not to pick more nits but in Hansen et al 1988 they say Scenario B is “perhaps” the most plausible one. It is in more recent papers that Hansen has coauthored that the conditional “perhaps” has been dropped.
After reading all limitations of the climate modeling and scenario producing that the authors freely admit, I would guess that Hansen was as surprised as anyone that, at least until recent times, the B anc C scenarios approached actual temperatures — from his choice of temperature data sets.
Re #13
They explain them on their website, changes are necessary because of changes in the satellites, melding the MSU & AMSU etc., they’ve used 9 satellites to date. Spencer and Christie do the same on their site, both groups have stopped using NOAA 16 for example.
RSS
http://vortex.nsstc.uah.edu/data/msu/t2/readme.03Jan2008
I wouldn’t worry too much about Rabett or Deltoid, Steve. Things must be tough on their end if they need to invent this sort of controversy.
Re Hansen’s Scenarios, however, the question that interests me are the trajectories Scenarios B and C take in the next few years and how well or not obs. temp. do in tracking them, especially considering the principal assumption Scenario C makes. Which brings to mind Pielke’s Snr request for a determination of the current radiative forcings of CO2, etc. and not for the difference between 1750 and 2005. We have two years in which some people will be very nervous, and this will coincide with some serious negotiations leading up to Kyoto II. We live in interesting times.
Re #17
Well Steve McI recentered them for the purpose of this comparison so they should be reasonably close, actually his color choice makes the GISS data disappear, very suspicious 😉
#19 Adding a dummy for the 98 El Nino doesn’t have much impact on the trends. It is statistically significant, but there is still a pronounced trend in 1993 to 2002 period. I’ve got to recrunch the numbers to clean up for serial correlation, but that just changes the details, not the overall picture. With the new numbers, I’ll include a variable for the 98 El Nino.
I guess this has been discussed before, but shouldn’t all the curves cross in 1988, the year the prediction was made? Scenario C is significantly lower than all the others in 1988.
RE 18.
Have you seen how hansen is sandbagging the 2008 projection.
This is the first time I have seen them come out and project that a year wont be that warm.
Cool strategy: If they are right, “it’s El nina” . If they are wrong, GHG overwhelmed El Nina.
Pretty damn slick. They are actually thinking one move ahead… Moshpit moves.. c5
26, the pipeline explains all.
And speaking of Bobby Fischer, will his death make more people want to move to Iceland for the thermal springs as we enter the possibility of a new little ice age?
Steve writes: “This graphic used the following file downloaded from RSS on January 16, 2006
http://www.remss.com/pub/msu/monthly_time_series/RSS_Monthly_MSU_AMSU_Channel_TLT”
You mean January 16, 2008?
Steve: yes. fixed.
There are two references to January 16, 2006 in this article. Are they correct?
Steve: Typos fixed – 2008.
This should be considered here:
Click to access 2006_Hansen_etal_1.pdf
Steve: any thoughts on the results if the current, lower forcing estimate (3 + or – 1 … I think) were to be run through the old model? Mightn’t the results be much closer? In all events, Hansen’s estimates (particularly B) aint too bad. Beats most economic models any day. Also, how does his “most probable” estimate compare to, say, similar estimates made in your business? I have no idea how to make a meaningful comparison (apples … oranges … etc.) but perhaps you could give a concrete, real world example from your line of work that would better put Hansen’s numbers into some perspective. That way people around here won’t just stand around gawking and saying “see, he was not exactly right — I told you global warming was a hoax.” All the best … and thanks!
Steve,
About Rabett, he’s invisible, like Harvey. Pay him no mind.
Anthony,
In a way, I agree with you. If not for Steve, I would not know who Rabbett is. As far as I can tell, he has made no positive contributions to the field. However, if Steve does not correct the record – some people will think Rabbett is right.
Keep up the great work, Steve.
Rabett is reportedly Joshua B. Halpern, Professor of Chemistry at Howard University (http://www.coas.howard.edu/PREM/Personnel/Faculty/Halpern.html). If this is correct, why is he so ashamed to associate himself with his pronouncements on what should be a valid scientific topic?
Thank you for pointing this out. I’ve been having a heated discussion with a true believer as to whether or not 2007 was the hottest year on record. Sadly, it appears the adjustment is in teh wrong direction for my case. Well, the data is what the data is….
MattN, you should point out that even GISS puts this as only the second warmest
And by the way, UAH 2007 is 4th.
Perhaps Mr Halpern is afraid that people will think less of him for being a jerk on his blog?
The moral of the story seems to be, trust UAH, not RSS.
It`s pretty simple. Either he made a silly error – or he has done this deliberately.
Which is it Rabett ? Did you make a error ? Or is this deliberate deceit on your part ?
Does he have the integrity to answer this ?
Andrew (#36) says,
The Hadley CRU index of urban airport and water bucket temperatures (HadCRUT3 global) shows 2007 as #6 (through Nov. only). Can’t win ’em all!
And what sound do lagomorphs make?
heh, heh. that makes no sense now.
bTW these plots were done without re-centering the scenarios. I experimented with re-centering on 1958-87 since that was the period available as at 1988 publication and it turned out – presumably by happenstance – that the 1958-87 mean for Scenario B was exactly the same – and I mean exactly to 8 decimal places – as the GISS surface mean for the same period. So the native scaling and 1958-87 centering are exactly the same. Re-leveling for A and C on the same basis would be slight.
The revised RSS temperatures still show a huge decline of 0.475C during 2007 from January to December.
It temps increased 0.7C from 1900 to 2006, we still lost well over half of that increase during the big cooling trend of 2007.
The latest sea surface temperature map shows that La Nina has strengthened even more throughout January. As a consequence, the lower atmosphere numbers from RSS for January will be down even more.
RE #38, as a thirty-something, I made a cardfile of openings and thought seriously about how to write a chess program, for exactly the same reason. But I made a wrong move somewhere along the way, and now I’m a published Climate Scientist!
Jim Arendt, thanks for your contribution, wherein you say inter alia:
While most of the players in the climate data game are paid for by the taxpayers, my understanding is that RSS (Remote Sensing Systems) is a private company. Thus, they don’t have to show you any steenkin’ badges …
Erik Ramberg, you say: [snipped]
If you seriously think that being rated the Best Science Blog means that every comment by every poster will be science at its best, you need more help than we can provide.
