The blog world is jump starting discussion of 2008 annual temperatures. Yesterday at 1:56 pm Eastern, NASA employee Gavin Schmidt and climate modeler, purely in his “private” capacity, posted an article arguing that the results were consistent with climate models – an activity that lesser minds might think relates to his employment.
Lucia commented here, perhaps redundantly, that Schmidt’s comment was “tendentious twaddle”.
As a result of Ross McKitrick’s “T3” concept (discussed at CA here), I’ve been following tropical [tropospheric] temperatures from time to time, as these are supposed to be “fingerprints” of AGW and decided that it would be timely to update tropical (20S-20N) estimates, using reported satellite data and collating tropical averages from gridded CRU, GISS and NOAA versions as these are not reported separately. I calculated results over the satellite period (1979-2008) for consistency (as I’ve been doing in the past.) In each case, I zero-ed the data over the 1979-1997 period – the choice doesn’t matter much – but I didn’t want to include the most recent period in the normalization period and I wanted a relatively long period. I also divided the satellite temperatures by 1.2 according to a factor sent to me by John Christy. This yielded the following:
Figure 1: Annual Tropical Temperatures. (Centered on 1979-1997)
We now have a 30-year period of satellite records. Within that period, 2008 ranked 26th out of 30 (5th coldest), 23rd for RSS, 16th for CRU and 15th for NOAA and GISS. Tropical temperatures in 2008 were lower in 4 of 5 indices than in 1988, the year of Hansen’s famous testimony.
Although climate scientists downplayed individual cold months early in the year, Schmidt seized on recent Oct and Nov data from GISS:
In GISTEMP both October and November came in quite warm (0.58ºC), the former edging up slightly on last month’s estimate as more data came in.
Figure 2: Monthly Tropical Temperatures. (Centered on 1979-1997)
In the tropics, temperatures in October and November have rebounded from lows early in the year, but GISS is running a bit warmer than the others (about 0.2 deg C relative to the 1979-1997 average).
When I squint at the tropical data, it has two quite different aspects depending on your perspective. On the one hand, you can look at it and conclude that the average level of the cycles is moving up, with the 2008 low being higher than prior lows, with the overall package moving up in an disorderly way. On the other hand, you can look at the data and surmise that it has consistent with 1/f noise or 1/f noise with a very minor trend. It’s hard to believe that this is the sort of result that Gavin Schmidt “wants”.
For people that look at data, it is obvious that the data (in each incarnation) is highly autocorrelated; the degrees of freedom in autocorrelated data can be much lower than people think and, accordingly, the confidence intervals can be surprisingly wide (a point made in Santer et al 2008, though there are some defects in their analysis that we discussed before).
Wide confidence intervals are a two-edged sword. For people who attempt to argue that this data shows no trend, the wide confidence intervals permit a trend much higher than observations (and perhaps high enough to be consistent with the less aggressive models). For modelers who argue (a la Santer) that the confidence intervals are wide (and wide enough so that observations are not “inconsistent” in some sense with models), the problem is that such wide confidence intervals will probably include the null hypothesis of no trend or a minimal trend. It’s interesting to watch, but I don’t think that you can conclude very much either way purely on the statistics of the data, as people so often try to do.
Schmidt observed:
The bottom line: In the GISTEMP, HadCRU and NCDC analyses D-N 2008 were at 0.43, 0.42 and 0.47ºC above the 1951-1980 baseline (respectively). In GISTEMP both October and November came in quite warm (0.58ºC), the former edging up slightly on last month’s estimate as more data came in. This puts 2008 at #9 (or #8) in the yearly rankings, but given the uncertainty in the estimates, the real ranking could be anywhere between #6 or #15
Climate Audit 
175 Comments
Willis, if you’re watching, notice that Schmidt aligns the starting point of the models in his starting year. I recall some sort of dispute a couple of years ago on you setting 1958 to 0 – a procedure that looks reminiscent of what Schmidt is doing here. Am I remembering correctly?
Re: Steve McIntyre (#1), you are correct that in Mann’s Senate testimony he had aligned the 1958 starting points of his “A, B, and C” GCM scenarios to a different point than the 1958 start of the relevant instrumental record.
Their argument was that they had run the GCM for 100 years using the 1958 conditions, so they simply started their data at the point where the 100 year run ended.
I thought the argument was bogus, and that if you are comparing GCM and actual trends, you should start them at the same point.
w.
Re: Willis Eschenbach (#25),
Willis, do you mean Hansen’s Senate Testimony?
Re: Mike B (#26), you are correct, Hansen’s testimony …
Re: Boris (#27), my bad, moving too fast. You are correct that they started it at the average 1958 value for the hundred year run, not the final value.
Re: Boris (#37) (and Deep Climate), you say:
Absolutely not. What I think you meant to say is that some of the models predict a tropical tropospheric hotspot from any warming, which is a very different thing. As Mark T. pointed out above, the IPCC doesn’t seem to agree with you either.
How can the surface heat up without heating the troposphere much?
Well, I live in the tropics by the ocean. Every day, I watch megajoules of energy being sucked up at the base of the thunderstorms and whisked up to near the tropopause, from where they slowly descend around the thunderstorm. This insulates them from most of the troposphere, since they bypass it without interaction either by radiation or mixing.