However, the poster you were objecting to said:[snip]
In this case, the data keeper and the modeler are the same person … which doesn’t bode well, and in many situations would be expressly forbidden. Doesn’t prove anything, of course, it’s just circumstantial evidence … but there’s five major global temperature records, and the one that shows the most rise, by quite a ways, is also under the tender care of someone who spends many hours of their working time not actually working, but instead giving alarmist statements to the press about how the global temperature is rising …
And if you don’t know which data he is referring to … then all bets are off.
w.
============================================================================
“Some circumstantial evidence is very strong, as when you find a trout in the milk.” Henry David Thoreau
(I said this to my daughter, and she said “But how would a trout get in the milk”? So I had to explain how in “the olden days”™, they used to add water to milk so they’d have more to sell, and etc. etc. … kind of humorous, trying to explain a corrupt business practice that hasn’t been seen in the US in a century or so …)
Willis Eschenbach #44 writes,
Now they call it 1% milk. Former “corrupt business practice” becomes heath food craze!
OK, OT, mea culpa…
Re #41
By these plots you mean the new ones as opposed to the ones in your previous post?
RE #43, BTW, I did write a killer 3-D tic-tac-toe program instead!
(Fortunately, Steve’s in bed by now.)
Hu, I’m banking on a “warm” December, obviously. 😉
Good luck with being a published Climate Scientist, now. You might need it.
I did some tests on Scenario B as though it were a reconstruction calculating r2 statistics for a “calibration” period of 1958-1987 and “verification” period of 1988-2007 and (since satellite coverage was short) the same calc for “calibration” 1958-1997 (the date of the first dispute) and verification of 1998-2007. The correlations between the GISS temperature data and GISS models was much higher than the correlation between the GISS model and satellite data.
For GISS station data the r2 for 1958-97 was 0.54 (1998-2007 0.08), for the GISS global data (which I’ve illustrated here) the r2 for 1958-97 was 0.49 (1998-2007 0.11); for the RSS (new) data the r2 for 1958-97 was 0.06 (1998-2007 0.03); for the UAH-MSU global data the r2 for 1958-97 was 0.01 (1998-2007 0.01); and (for a change of pace) for the Angell radiosonde lower troposphere (850-300 mb), the r2 for 1958-97 was 0.21 (1998-2007 0.07).
Your graph didn’t make it. Those are pretty dreadful verification r2 .
Please forgive me for being a slow lurker, but did RSS’s 2007 correction make the result warmer or colder. I think that I have seen both stated here in various posts.
#51 You’re kidding right? Look at the 2nd and 4th graphs. You know how to read graphs?
Where are the opposing statements? Point to them.
#49 No correlation at all. I just read the RC post of 11 January about the distinction between ‘weather’ and ‘climate’ and how it was so important to remember that the GCMs predict ‘climate’ not ‘weather’. But my guess is that if they performed with any correlation at shorter intervals it would be all over the evening news, so to speak. I think there is no real distinction, just a continuum, and any prediction is better than no prediction. So the ‘weather/climate’ distinction seems like more of a rationalization for poor model performance to me.
If an error of 0.1 degrees was made in the math from earler years, how does the correction in the last year become greater than 0.1 degree?
A simple explanation by Carl Mears, RSS, of the January 16 2008 graphic correction would suffice. In the final analysis, digging deep, was the adjustment made to to absolute temperature or was it to a temperature anomaly? Did those danged thermometer things give the incovenient answer again?
http://www.climateaudit.org/?p=2648#comment-202481
Re #31 Eric McFarland
There are many people from geological backgrounds, like me, commenting on CA. I suspect this is because of some commonality of method and some objection to alterations in the accepted way to do things like stats (see Wegman report).
You ask if performance figures can be given for Steve’s line of work, of which I do not have many details. Climate work has a similarity to the estimation and distribution of economically recoverable grades of ore in a deposit partially understood by analysis of a relatively small number of drill holes, so we can make a quick and dirty comparison.
There is no global estimator of performance, because the grades of various types of minerals do not follow a uniform regularity from one deposit to another. For example, gold can be pretty “noisy”, while alumina is more predictable. Apples and oranges again. Grades are not constrained to a statistical distribution like Stefan’s law, but a distribution is invariably calculated in the routine process, not for prediction so much as for quality control.
A significant pre-mining expenditure is estimation of the number and attitude of drill holes required to give a sought reliability, the assay interval chosen down the hole, and the quality control of assays. Typically one assay in 10 to 50 is replicated. Part of the drill material is physically stored under careful conditions for later use. More than one method of assaying or analysis is commonly used. Analysis variances are investigated until correction is understood. I think proxies are about absent as a method, depending on fine definitions.
So what is an acceptable variance? First, the accuracy has to be acceptable, usually assessed by comparing different analytical methods for a start. Second, the precision has to be acceptable, derived from stats of repeatability of multiple assays. Third, the extraoplation between drill holes has to be ok, another stats exercise that commonly leads to more, expensive drilling. Fourth, the boundary between economic and sub-economic ore has to be estimated with acceptable certainty – and product prices can change in the process. Fifth, the amount of total metal (or other ore) that can be metallurgically extracted has to be estimated. Then, if the deposit is polymetallic, the above have to be repeated for each item of interest (e.g. lead, sinc, silver, gold). Sixth, hindcasting for holes already assayed has to be aceptable. If you think about it, there are some similarities to climate modelling, except the temporal aspect is absent.
In the final defition, an acceptable estimate on a deposit rich enough to mine is one that will return a profit.
That is, acceptability is not based on tweaking of models, or on taking a liking to a particular forecast. It is based on actual data on metals or ores that can be seen and weighed as they are produced.
The mining world has a large number of technical successes, where the grade was just too low to make a profit. These can be mined by bad decision, or left for an upturn in demand.
But the point I want to make is that it is not a statistical or mathematical figure that expresses success or failure of variability estimates. It is the cold, hard reality of money going into the bank, or the company going out of existence.