When the tropics heats up, more energy is moved by the thunderstorms … but that does not warm most of the troposphere, because it is being bypassed. In addition, the heat is stripped out of the rising air by the condensation and rain. So the descending air, which does mix throughout the troposphere, is generally cool. This, of course, cools the troposphere somewhat as it descends.
So yes, it is definitely possible for there to be tropical surface warming without much tropospheric warming … but as you point out, you won’t find it in the models …
w.
I don’t know if this is relevent to a discussion on the world’s temperature feeds, but I’ve just had this thrust at me from another blog:
I looked at it, and did not recognise the RSS feed at all. Does anyone know where this has come from, and what they have done with it?
For model-data comparisons see the GCM forecasts from IPCC Figure 10.7. As of the 2011-2030 interval (leftmost column) the troposphere over the tropics looks like it’s projected to be about 1.5 oC warmer than the average temperature over the 1980 to 1999 interval (using the A1B emissions scenario). Comparing interval midpoints (1990, 2020) this implies we should be seeing a current average warming of 0.5 oC per decade in the mid-troposphere, and the IPCC text (pp. 764-765) says this change reflects a rapid response of that area of the atmosphere to the forcing change, and should be observed even in the early-century time period. So we’re well into the time interval win which a strong upward trend should be visible. Also bear in mind that the CO2 content of the atmosphere rose about 14% over the 1979-2008 interval.
Steve, can we see the mid-troposphere RSS and UAH data too? (Or more to the point can everyone else, since I already have)
A couple of quick comments:
And what was the result when the CORRECT data was put in?
Second:
It appears that they recentered the zero (baseline), and since all three use different baselines, why are they continuing to use 28-year old baselines? If they were going to recenter anyway, use one that ends at 2000 (as I believe NCDC uses.)
Yes, I realize the TREND won’t change, but they’re still trying to show large anomolies, that is, values above “zero” or “normal” (strictly for presentation purposes.)
Huh? Why do you think tropical temps are a fingerprint for AGW?
Re: Boris (#5),
Because that’s what the models predict.
Re: Not sure (#10),
It’s Physical Meteorology 101. In the all other things being equal case (please don’t bother replying that all other things may not be equal, everybody is, or should be, well aware of that), relative humidity remains constant with temperature, so if temperature increases, specific humidity increases. In the tropics, the limiting lapse rate (the rate of change in temperature with altitude) is determined by the moist adiabatic rate, which decreases as specific humidity increases. A decrease in the lapse rate means the temperature at altitude increases faster than the surface. QED.
Re: Boris (#5)
The tropics receive the greatest insolation, thus any changes in heat trapped by the atmosphere would be most evident there. Furthermore, they have the least natural year-to-year variabilty, thereby making any secular changes more readily apparent. At latitudes above ~40 degrees, the temperatures are increasingly more dependent upon heat transported from the tropics by global circulation and vary much more, due to the chaotic features of that transport.
What never ceases to amaze me is the preoccupation with yearly rankings and with “trends” in records much shorter than the 88-yr Gleissberg cycle. Natural extremes have a strong tendency to cluster because of cyclical influences. I would suggest that we need records at least twice the length of the G. cycle to obtain meaningful estimates of small secular changes in temperature levels–those within a standard deviation of the year-to-year variabilty. The f-inverse noise model, though by no means proven to capture all the features of surface temperature variations, would suggest that the whole simplistic concept of linear trends is plainly inapplicable.
Re: John S. (#11), I observed something like this when the whole temperature record from 1850 was posted up recently. Not saying its definitely solar but enough to make you not discount it. Is it my imagination or is there a positive trend only in the cru, noaa and giss but not in the msu or tlt3?
Re: MC (#15),
MSU is available only since 1979 and is too short a record for meaningful estimates of secular temperature changes, although the rise to 1998 and subsequent decline are plainly evident. The HADCrut3 anomalies from 1850 might be a long-enough series, but the anomalies are piece-mealed together indiscriminately from urban station records and SST observations, using wholly unproven “adjustments.” The series is not a bona fide average of continuous observations at fixed points around the globe. The apparent “trend” of ~0.6K/century is an artifact of their rigorously indefensible method of construction. Urban station records often show strong positive trends due to intensifying UHI effects. SST’s, of course, are not air temperatures and reports from ships of opportunity do not constitute proper time series. In short, it’s a data sausage of suspect ingredients that is being presented as the “global temperature.”
Re: John S. (#21), I know. Data sausage is a good term to use. In reality it looks like the best data, as in the data that can be characterised, is the last 30 years then. The next few years will be interesting.
Re: John S. (#11),
This is an important point/observatoin, as a sort of corollary to the Sampling Theorem. That theorem, interested in “high” frequencies, requires that in order to detect a component w/ frequency F, you have to sample at frequency 2F. The corollary is that in order to have the frequency-domain granularity required to distinguish low frequencies, you must sample for a longer period of time.
If I squint at the satellite data I see an upward trend of about 0.06 C per decade which is perfectly in accordance with the idea that AGW is real, minor, and grossly exaggerated by people like Schmidt.
Like Henry, I wonder why they use 1951-1980 as a baseline. Could it be because these were colder years?
The first thing that jumps out at you when looking at these graphs is that there is no trend and the only way you can find one is to use the data like so much play-dough until it tells you what you want to here. Or do like Tom C and squint really, really hard.
What is truly fascinating about the precipitous drop in global temperatures for 2007 and 2008 is the remarkable correspondence with the declines in solar activity.