Climatologists seem to underplay the significance of the last paragraph. “Accountabilty” is a word that comes to mind.
#55 Thats all very well. Gladly we live in a brave new world,
miners are honest folk, the Bulletin is no more
and Pierpont is only seen at Christmas.
http://bulletin.ninemsn.com.au/
Re #55, #56: Lately I have chosen to play a lurker role and let the drama unfold.
However, I feel the need to speak up in Geoff Sherrington’s support, and cannot allow the important point he makes to be trivialised.
As he has perhaps demonstrated in this post, Geoff can be loquacious. His main point though is that geologists and mining engineers are daily accustomed to dealing with uncertainty and risk assessment, and working diligently to minimise risk exposure (not just investors, but to the environment, local community, and other stockholders these days).
Geologists and mining engineers have learned to be very disciplined and a strong body of good practice has accumulated over the years. For a good example, have a look at the JORC Code: http://www.jorc.org/pdf/jorc2004print.pdf
The benefit of a Code of good practice such as the JORC Code is that it is a very effective means of educating practitioners in the field, and a means for holding them accountable. Breaches in the public domain are reportable to the AusIMM Ethics Committee which takes its responsibility to uphold good standards very seriously.
The Corporations Law (in Australia) and the Australian Stock Exchange endorse the JORC Code, and play a key role in enforcing it. As a result, standards have improved dramatically, and the satire of Pierpont is, somewhat at least, less relevant these days.
I think that climate science would benefit from the development of a similar code of good practice. In time, Steve McIntyre (and many others) will be recognised for the valuable contribution they have made to society in raising standards of accountability, just as those behind the development of the JORC Code over the past 30 years or so have rightly been applauded.
Please note that I am writing from an Australian perspective. Many companies are now adopting similar codes to the JORC Code, in many cases, closely following the JORC template.
Having demonstrated my capacity to be just as loquacious as Geoff, I now retreat to my lurking role.
Re 56 david
Pierpont (Trevor Sykes) got burned by Poseidon and did an audit function of poor quality sompanies for some decades. What’s wrong with an audit?
re # 57 Trevor
In younger days I was Fellow of Aus IMM, wrote the corporate manual on quality control of assaying, did perhaps the world’s first large corporate CO2 budget and contributed heavily to corporate governance principles for the resources industries, the precursors of JORC.
For my 30 active years it has mostly taken a long response to counteract a short throw-away line.
Please look up the meaning of “loquacious” to better understand my lengthy attempts to write text (typing excepted) accurately.
Dave Stockwell:
What does it mean when the satellite record (either version) has such a poor correlation with the GISS surface record?
When Christy and Spencer adust their data due to an acknowledged error identified by Mears (as has happened before), the Rabett, Bloom, etc, crowd jumped all over C&S and claims their UAH analysis is bunk. When the opposite occurs…silence.
I think his students, posts, and blog have already revealed he’s quite a poor one.
Re #60
Well they’re not actually measuring the same quantity. However according to the graph Steve McI posted above they’re an almost perfect match, so much so that you can hardly see the GISS result at all.
By the way Steve, RSS had a notice on their website of the correction since at least yesterday http://www.remss.com/msu/msu_data_description.html#version
That the RSS data was in error should not have come as a complete surprise, as we discussed Christy’s thoughts about it back on January 9th in an update to our January 8th piece in which we thought the error was in the UAH data (as did the UAH folks as recently as mid-December). However, upon further analysis, the UAH folks grew more confident in their product and determined that the error must lie with the RSS. This process apparently played out over the last weeks of December 2007 and the first weeks of January 2008. According to Christy’s readme (where he always announces the tweaks he makes to the UAH data) by Jan. 3rd, he pretty much figured that the UAH data was correct. That Mear’s also documents corrections in a readme that accompanies the data doesn’t seem unusual, in that respect. Although, admittedly, I, too, have been nosing around the RSS site on several occasions (including post Jan.18th) to see if they had announced any sort of correction and did not come across the updated readme file until I saw Rabett’s coverage. I guess it is just part of learning how different data managers document tweaks to the data.
Eli made a good point.
Steve did include the RSS data into his graph, BECAUSE of the down tick. he wouldn t have done that with the corrected data.
#60 John, the GISS and satellite records are well correlated aren’t they? Do you mean scenario B and RSS?
#58 Geoff, no offence meant, I am with you on that. Pierpont always gave me a laugh. Actually, in reference to matters like this thread, I sometimes struggle to find a better response than condemnation to the things I see going on in the AGW world. Pierpont to the mining world seems like the kind of gentle mocking Steve does well to the climate science world. Your response re #59 is/was a productive too.
#31. Fair question. You say:
I’ve not suggested that econometric models are any sort of gold standard here. When I was young, I spent some time looking at econometric models for the purposes of forecasting copper prices and concluded that they had little to offer. Some of the problems were pretty fundamental – in order to make projections, you had to be able project things like Chinese or Russian economic activity, when (At the time) there was virtually no information on the economies of these countries and no real way to project what would happen. If you lacked information on such items, modelers inserted feedback loops with unpredictable characteristics and all you really got out were the properties of the loops. So I’ve never set much store on econometric models nor would I hold them out as an example to climate modelers – other than some of the hard lessons in econometrics should be studied.
My own guess is that GCMs introduce a wide variety of irrelevant complications into the analysis of CO2 impacts and that if the only GCM output that you are interested in is the course of global temperatures, that there’s a lot of algebra that more or less cancels out and that you can pin down the essential items in a 1-D or 2-D model that could be intelligibly expressed to scientists from other fields. In this respect, the early radiative-convective models and their developments and updates would be the sort of thing that I would have built up if I were designing an exposition for policy-makers.
Let’s suppose that I said that I had a system for predicting football scores and in Super Bowl week last year that Indianapolis was going to win the Super Bowl and they did. Maybe I even did this 2 years in a row. Would that prove that my system worked? Obviously not.
Let’s suppose that a few years the stock market was at 10000 and I gave three scenarios: A(business as usual) – the stock market goes to 16000; C- it goes to 11000; and B – it goes to 14000. I say that, if I were “Forced” to choose, that B was the “most plausible” and elsewhere that the “most probable” course is that it went somewhere between A and B. Then the stock market went to 12800. Did that prove that my system worked? I would have made money, but it wouldn’t prove that my system worked. But obviously Warren Buffett is better at the stock market than you or me, so I wouldn’t say that knowledge is impossible.
Where do Hansen’s predictions sit in this sort of scenario? In statistical terms, I don’t think that the outcome of the past 20 years is any more definitive than in the stock market case. Hansen supporters would be entitled to say that their model embodied physics rather than mere correlations and they would be entitled to do so. This certainly adds weight to the prediction that does not exist in the Super Bowl example. As to how one allows for that in assessing statistical significance, I’m not sure. IT seems reasonable enough. If someone has some statistical references where this type of problem has been considered, I’d be happy to consider these references.
Re #63
See #62 and #21 They documented it on their ‘Versions’ entry on the Website as well as in the ‘readme’, it was certainly there yesterday afternoon.
#62. You say:
I checked their website when I made this post and, while it’s not impossible that I missed the notice, I think that the notice was put up subsequent to my post and perhaps even in response to it. It would be interesting to know exactly how Rabett became aware of the change.
In his usual ungracious way, Rabett conceded nothing about the fact that, on Jan 16, 2008, I could not use a data set issued only on Jan 18, 2008. Instead, he sourly observed that I continued to use the “old” RSS version after January 18, 2008. In practical terms, I updated the graphic within a few hours of becoming aware of the update (which I learned of via Rabett), but I observe that it’s impossible for me to troll through the RSS website on a daily basis to check whether a readme has been inserted saying that a new data set has been substituted. For that matter, it’s impossible for me to check the RSS webpage on a daily basis to see whether their data version has been changed. I updated the graphic promptly when I became I aware of the update – and in this case, the update was made within 5 days of the update despite the somewhat “quiet” nature of the update.
Wow, so you can read minds?
Even the corrected data is well off of Hansen. It would be just as worthy of a post!
The correction of 0.12-0.14 deg C is well within the “the error doesn’t matter” range established by Gavin, Hansen, etc. It still puts the RSS further away from Hansen than the GISS surface record.
As Steve says, “I observed that Scenarios A and B were both at very elevated levels by 2010 and that noticeable increases in RSS temperature would be required to keep pace even with Hansen Scenario B.” This is still quite true. The RSS correction just reduced how much they would have to increase. A substantial warming in the RSS record would still have to occur.
#64. C’mon. I showed the GISS version and the RSS satellite version. As it was, the GISS line is hard to pick up in the later portion where it overlaps RSS, but is quite distinct in the early portion. If I added more spaghetti, it would have been harder to see. The MSU line was in between the GISS line and the RSS line. Had I not shown the GISS line, you might have had a case, but I showed it.
The point of the post remains: the argument over SCenario A or SCenario B becomes moot as both show substantial increases by 2010. To keep pace even with Scenario B, there needs to be a dramatic increase in temperatures in the next few years (regardless of index). This could easily happen and would be very strong support for Hansen’s parameterizations. If temperatures are about at present levels in a few years, there would be enough separation between Scenario B and observations that one could reasonably suggest that lower parameterizations were being indicated.
#69. Ironically, the 0.15 deg error in RSS is almost exactly the same as the amount of Y2K error in the U.S. data.
Hi,
#44 My point stands, why do they have to be so secret about this. Its not national security. Simply give they raw data so others can review, is that too much to ask. Hansen is great at hiding all of the numbers and then 5 years later putting it out with an update on past temperatures.
#72. the data issue has been discussed at length here. Thompson’s Dunde ice core was taken in 1987 and sample details remain unarchived (he has only, after my complaint, archived 10-year summaries of dO18 for a portion of the core.) Not just Dunde, but other cores as well.
Re #68
Since he quoted the ‘readme’ rather than the web announcement perhaps he tried to download some data and came across it there?
I wouldn’t expect you to constantly check the RSS site, the reason I posted the web announcement was so that you could update the statement in the original post so that any reader could read it for themselves.
SteveMC. I raised a similiar issue WRT the data archives of GCM data held at the IPCC.
The onus is on the user of the data to check back and see if any errata have been published.
They don’t proactively inform users of the data that the data has changed. I get better notices
about flaws in my damn car than flaws in climate science data. It is a trivial matter
update subscribers when data changes.
Ironic that its called RSS.
http://en.wikipedia.org/wiki/RSS_(file_format)
One enigma related to this matter (Hansen’s scenarios) seems to have escaped notice so far, though not in the literature. Methane is presumably the third most important GHG (after water vapor and CO2). Methane concentration has been steadily increasing in the atmosphere over the past couple of hundred years, as presumably have anthropogenic emissions, via mostly agriculture, and ruminants farting. However, the rate of rise in methane concentration has started slowing down since the 70’s, to the point that the concentration has been flat, and even slightly decreasing over the past ten years, despite the fact that emissions have been increasing. Obviously, there is something we don’t understand about the methane cycle, and various hypothesis have been proposed to explain this behavior, though none seems to have been definitely proven.
Methane used to make up for 25% of the anthropogenic forcing 20 years ago, but now is only 20%. Surely this would affect Hansen’s scenarios in a totally unexpected way.
Funny too that even though methane decreases, Greenpeace just issued a report claiming that we must reduce methane emissions, and they even proposed that everybody switched to vegetarianism, which caused a minor uproar in our local cattle raising business. Greenpeace failed of course to mention that one of the largest sources of anthropogenic methane comes from rice cultivation, and that pork and poultry only emit about one fourth as much methane as ruminants.
Re #76
Yes although his scenario C was eerily prescient on CH4, quite remarkable considering we still don’t really understand why it happened. The thought that CH4 is a good early target for early mitigation is not a bad one since many of the industrial emissions are small scale accidental releases unlike CO2 which is a large scale waste product in combustion processes. Not much that can be done about termites and ruminant emissions though (could shift to white meat rather than vegitarianism).