You can check the sunspot data for yourselves.
http://www.ngdc.noaa.gov/stp/SOLAR/ftpsunspotnumber.html#american
Steve: snip – no need to editorialize. Also please don’t go a bridge too far in this. The decline in temperatures has been noticeable, but hardly “precipitous” and not out of line with downspikes in other years.
Gavin Schmidt appears to have difficulty counting to ten.
According to his own blog, 2008 was cooler than any year since 2000 (that’s 7 years) and cooler than 1998 (7+1=8 years) so cannot be higher than #9. In fact it is #9 for GISS and #10 for HADCRUT3.
He also omitted to give the global 2008 (Dec-Nov) rankings according to RSS (#14) and UAH (#17). I wonder why? Another puzzle is why when I pointed this out on his blog my comment mysteriously vanished.
As a matter of fact if you look at things historically it is obvious that the tropics are pretty tightly “termostated”, they don’t grow much colder during glaciations and were not much warmer during the previous (oherwise very warm) interglacial. Even back in the Paleogene when climate was warm temperate to the poles the tropics were not dramatically hotter than now.
The GCM’s may have other ideas, but my money is on the historical record.
Are we officially to a full thirty years of satellite data? Because, if true, that should require a shift of baseline using their own criteria.
The ‘instrumental period’ isn’t the full extent of instrumental temperature measurements – just that prior measurement records were considered insufficiently regular, accurate, precise, and disbursed. (Things like monastery logs, ships logs, etc.)
But here we have a collection of surface station measurements… and a collection of satellite measurements. One of which is more regular, more accurate, more precise, and more disbursed.
Shift, then try to find a calibration for individual surface stations to the nearest measurement in the satellite record – if you want to continue using the older data for anything constructive.
The only thing I conclude is that looking at only a few decades of data is ridiculous unless one believes that all possible cyclic influences have come and gone a few times during that period. As we all know that’s not happening there’s little reason to believe the observed trends are something layered on top of all that influences the climate on a long term basis.
When talking about tropical temperatures, you have to note the strong correlation with the ENSO. In this case, the Nino 3.4 region anomaly(lagged three months).
#15 John S.
That’s the first reference I’ve seen an “88-yr Gleissberg cycle”
When I went to learn about it the following popped up:
Exploratory Analysis of Similarities in Solar Cycle Magnetic Phases with Southern Oscillation Index Fluctuations in Eastern Australia
ROBERT G.V. BAKER 1
The abstract states: Using data from 1876 to the present, the exploratory analysis suggests that when the Sun’s South Pole is positive in the Hale Cycle, the likelihood of strongly positive and negative Southern Oscillation Index (SOI) values increase after certain phases in the cyclic ~22 yr solar magnetic field. The SOI is also shown to track the pairing of sunspot cycles in ~88 yr periods. This coupling of odd cycles, 23-15, 21-13 and 19-11, produces an apparently close charting in positive and negative SOI fluctuations for each grouping. This Gleissberg effect is also apparent for the southern hemisphere rainfall anomaly. Over the last decade, the SOI and rainfall fluctuations have been tracking similar values to that recorded in Cycle 15 (1914-1924). This discovery has important implications for future drought predictions in Australia and in countries in the northern and southern hemispheres which have been shown to be influenced by the sunspot cycle. Further, it provides a benchmark for long-term SOI behaviour.
Link at http://www.physorg.com/news147456732.html
I also read someplace that GISS is no longer adjusting for the heat island effect. Can anyone verify that as true or not?
thanks
Ed
What does that factor of 1.2 represent?
It’s too early to call a top in this market but I don’t think I’d be long. Profit taking time.
Chartists are the worst at calling tops – my motto leave the top 20% on the table and v.v
Don’t you just love statistics.
How does that make trop. temps fingerprints for AGW?
Re: Boris (#23),
Ask the IPCC:
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1154157
Re: Not sure (#29),
I see nothing about a fingerprint in that quote. A tropical tropospheric hotspot is expected from any warming.
Re: Boris (#37), Really?
Sounds to me like they’re saying only greenhouse gas emissions should cause such a thing… that whole “no expectation that other climate forcing…” comment and all.
Mark
Re: Boris (#37),
Ah yes, the “nobody ever said tropospheric warming is a result of increases of GHG” axiom.
From Gavin et al
http://www.osti.gov/energycitations/purl.cover.jsp;jsessionid=8700E9E9B8A318E483BE80771E328BC8?purl=/881407-xk2Sdg/
PLEASE – solar people, go to your own thread.
No. They started the data at the average for the 100 year run, to remove as much internal variability as possible. The offset was actually part of the prediction: it was Hansen’s prediction for what the average 1958 would be. When you lined up the scenarios to the real world temperature, you threw out part of the prediction.
Steve M. is right that what Gavin is doing is similar to what you did. His 1990 and 1979 plots give a better idea of where things stand. But in any case, we are still left with the obvious: real world trends are well on the low side of model projections at the moment.
I like the plot of five tropical temperature measurements, but would /love/ to be able to download the actual numbers that were plotted. Is this possible?
I hope so!
Robin
Re: Robinedwards (#28),
http://data.giss.nasa.gov/gistemp/tabledata/ZonAnn.Ts+dSST.txt
http://vortex.nsstc.uah.edu/public/msu/t2lt/tltglhmam_5.2
http://www.remss.com/data/msu/monthly_time_series/
http://hadobs.metoffice.com/hadcrut3/diagnostics/regional/30-30/
The UAH and RMSS addresses will get you 20S-20N and the others close.You may have to dig about a bit.