I thought I’d give communicating with AGW alarmists a try. So when a poster on Deltoid claimed that you couldn’t have possibly corrected the “scenario B” error at 9 am since it there wasn’t a post about it until ~6 pm, I carefully reviewed the timeline of CA posts there and pointed out that, while there wasn’t a post regarding the correction until 9 am, the correction had, indeed been made at 9 am. That didn’t work. I repeated the timeline in more detail (and very politely). The poster still failed to get the point. I referenced the sailent part of my timeline and asked to be corrected if I’d gotten something wrong.
They never responded.
So much for communicating with the “other” side.
Steve: As you say, the change was made at about 9am Eastern and I noted the change about 6 pm. Didn’t you mean to say that “while there wasn’t a post regarding the correction until 6 pm” rather than 9 am?
BTW I often edit posts quite a lot in the first hour or so of their life. I write straight to WordPress these days and often edit when I see the appearance of the post. If people are hanging on every word that I write in real time, I may have to change this, but so far it hadn’t seemed like a problem.
#74. You wouldn’t notice the change in many forms of download programs. For example, I have an automated script to download the current RSS results which uses a file name. The data set under this file name is still there and nothing explodes when you use it. There’s a tradeoff between maintaining a record of the old data and making a correction.
One further precaution that they ought to take: they change the name of the old file to say 3.0_old so that automated references to the file are disabled, while still preserving the record. That way, someone using an automated script would have to go and check when the read failed. Right now, an automated user would have no way of knowing that he’s picking up a now obsolete version.
#77
eerily prescient?
voodoo climatology
Didn’t know he acknowledged your existence. Isn’t this progress?
Phil. says:
Prior collective panic attacks such as the population bomb have a common theme: factors that were not well understood at the time ended up rendering the problem non-existent. The CH4 data is an example of an unforseen factor that demonstrates the original GHG alarmist claims wrt CH4 were exagerrations and there was no need for punative economic measures designed to reduce CH4 emissions in 1988.
If the current flat temperature trends continue for another 5-10 years then we will likely be able to make the same conclusion wrt CO2 (i.e. there is probably an effect but it is too small to worry about).
The CH4 data tells me that we would be better off waiting another 10 years before committing to any anti-CO2 measures that do not offer benefits unrelated to CO2 (i.e. reducing dependency on oil imports is a good thing no matter what the science says about CO2).
To be fair, a significant jump in temp across all of the datasets (i.e. not just GISS) during the next (expected to be weak) solar max would suggest that the alarmists managed to be at least partially right and that we better step up the efforts to do something about CO2.
For me, the temps in 2013-2015 would have to exceed the 2003-2005 values by at least 0.2 degC. Anything less would suggest the CO2 effect is below the range predicted by IPCC.
They have changed the name to:
RSS_Monthly_MSU_AMSU_Channel_TLT_Anomalies_Ocean_v03_1.sav
I don’t know how that would affect your automated program?
The 3.0 product isn’t being updated so presumably the user would soon notice and investigate.
Steve: They didn’t “chane” the name. They added a file named RSS_Monthly_MSU_AMSU_Channel_TLT_Anomalies_Ocean_v03_1.txt without re-naming RSS_Monthly_MSU_AMSU_Channel_TLT_Anomalies_Ocean_v03_0.txt . Anyone using the RSS_Monthly_MSU_AMSU_Channel_TLT_Anomalies_Ocean_v03_0.txt in an automated script would have no notice of a problem. The next update isn’t for a while so why would anyone notice right now?
HadCru. 2007 8th. December anomoly 0.211
I’ve added the following to the above post:
Here is a version showing including the MSU version centered to synchronize with GISS for the period up to 1987. As you can see, the MSU version was bracketed by the GISS and old RSS versions. These spaghetti graphcs get pretty messy pretty fast and since it was bracketed, I did not include it in my prior version. Eli Rabett has made an issue of this and I have accordingly shown the MSU version here on the same basis as the prior graphic. As you see, it is no longer bracketed by GISS and the new RSS version, but is slightly lower than either, the point of my post remaining unaffected.
They wouldn’t, that’s why I said ‘soon’, i.e. when the january update didn’t come (they appear to have updated for Dec on Jan 3rd so that would be just over a week from now).
#86. I’m not sure that their updates are done on a regular monthly schedule. It doesn;t say that anywhere. Ask them and see,
#42 John Lang — The drop from January to December in 2007 didn’t suddenly erase half of a century of temperature change. It didn’t even erase half of the change we’ve seen since the beginning of the satellite record; to do the latter, it would have to drop back below the zero anomaly line, which is no longer the case with the revision in the RSS data. But there’s still something interesting going on, to be sure. Here is another way to look at the data:
I’ve fit separate trend lines for 1979 to 1992, 1993 to 2002, and 2003 to 2007, with a dummy for the ’97-98 El Nino. FWIW, all variables are significant for the middle period — intercept, trend, and dummy. For the first and third period, only the intercepts are significant, not the trends. Between the first and third periods there is clearly a “warming,” but it is shift, not a trend. Since 2002, it is tempting to say “global warming has stopped” but it is a bit soon to say “global cooling has started,” though I’m with you in your interest seeing what impact the La Nina will have.
re #78
Yes, that is what I meant to say.
Deltoid Update from Tim Lambert:
In the above post I said:
Tim Lambert has now analyzed my posts and observed that I have a comment at 11 am blog time (eastern time is a little later and this would be lunch eastern). I note that my next comment was after 6 pm blog time – about 7 hours later. So you may interpret this as at least evidence that I was out. As to how I spent my morning, jeez – is this what Lambert is reduced to? I don’t keep a daybook. But for interested parties, I did not “rush” out to play squash; I played squash in the afternoon. I did some chores in the morning; I would have been home for lunch and it’s not unusual for me to check the blog. I was mostly doing other stuff on Friday, but it looks like I made a post around lunchtime [11 am blogtime]. Perhaps I made the changes around 11 am blog time rather than 9 am eastern, but I’m pretty sure that I did them pretty soon after getting up rather than at lunchtime. MAybe Lambert wants to know what I had for lunch.
This is exactly what the squealing alarmists are reduced to. Microanalysis of Steve M’s daily schedule as opposed to Hansen’s GCM scenarios. It’s patheticomical.