Remember that the early part of the time series contains two whopping cooling non-climatic volcanic eruptions: El Chichon and Pinatubo
Re: John S. (#11), such suggestion would compel an awfull lot of people to look for real jobs and bankrupt an awfull lot of carbon “assets”.
As Boris suggests, this statement requires amplification and justification.
some clarification.
As an ensemble, the models project amplified warming in the upper troposphere 10 km above the tropics, not in the lower tropsphere or at the surface. Also, the effect is similar for both natural and anthropogenic forcings. However, models do show a difference in the lower stratosphere between natural (e.g. solar) and GHG forcing, so this may be a more suitable “fingerprint.”
Re: Ross McKitrick (#3),
As I read it, the projections show between 1.0 and 1.5 deg in the upper tropical troposphere, implying a rise of between 0.33 deg and 0.5 deg per decade in that area. Moreover, the RSS TMT (mid-tropsphere)analysis is centered on an altitude of 5 km, although the entire troposphere is covered by the weighting function. Here the models project between 0.5 deg and 1.0 deg increase, so the correct figure for model ensemble comparison to TMT is probably around 0.23 deg per decade, which happens to be the trend used by Douglass et al.
Boris and DC want to “Buy an Argument” today. Here is in graphical detail exactly what Not Sure and John S are talking about. Both the lower and upper tropical troposphere is not warming as the models predicted and are actually falling on no change. 2 quatloos to buy the next argument.
Sorry forgot, The stratosphere is cooling because we are at solar minimum and therefor less UV interacting with the stratosphere. UV warms the stratosphere by interacting with the oxygen. I think most know this.
RSS TMT in 2008 is 26th out of 30 and UAH T2 is 27th out of 30. They are 0.16-.22 deg C lower than Hansen’s 1988. There’s no comfort for Gavin Schmidt in the tropical troposphere based on the present data.
I’m not saying that this proves or disproves anything. As noted above, these are highly autocorrelated series and IMO may well be “consistent” with a relative wide range of “true” trends – though, as noted in the Santer thread, even with the autocorrelation, the data is not supportive of the more aggressive models.
Boris
Do you mean the plots that are not displayed in Gavin’s post, which he appears to have created and linked after people commented on the graph with the cherry picked start date, which he created, posted and which he used to compare model and observational results in his blog post? The graphs only called out in an update with an rather unpreposesing text mentioning that he created them and providing no explanation of what he thinks they show or how they relate to his main post?
Yes those, Lucia. Evil Gavin at it again. But telling fairy tales about the CO2 lag is just hunky dorry!
What you fail to grasp is that the models in this case are based on the moist adiabatic rate, which has been confirmed many times over. That’s why it’s in the model, because it is a physical observation. Claiming that models are bad because they include a well grounded physical observation is silly. Claiming that a response to any warming is a fingerprint of anthropogenic warming is twice as silly.
One could always remember that air mixes globally. That thing called wind, which surprising acts differently depending upon altitude at any give spot, say a nice one meter square at at certain lat/log DMS location.
On the other hand, we all know that the urban heat island effect is only local, so no need to fret. It’s only a few tenths of a degree anyway.
“Move along, there’s nothing to see here.” {hides A/C vent exhaust that’s blowing onto a temperature sensor}
Boris–
I know you object to something or other in a particular document that I have ignored, and would never even have skimmed has you not brought the topic up in comments. But if want to criticize something, it would be useful if you could be specific about the facts or ideas you consider incorrect. For some reason, when you are pressed to do this, you decide you are too busy. So, as it stands, while I have never endorsed or even commented on the article you kept objecting to in comments on my blog, I also have no idea what precisely you wish me to find objectionable.
In generalities I have already agreed with you: The article by who-zi-whats (whose name I don’t even remember) seems to present an argument I find less plausible than others. But for some reason, you want me to complain about something more specific.
As for Gavin’s article: He happens to write tendentious twaddle on topics that overlap what I am specifically looking at. So, yes, I will comment on his stuff.
I could have said more about Gavin’s article. For example: If you read the first few paragraphs, he informs us that he will be presenting an rebuttal to an argument that he imagines some unamed person will make sometime in the future. That is: He creates his own strawman to counter argue.
But I skipped most of that and focused on the cherry picked figure
Steve M: Sorry if this is too off topic.
The tropical upper troposphere is not homogeneous. Some regions (Western Pacific, for example) are composed of the tops of thunderstorms, cloud, etc, and are driven by convection from below. Other regions (the vast Eastern Pacific, for example) are where clear air radiates, cools and sinks, and are driven by that radiative cooling.
I imagine that the relative temperatures of these regions vary with varying types of forcing (CO2, solar, ocean, etc). Forcing from increasing CO2 may affect the temperature of the Eastern Pacific upper troposphere differently than, say, forcing from a warming ocean or strengthening sun.
If the models and satellite data allow sufficient regional resolution then regional comparisons, not global, may be a better way to search for any CO2 fingerprint.
Re: David Smith (#43),
Thank you and John for your attempts to understand our climate and sharing it with us. Please keep up the good work.
As background to this discussion, readers of this thread may want to read the paper “Limits on CO2 Climate Forcing from Recent Temperature Data of Earth” by John Christy and myself which will be published in January.
Variations in the tropical temperatures were shown to be correlated to ENSO. One of our observations was that the temperature anomalies in 2008 are less than they were in 1979. The paper can be downloaded at http://arxiv.org/ftp/arxiv/papers/0809/0809.0581.pdf.
Steve, can I re-post my temperature reconstruction results for the tropics. The residual warming trend drops to 0.045C per decade from 0.13C per decade in the raw data.
#41
Boris appears to be right; the tropical tropospheric hotspot does not distinguish between types of forcings, at least not in the GISS model. For example, here is the GISS model output for marked solar forcing, clearly showing the tropical hotspot.
Compare this to the model output of the equivalent GHG (2 x CO2 equivalent) forcing (also note the lower stratospheric cooling which is different):
I understand that to mean that the actual observed and estimated natural forcings in the last 50-100 years were considered insufficient to cause the warming, not that a natural forcing of equivalent magnitude (if one had occurred) could or should not cause such warming. Now I admit I don’t know the details of the other model outputs under the assumption of large natural forcings; however, Boris’s assertions make eminent sense and are backed up by the above graphs.
Re: Deep Climate (#47),
These comments tend to derail the important points of the discussion. The troposphere to surface temperature trend ratios coming out of the climate models are primarily and necessarily the result of GHG forcing.
Therefore, while one can show model runs as Gavin Schmidt did at RC a few months ago that give similar surface to troposphere gradients (except for the stratosphere cooling) for solar and GHG forcings, the important points to be debated are those from Douglass et al. (2007) and Santer et al. (2008). Analysis and discussions of those papers have been made rather well and completely, in my judgment, in other threads here at CA.
It is also important to keep in mind the point that Steve M makes in the introduction to this thread that showing the uncertainty in the model and observational results in order to avoid a judgment on the models getting the surface to troposphere wrong can also put a zero temperature trend into play and most certainly showed that model results have uncertainties that might question the practical use to which these results can be put.
Santer et al. in attempting to show new radio sonde adjustments for tropical troposphere measurements (favorable to the authors point) ended in my judgment showing how much those “adjusted” readings can be varied over the range of the troposphere and why Santer would only make qualitative statements about a certain part of the range for these adjustments.
A couple of questions about data derivation:
Is this area weighted (it is too bad these zonal numbers aren’t avaialable)?
Could you explain this factor please?
Thanks.
What is the point of attempting to compare Global Temp ranks (like 2008 is the 9th hottest on record) with tropical temp ranks (…2008 ranked 6th out of 30…)
It sounds like you doing exactly what RealClimate was suggesing was bad.
“Within that period, 2008 ranked 26th out of 30 (5th coldest), 23rd for RSS, 16th for CRU and 15th for NOAA and GISS.”
Shouldn’t that statement indicate that you are discussing tropical temps only?
On the question of whether a tropical troposphere hotspot would emerge under other forcings, here’s a simple test. Look at IPCC Figure 9.1, which they have conveniently (for our purposes) posted without a Figure caption. This diagram is a backcast run from the Parallel Climate Model inputting observed 20th century forcing changes (1890-1999) one at a time to show their individual effects on the atmospheric layers, assuming the world works the way the PCM is set-up. (Probably has the moist adiabatic lapse rate and lots of other great stuff.) The bottom right panel shows the SUM of all forcings. Now try to spot the GHG-only panel and the SOLAR-only panel. Here’s a hint for finding the GHG one: hum to yourself the old Sesame Street song, “One of these things is not like the others…” The solar one: lotsa luck. I don’t know where the graph came from that DC used in #46, maybe it’s a simulation of the climatic response to the sun going nova. But in response to the small solar changes we have actually observed over the 20th century, the IPCC Figure 9.1 says you’d have to squint to see any tropical amplification.
It’s a bit of an irony. When it comes to explaining surface trends the warmers insist that the sun is a big nothing, and they are quick to say that only GHG’s could do it. But when attention turns to the conspicuous prediction of a tropical hotspot as also arising from GHG’s, all of a sudden the warmers say the sun can boil the sky just as much as GHG’s so we daren’t single out GHG’s. The IPCC Figure 9.1 shows that models are parameterized such that GHG-induced changes are huge in comparison to other forcing responses. That’s convenient for the purpose of saying surface warming must have been due to GHG’s. But it creates a problem when that same model tuning yields such a strong, unique prediction for a large region of the atmosphere that isn’t showing the predicted change. They can’t now go back and disavow the GHG-induced tropical hotspot without disavowing the current GCM structures with it.
Re: Ross McKitrick (#50),
Ross:
There seems to be a problem of understanding our language going on here. I found it interesting that people who quote IPCC have selective forgetfulness when any one finds differences in projections and reality. they claim that you can’t prove IPCC said that even after oyu give them the page number. Go figure!
That quote in 29 is from McKitrick, not the IPCC. People can go to page 764-765 and read what the IPCC actually says:
Click to access AR4WG1_Print_Ch10.pdf
The reason why you have to “squint” to see any amplification due to solar in figure 9.1 is because the Sun didn’t do much of anything over the 20th century, in stark contrast to GHG. Little change vs big change. The images from 46 compare a 2% increase in solar irradiance to an increase to 550 ppm of CO2 (doubling preindustrial) — examples chosen because they cause about the same amount of warming. Tropical troposphere warming is predicted regardless of the cause of warming, and is not a “fingerprint” of AGW.