So ridiculous the difference between the standard they have for Steve and the one they have the orthodox.
We should all take a moment to just lean back in our chairs and have a very good laugh at how the honest efforts of Steve is driving these people absolutely nuts.
Their pettiness properly reflects on the worthiness of their agenda. (and the lack thereof)
Phil.:
This is at least the second time you’ve mentioned the eerie prescience of Scenario C and CH4, so it seems you somehow feel it’s significant. But for Scenario C, Hansen didn’t assume some unknown phenomenon would cause CH4 to level off. He assumed the imposition of draconian caps on greenhouse gases. I don’t see how happening to be right for completely wrong reasons qualifies as prescient, much less eerily prescient. Particularly since Hansen offered Scenario C more as an unrealistic “what if?” baseline than a projection.
I noticed at RC, their Model E results (in the water vapor feedback or forcing post) do not include methane.
As far as complaining about using data that’s not even available yet, as well as not being predominently displayed once corrected, I sometimes wonder what kind of people are so zoned out they don’t think others will notice their contradictory and hypocritical behaviours.
As far as the anomalies, it is helpful to not just mention what it was, but where it was from. The global climate at a glance at NOAA’s NCDC has different numbers on the chart than are in the GISTEMP file and the latter is more up to date and includes seasons, monthlies, etc.
2005 is the year with the highest yearly anomaly in the GISTEMP file and 2007 is tied with 1998.
The smoking gun:
Re # 65 David Stockewll
Point taken and appreciated. No probs.
This flows from your suggestions and is not a typo:
“Brevity is the soul of it”
Tim is working on the indictment of the ham sandwich as I write.
=======================================
How many years have we been waiting for Hansen to correct some of his mistakes (The rain in Seine for an example)? Yet if Steve doesn’t correct something in a matter of minutes, it’s evidence that he’s careless with the truth.
MJW, #94:
When you’re in love, little things like reality and causality are often ignored.
@Phil Re 46 asking about 23.
I’m pretty sure Steve means that if you take the average of 1958-1987 GISS measured GMST and the computed values from the model B for the same time frame, they come out almost exactly the same.
I know there are lots of people who run numbers here, so I guess they could all check to see what they get when they:
1) Calculate the average temperature anomalie for the GISS model B run from Gavin’s file.
2) Calculate the average GISS temperature for the same case from GISS annual average data. (Met data, not land/ocean– right SteveM?)
Compare to see if they match.
Re #101
Lucia, the reason I asked was because I’d answered a question saying that in his calculations Steve had recentered (he’d said so in his earlier post). Then later Steve posted (but not replying to my comment) that there had been no recentering, so I was trying to get clarification (with no luck).
I’m sorry I didn’t come up with that post I promised you, final exams took precedence! I’ll try to get back to it next week.
Steve: Phil, I experimented with centering – my objective being to get apples and apples as centering is something that comes up all the time in proxy studies. As it turned out, if you centered the Scenarios on 1958-87 which was a plausible suggestion, Scenario B came out at exactly the same level as the present GISS – so re-centering Scenario B on this basis was moot. For other scenarios and for other plausible reference periods, the re-centering was always less than 0.1 deg C and typically under 0.05 deg C – so for all the huffing and puffing that any given re-centering hypothesis was wrong , the choice of centering reference period had minimal impact on the comparison to GISS. I centered the satellite series on the 1979-87 period to match GISS over the same period, as noted before.
Phil– Don’t worry about the post.
I saw Steve hadn’t answered your question. That’s why I answered. I would have said why I think that’s what he means, but I’d be relying on a fuzzy memory. But, I remember before I decided to compare model predictions to data using post 1984 data, I did some comparisons of baselines with 1951-1987 data, and… well… yes. One set of agreements is hauntingly similar. (I don’t remember which.)
I didn’t attribute much significance to this, but it did send me to reread the figure caption on Figure 3 in Hansen 1988 to see if they haunting similarity was due to choice of zero baselines for the computations and the experimental data.
So, now that Steve mentions this, I suspect that’s what he means. If someone checks, they could see if that’s the comparison he finds hauntingly similar (to eight decimal places.)
Here is the corrected RSS global temperature anomaly time series, with ENSO “removed” and with the two volcanic eruption episodes erased. The method for ENSO removal was covered in a prior post.
Re #104 “Removing” ENSO, by the way, does not affect the overall 1979-2007 trend. It does affect short-term swings and may be useful in a search for other factors affecting global temperature.
David Smith– Could you provide us the link to the post where you describe how to remove enso? Your name doesn’t have a url, and it wasn’t obvious to me based on the url of the image.
Also, why remove or erase ENSO or Volcanic eruptions from the temperature anomaly series?
Re #106, yes but I’ll have to look for it after fajitas and margaritas 🙂
What I’m looking for are patterns that may otherwise be masked by ENSO variation or by the immediate aftermath of the two volcanoes. For instance, the swing from 1987 to 1989 intrigues me – what was behind that, if anything? Did something similar happen in 1998-1999? Maybe it’s random variation or maybe it’s driven by something. Simple exploration and playing with data.
David, might I suggest comparing it to solar cycles or Cosmic Rays? Like in this paper:
Click to access Scient_No._3.pdf
(I know about the trend, but I blame the heat capacity of the oceans. That’s Nir’s and my story and I’m sticking to it)
All: My participation regarding the RSS error is as follows. Last fall the divergence between UAH and RSS was becoming significant. Since we had only NOAA-15 contributing data at the time and RSS had applied a diurnal correction to correct for spurious warming in their NOAA-15 data as the spacecraft drifted backward into a warmer time of day, we assumed our uncorrected LT data were spuriously warm, and made a note in the readme file. However, after accessing data from NASA’s AQUA satellite, which has onboard propulsion and so does not drift and so requires no correction, we found our NOAA-15 data and AQUA were almost identical (i.e. NOAA-15’s diurnal drift was tiny at this point.) Carl and I shared a number of emails and I pointed out in mid-January 2008 that the real shift between the datasets was actually Jan 2007 when I looked a little closer. He checked it out and discovered that a small change in his code in Jan neglected to adjust the post-2006 data with adjustments already applied by the data to that point. He fixed it quite quickly.
That said, the recent publications on a number of tests for precision indicate that UAH data are more consistent with a number of other realizations of trends and correlations (Christy and Spencer 2005, Christy and Norris 2006, Christy et al. 2007, Randall and Herman 2008). Though we’ve issued a few press releases on these papers, virtually nothing was picked up by the media, so these results remain in obscurity. Nothing new.
Now, I have one misrepresentation to point out on Steve M.’s charts. The temperature comparisons shown are not apples to apples. All climate models indicate the global tropospheric temperature should warm at a rate of 1.2 times that of the surface (1.4 times that of the surface for the tropics – see CCSP SAP 1.1. or Douglass et al. 2007). So, to put surface temperature projections from models on a chart with observed tropospheric temperatures, one must reduce the tropospheric temperature trend by a factor of 1.2 for the comparison to be legitimate. I think the result would be of interest to the readers, and it is entirely defensible as shown in numerous publications.
Re # 109 J Christy
Forget the models for a moment, please. What do the measurements say?
There is an ever-increasing populace who have lost faith in model predictions.
Re #110 Good paper
#111 That’s the one that G. Schmidt tried to refute by showing how massive the error bars are on the GCM output. ‘See, they overlap!’ That’s how desperate he was.
#112
http://www.realclimate.org/index.php/archives/2007/12/tropical-troposphere-trends/langswitch_lang/in
Text search on “Not quite so impressive” will take you to the figure caption.
lucia, here are links to the posts using global and tropical data. As indicated, I’m always looking for better ideas so any thoughts or critiques are appreciated.
An alternate view, using unadjusted data, is here . In this one I’m looking just at the 1980s. The lines diverge around early 1982 and converge around 1987. Is that simply random variation? Or, was the divergence related to El Chicon? If it was, why did it last so long? Did PDO/NAO/AAO etc play a role? Did Indo-Pacific Warm Pool activity play a role? Is there some other poorly-identified factor? Those are the types of questions I explore, as a learning exercise.
@J Christy–
I just want to be sure I understand what you are saying. Are you suggesting that with regard to comparisons between Land temperatures and RSS measurements, we need to scale the change of 0.5C in RSS measurement should be knocked down to 0.5C/ 1.2 ~ 0.417C?
That would make the GISS models overpredict by a larger margin– correct?
Or do you mean the opposite, the RSS 0.5C change over the recent time span should be increased to 0.5* 1.2 = 0.6 C?
109 J Christy
What is the actual physical reason for the warming of the upper troposphere produced by the models?
Do we know?
#115 To compare global-average, model-produced (or measured) surface (land and ocean) temperature TRENDS with observations of tropospheric temperature TRENDS (whether UAH or RSS or radiosonde tropospheric temperatures), one must either multiply the surface by 1.2 or divide the troposphere by 1.2. This is the modeled difference between surface and tropospheric trends if greenhouse forcing is responsible for the surface warming. Example, if a model indicates there should be a warming of the global average surface over the past 25 years due to the greenhouse effect is 0.20 C/decade, then the troposphere (model) shows a warming of 0.24 C/decade.
This 1.2 factor is for the global average only. For the tropical average it is 1.4. For polar areas it is less.
This is the so-called negative lapse rate feedback effect as models force the atmosphere to warm in response to a warmer surface in a fairly rigid vertical relationship (which evidently the real world doesn’t recognize.) This assumes all of the greenhouse factors (clouds, water vapor etc.) respond in a consistent, positive way to an enhancement of the small CO2 warming effect. The negative lapse-rate feedback is observed in short-term fluctuations, but not in long-term, subtle variations.
I will not have time to follow up with further responses.
Re # 117 J Christy
Clarification.
The aim of my brief comment in # 110 was to encourage observations to be reported in actual physical terms. The suggestion was –
If your objective is to make a scientific point, please use the fundamental data, elaborated as needed.
But, if you want to commend or condemn a model, you might choose to say that your observations are to be read with that objective in mind.
Non-specialists can get the two objectives confused.
Based on what John Christy says, we should conclude the surface warming records from GISS, Hadley and the NOAA are diverging even more than we thought from the satellite atmosphere records.
Re #117
So first we are told because the models predict a factor of 1.2 for tropospheric trends vs surface trends we should allow for this when comparing RSS (UAH) measurements with surface measurements (i.e. the models are right).
However, when asked to explain the mechanism Christy says the models are wrong on this: “which evidently the real world doesn’t recognize”!
Just to clarify Dr. Christy’s citation in post #109 (here)
the complete reference is:
Randall, R. M., and B. M. Herman (2008), Using limited time period trends as a means to determine attribution of discrepancies in microwave sounding unit-derived tropospheric temperature time series, J. Geophys. Res., doi:10.1029/2007JD008864, in press.
accepted 10 December 2007)
It’s still in press but no doubt will be officially published shortly.
Re #120
I think the point is that when testing Hansen’s model, he reported surface temperature trend predictions, whereas his model would have predicted tropospheric temperature trends 20% higher. However, if Hansen did not specifically report the increase in expected trend in the troposphere (and for all I know, he may have) or it wasn’t in his model at the time, then this point is moot.
John Christy observed below that “one must either multiply the surface by 1.2 or divide the troposphere by 1.2″ to gave apples and apples. I will re-plot the graphic once I clarify a point on this and I’ve sent him an email on this. The standard deviations for annual values for the various series for the period 1979-2006 are shown below:
# tlt3.glb giss.glb msu.glb
#0.2076278 0.1710205 0.1835279
Left to my own devices, if re-scaling were necessary, I’d have applied this info. John, what is the basis for the 1.2 factor?
It’s a rather absurd argument, ‘the models predict that the ratio of tropospheric to surface trends should be 1.2 whereas it’s actually ~1, so if you want to compare model results for the surface T with tropospheric measurements you should scale by the fictitious ratio’!
Phil @124,
My inclination would generally be to compare as close to like as like. So, I’d compare predicted surface temperatures averaged over land only to predicted surface temperature averaged over land only. I’d compare surface temperatures avearged over land/ocean to measured over land/ocean and so on.