#46 #50
People who want to know the answer to Ross’s little quiz can find it on page 675 of AR4 chapter 9. Ross wonders where the figure posted in #46 came from; suffice it to say that it came from realclimate and that it bears no resemblance to the IPCC’s figure. The perfect match of the dark red patch of the two pictures in #46 – coming from two completely different physical mechanisms – is quite comical, and a tribute to the art of Gavin and his team (I wonder how many times they had to tweak the parameters in the model and the contour levels of the plot to get this?)
Perhaps deepclimate and Boris could explain how these two very different heating mechanisms, one based on shortwave visible radiation that is not absorbed by the atmosphere, and one based on atmospheric absorption of long-wave back radiation, should give almost identical atmospheric temperature profiles? If not, maybe they should stop attempting to propagate this myth.
Annual data is noisy, if you want to compare values decadally, use Mann’s
[smoot,icb,ice,mse0]=lowpassmin(data,1/10);
UC–
Can’t you find a Mannian smooth filter that shows the “right” answer? (That would be monotonically increasing for both sets.) 🙂
Re: lucia (#59),
No, none of the lowpass.m options will do that, and thus even Mann08 method cannot dodge this decadal cooling 😉 And if 2009 (HadCRU ) is less than 0.45, same thing. Then they’ll need to invent a new smoothing method.
One point about the ENSO’s impact on tropical and global temperatures (and perhaps david douglas is interested in this) is the differing temperature response from the 1982-83 El Nino and the 1997-98 El Nino.
Both were “super El Ninos” of basically the same magnitude and duration, yet tropical (and global) temperatures increased by 70% to 80% more in the 1997-98 El Nino.
In fact, the temperature response to the 1997-98 El Nino is generally about twice what should have been expected from the Nino alone.
The difference is that the AMO index also spiked hard in the 1997-98 El Nino but did not in the 1982-83 El Nino.
There was a similar response to the other super El Nino in the last 150 years, the 1878-88 El Nino in which global temperatures spiked 0.7C above the background trend at the time. The AMO spiked again similar to 1997-98.
Generally, I think we need to look at both the ENSO and the AMO when trying to explain these temperature changes. When you pull out the influence of these natural climate cycles, the warming signal from GHGs falls off to very low levels.
Re: Bill Illis (#61),
Bill, I think it is important to keep in mind that the point of contention is the ratio of the surface to troposphere temperature trends in the tropics. As I recall, those ratios are not that much affected by El Nino events, but, if you have some data on those ratios to the contrary, I would be most interested in seeing it.
Re: Ross McKitrick (#63), Re: lucia (#64),
Good posts for keeping the discussion on track.
Sorry, that should say the 1877-78 super El Nino.
#55: I have no reason to dispute that increased solar activity would generate similar effects in the troposphere as increased GHG’s: see also Bengtsson’s paper. But that only means that if we had lived through a few decades in which GHG’s went up and the sun sharply intensified and we observed a hotspot we might have trouble deciding what caused the hotspot. But in the context of the model backcasts and the available observations we do have a unique causal chain embedded in the hypothesis. Read your own words carefully. Solar would look like GHG if solar had done much. But solar didn’t do much, while GHG did go way up. So we expect to see a big hotspot by now, solely due to GHG’s (a la IPCC Fig 9.1). The presence of a hotspot would (according to the model scheme) implicate GHG’s. The absence of a hotspot means none of them had much effect. It’s no good appealing to a potential similarity in atmospheric changes under hypothetical historical forcings that we didn’t actually experience. It’s like the defence attorney saying that if, hypothetically, the police had found two sets of fingerprints on the gun, they couldn’t implicate your client. True enough–but there was only one set of fingerprints.
And as for the stratosphere, don’t be fooled by the trend line. There are 2 major volcanic events that coincide with steps down in the average temperature. Between them there is no trend, and from 1994 to the present there is no trend. I don’t know what this means, but any discussion of the stratosphere needs to start from the observation that there is no recent trend, at least in the last 15 years.
Ken–
The other difficulty is that, while Gavin showed runs where the hotspot appears with solar forcing, the IPCC AR4 tells the world otherwise.
While section 9.2.2 of the WGI-AR4 may be incorrect or distorted, and the GISS model or some other may be correct that the hotspot occurs with solar forcing as well, way back in 2007, the authors of the WG!-AR4 felt inclined to tell the world that the hotspot is a fingerprint of global warming.
Ross–
Of course, your argument is correct. If, as suggested by RC and Bngtonsson’s paper, a similar effect is expected with solar forcing, the absence of a hot spot would tend to suggest deficiencies in model’s ability to simulate climate features. So, the odd thing about the argument that we’d expect it with solar forcing also is that the modelers still need that hot spot to appear. Currently, their best spin is that it may be there but the data are not sufficiently accurate to reveal the feature. This makes the theory of the hot spot untestable using currently available data.
Crowley has an interesting paleoclimate paper in the 1990s (in a little publicized conference proceeding) that I’ve meant to post on in which he wonders about paleoclimate evidence which indicated to him that lapse rates in the LGM were different than at present, notwithstanding the seemingly firm belief embedded in climate models (based on strong opinions) that the lapse rate is an iron law of nature; Crowley wondered whether there shouldn’t be more play in the lapse rate.
In mathematical terms, if you permit some variability in the lapse rate, it seems to me that you could develop quite different surface scenarios. It also seems to me that something that might be possible mathematically would have to actually occur to some degree physically if a “skeptic” argument were to succeed.