I’m assuming the difficulty is the models didn’t publish time series for troposphere temperatures and people want to compare something?
Re #125
Me too.
OK so why not compare with the surface record? If you want to compare with the tropospheric data (RSS) then why not use the measured anomaly since the measured surface and tropospheric data appear to track each other reasonably well? Which is exactly what’s been done here.
It’s only absurd if you don’t understand it.
Re #127
Larry you have that backwards, it’s absurd when you do understand it. Since you claim to understand the argument would you care to explain it to us?
#125 yes, comparison must be like to like. and, #124, christy is making the perfectly rational argument that tropsphere-to-troposphere would be better than surface-to-surface because the GHG response is predicted to be warmest in troposphere. tropsphere-to-troposphere would give you the strongest, fairest test of the hypothesis.
whether christy’s suggested adjustment is appropriate, i will leave others to judge.
128, Go read Christy’s paper. The link is in 111.
Ref 124 Phil said
Warming by GHG should start in the Troposphere where the GHGs reduce (contain) outbound radiation. So to compare Hansen’s projections to troposhere temperatures you would have to use his troposphere temperature projections (1.2 times surface temperature projections). The ratio by be fictitious, but that is what the models were predicting and the point is to test the model’s predictions, right?
Phil, what would you advise as an alternative to bring the predicted and observed to a common basis in the troposphere?
Re #132
We don’t have a predicted troposphere in this case so compare what was predicted with the measurements, don’t create a fictitious
prediction using an ad hoc correction factor.
Wish there was an edit feature. For the graph shown with with ABC and GISS surface, dividing RSS by 1.2 would show the divergence. Personally I would use two graphs, one for surface temperature and one for Troposhere with AB and C adjusted and noted.
just my two cents.
#133 How about re-running the 1988 model and extracting the tropospheric output from it? You would be ok with that?
Re #135
With one proviso, if I knew that I was going to be comparing with an MSU tropospheric product I would calculate a value that came from a similar profile. I.e. use weighting functions such as the following for TLT:
http://www.remss.com/msu/msu_data_description.html#msu_weighting_functions
When I worked with modellers we often found that when comparing our computed and measured result that it was best to generate an equivalent quantity from the computation rather than try to adjust the experiment.
#136 I think if you wanted to publish the result one should do it the way you suggest, Phil. i.e. I think your concerns are valid. However, if one simply wanted to write a grant proposal to get funding to test the hypothesis, I would suggest that doing a christy-esque post-hoc data adjustment would be ok. In that context, what kind of data adjustment would you advise? I know you don’t like the idea of “creating a fictitious prediction using an ad hoc correction factor” – but that’s what a hypothesis is – a fictitious prediction. Suppose there’s 400K funding on the line. How would you generate that hypothesis using an ad hoc correction factor?
Re #137
I’d certainly want to verify that the Hansen 88 model did predict a more rapid trend than for the surface and try to determine an appropriate ‘correction factor’. The term tropospheric needs to be nailed down too since the plan is to compare with a result that incorporates data mostly from the near surface (0-2km) with some contributions from ~12km. The problem is that in Christy’s 2007 paper the difference between the two MSU products was huge (RSS was ~3x UAH). The RSS value for TLT actually overlapped with the models. From their paper you would conclude that the RSS TLT product trend was the same as the surface trend so I think I just talked myself out of a 400k grant!
What’s the observed ratio? Duh.
You sure did. If the deadline is in four weeks you don’t have time to do everything you just recommended.
If today’s models suggest a trend difference of 1.2, I would think it would be ok for a quick and dirty analysis to go with that. As long as you explain in the graph exactly what you did.
David Smith posted this comparision of global TLT versus tropical TLT that I found interesting. The global trend difference versus surface should be 1.2 times and the tropical trend versus surface temperatures should be 1.8 times based on the models.
#141 figure legend?
The dark blue is tropical and the pink is global.
re 143. the tropical should always be done in flamigo pink. Just sayin
131 captd
That’s an interesting statement. Where does it come from? Do you have a reference? (I ask because I’m trying to figure out what is going on physically, not to dispute it).
ref 45 Pat Keating, Sorry I can’t recall the paper, but it was a discussion on the GCM’s. This one discusses the IPCC predictions. J. Christy would be the better person to discuss this point. As I understand it, warming due to GHG should be most noticeable in the tropical Troposphere because there is more water vapor to warm. If the models are correct, the tropical Troposphere warming should be about 80% more than tropical surface warming tapering to virtually no difference between surface and tropo at the poles.
146
Thanks for the response, capt.
I’m still trying to nail it down. John Christy gave a partial answer in #117, but it’s too general to indicate a specific mechanism.
RSS data has been updated with Jan 08 data: ftp://ftp.ssmi.com/msu/monthly_time_series/rss_monthly_msu_amsu_channel_tlt_anomalies_land_and_ocean_v03_1.txt
Coldest month for the planet in 8 years. 2nd coldest January in the last 15 years. Both hemispheres (scroll over all the way to the right), posted negative temp anomolies. 1st time that’s happened in 8 years.
Steve McIntyre says:
Did you ever clarify this point?
Somehow the whole point is being missed.
Global CO2 emissions continue to increase and the Hansen “A” and “B” projections both fit this emissions increase.
None of the actual temperature data whether corrected or not reflect this continued increase in global emissions.
This is not consistant with a direct causal relationship between CO2 emissions and global temperature and by normal scientific practice this would falsefy the AGW hypothesis.
Doesn’t science adhere to the principle that “even if there is only one observation that is not explained by a hypothesis the hypothesis must either be dropped or modified to fully explain all observations”?
It is not the amount of temperture difference between the projections and the data sets that is important; it is the fact that the two are not moving in the same direction, and the observed temperatures are not moving in a direction that is consistant with the increase in emissions from fossil fuels.
2 Trackbacks
[…] January 2008, I discussed here and here how Hansen’s projections compared against the most recent RSS and MSU data, noting a downtick […]
[…] set, which was later adjusted, slightly reducing the downtick in observations. On Jan 23, 2008, I updated the graphic comparing Hansen projections using the revised RSS version. Today I re-visited this data, posting a […]