I wish that IPCC had provided a proper exposition of this issue and would very much appreciate a reference to something that does provide a proper exposition (one that rises above assertion.)
Re 55
A 2% change in solar output? That would be about the amount of change expected over a 200 million year period. Somehow this doesn’t strike me as a very realistic scenario. So that hotspot would seem to be a GHG-exclusive fingerprint.
I respectfully and partially disagree with Ross and Lucia. Bengtsson can only get similar pictures from solar and GHG by assuming that the dominant affect is positive water vapour feedback. (The primary heating is very different, see his fig 1). By this argument, any warming gives a hot spot somewhere up in the atmosphere. This is not observed, despite recent warming, therefore it’s fundamentally wrong.
#66 tty 2% change in solar output is 27w/m2. Way to large for the sun. I think the most it has varied in the last 200 years is .35 w/m2. You can see that a 2% change is way beyond standard deviation.
#67,68: That’s why I conjectured in #51 that it it might be a simulation of the sun going nova. You can probably get a tropical hotspot in a GCM in lots of ways if you’re willing to play those kinds of games. But what counts is whether the models tell us the observed changes over the last few years could have done it. The IPCC report says: GHG’s yes, sun no. Hence the absence of a hotspot is a problem for the IPCC model schemes since it is inconsistent with the GHG sensitivity. The 2 themes emerging from the Team are: (i) maybe the model schemes are so uncertain that we don’t necessarily expect a hotspot after all (see Santer et al); (ii) the data must be wrong (see CCSP 2006). But the catch with (i) is that the models that fit the data have very low GHG sensitivity, so they don’t really want that one to be true. (ii) may be true but it came at the end of a 6-chapter report in which they failed to find any problems with the data beyond those corrected as part of the report-writing process, so it’s pretty groundless.
And don’t forget how the IPCC embraced the new Wang et al. solar reconstruction, that disputes whether there was even any solar intensification over the 20th century, so even if they dial up the solar sensitivity in the models, they dialled down the reconstructed irradiance trend. It is more than a little ironic for realclimate now to be toying with the effects of a 2% irradiance increase when the IPCC Summary highlighted the Wang et al. reconstruction that says it was only 0.04% since the Maunder Minimum.
#55
Yes, cce is right. The point of the graphs in 46 was to show the pattern produced by similar magnitudes of forcing that would result in the same level of overall warming.
As cce and I have both noted, the pattern of warming described in the IPCC and displayed IPCC Figure 9.1 (a) is based on the observation and estimate of very slight positive forcing from solar irradiance in the 20th century. But even here you can clearly see a proportionally enhanced warming (again slight) in the mid to upper troposphere above the tropics.
A careful reading of Chapter 9 makes clear that the pattern of atmospheric warming is similar whether driven by either GHG or solar forcing, except of course that GHG forcing results in stratospheric cooling. That’s the true “fingerprint” that distinguishes GHG warming from solar.
http://ipcc-wg1.ucar.edu/wg1/Report/AR4WG1_Print_Ch09.pdf
The text of pages 674-5 states:
The executive summary states:
Nowhere in chapter 9 can I find support for the supposed “fingerprint” of differing tropospheric warming patterns under equivalent solar or greenhouse gas forcings. To the contrary, the “fingerprint” of AGW is the combination or “pattern” of tropospheric warming and stratospheric cooling.
71: But you are only considering the “in principle” pattern. You have to look at the patterns in response to observed historical forcing changes. If when you look at the Figure in Lucia’s note (#64) you can’t see a difference between panel (a) and panel (c) then you need to look more carefully. And surely you can see that panel (c) determines the summation pattern in panel (f). So if panel (f) is the expected pattern and the observations show no such hotspot, which underlying response scheme is thereby called into question? Answer: panel (c), since panel (a) didn’t produce the prediction.
71:
Actually the forcing represented in panel (d) also produces stratospheric cooling with warming in the troposphere. Panel (d) forcing is change in Ozone.
Re: GMeurer (#73),
And ozone was affected by CFC’s, which have, I believe, declined in recent years. Does the model somehow tie in to a level of CFC’s (steady, increase, decrease)?
#70 Look at it this way. The sun does vary by 90w/m2 every year (orbital variation) but we don’t see this “Hot Spot”. The Stratosphere cools and warms to the solar cycle and the when you have consecutive large to smaller cycles you will have stratospheric cooling.
Click to access TSI-LEIF.pdf
I just reread Gavins post and I must agree with him when he says there is no statistical significance in the records since since AR4 to prove that the climate is changing in any way. Thereforee I agree with Boris,DC, and cce that we do not know what the climate will be as there is no value in the GCM,s as they stand. At least that is the way I read their posts.
Hehe, it would be the “denier’s” dream if it were truly that easy. Of course, it would also be the skeptics dream if the burden of proof that GCMs are valid were to actually lie with those that claim they are, but alas, such a dream world does not exist.
Mark
Re 68
I said 200 MILLION years (200,000,000), because at the rate the suns brightness is changing, that is the time it will take for the brightness to increase 2 %.
OK DeepClimate, let’s see if I have this straight: The AGW “fingerprint” is composed of two parts, part 1 is the Tropospheric heating, and part 2 is the stratospheric cooling. So AGW-fingerprint = part-1 + part-2. But part-1 is false, so part-2 (or even additional parts) don’t matter because without part-1 AGW-fingerprint == false every time. All this talk about what “could” cause part-1 to be true is a red herring because it is NOT. So Steve is correct and all your objects are mere noise.
I read gavin’s statement where he stated that the 2008 average could be anything from the 6th to the 15th warmest which would meen no stastical significance Therefore no climate change. I read the posts by the three to mean that they disagreed with statments made by IPCC AR4. I take that to mean that they feel that you should not rely on projections or statements made by the IPCC and I agree.
#79. One of the editorial policies of this blog is that I’m not interested in people trying to debate all climate theory in one paragraph bites. Otherwise every single thread quickly develops into the same argument.
The pattern of stratospheric cooling and tropospheric warming was certainly projected by Ramanathan in the 1970s and I consider it to be evidence of CO2 impact.
The devil is in the details. In my opinion, a “skeptic” theory would have to demonstrate that there is potential elasticity in the lapse rate, such that one could have enhanced warming in the upper troposphere with lesser warming in the boundary layer that we inhabit. From superficial knowledge, the models seem to embed the concept of a fixed lapse rate in one way or another. An assumption which might even be true, but it would be nice to see it better argued.
I strongly believe that understanding can best be developed by finding solid ground and working from that. That’s why this blog tries to focus on details and specifics and discourages the attempt to argue everything in climate in one paragraph bites.
#81
I read Steve’s postings precisely for the details and specifics, as I am able to understand them. To the extent that others augment Steve’s postings, I look forward to those contributions and subsequent comments. Anecdote, while colorful, detracts from his (and his collaborators) singular efforts, which I find nowhere else in climate conversation.
Thanks, Steve.
In a fortnight or so I will have completed a draft paper which looks inter alia at rural undisturbed tropical sites in Australia (also some sub-tropical). Of the various sites I have graphed to date, only one (Cobar, NSW, sub-tropical) shows the decade 2000 hump, then mainly in the raw Tmax more than raw Tmin.
In the last 40 years, it is emerging that the annual averaged data I have calculated are montotonic, with a slight rise,though some show a slight fall. I don’t have the grand mean yet for slope.
Any theory that accepts the decade 2000 hump has to explain its comparative absence here. The first guess is that the hump is an artefact of processing urban dta and possibly a failure to process out UHI. But didn’t John Daley and Warwick Hughes write similarly some time ago?
Anyone wish to proof & comment with me?
Re: Geoff Sherrington (#84), Geoff: I would be happy to look at what you have done. cloehle at ncasi.org
Deepclimate
If so, the point of the graphs in 46 to introduce a red herring. Those graphs are a) not using the term fingerprint of global warming the way the IPCC uses it b) are based on fictional forcings.
It may be an interesting exercise to think about what would happen if the sun did something it did not and possibly cannot do. But if we restrict ourselves to reality, the hot spot is anticipated to occur only in the presense of GHGs. The IPCC uses that term and terms like patterns or signature to distiguish between effects that arise from GHG as opposed to responses to other changes in forcing that actually occur on earth.
Arg, I hate finding typos later when I reread my own postings. I meant to say “objections” instead of “objects”. Sorry for any confusion.
Re: Paul Penrose (#86),
Me too. Please correct to “Daly” and replace “monotonic” with “near linear” in my #84.
The statement is that any source of heating will give a tropical tropospheric hot spot. This is then illustrated by one example using something that is extremely unlikely to actually happen. One is not any. Why not use as a model input a forcing that has been proposed by a reputable climate scientist, Roger Pielke, Sr., who maintains it may be as significant an effect on the GMST as ghg forcing: land use/land cover changes.
Wont the lapse rate vary with dewpoint? The energy budget also includes latent heat, which while ultimately a zero sum phenonema, is not necessarily in equilibrum at every pressure. Paraphrasing the late great Deep Throat “Follow the water vapour” (especially at the phase change boundaries).
Ratio factor:
Climate models display a fairly robust ratio of the troposphere anomalies/trends represented by the surface variation vs. the LT profile of MSU temperatures of 1:1.2 globally, and 1:1.3 in the tropics (Steve: it is 1.3 for the tropics and 1.2 for the globe). These have been called amplification factors. Models indicate that anomalies and trends of the surface become larger by these factors in the LT profile (See CCSP report for example or Douglass et al. 2007 for model results). Observations have not agreed with these model-calculated ratios for long-term trends, but do agree on monthly scale anomalies. (see Christy et al. 2007 on Tropical Tropospheric Temps) and Douglass et al. 2007.
Re. Ross 63 and Lucia 64. Thanks, very informative and to the point. It is like a magician saying he is going to pull a rabbit out of his hat. He reaches in, and pulls out nothing. He hides his surprise, and then goes on to explain how other magicians have used other methods to pull rabbits out of their hats, and then stands there waiting for applause
Jim Arndt
This exhibit shows that the lower stratosphere is cooling regardless of the solar maximum or minimum. There is a steady decline regardless whether the global lower tropospheric and surface anomalies are dropping , flat or rising .Any comment?
Alan S. Blue said.
December 18th, 2008 at 12:29 pm
Are we officially to a full thirty years of satellite data? Because, if true, that should require a shift of baseline using their own criteria.
The ‘instrumental period’ isn’t the full extent of instrumental temperature measurements – just that prior measurement records were considered insufficiently regular, accurate, precise, and disbursed. (Things like monastery logs, ships logs, etc.)
But here we have a collection of surface station measurements… and a collection of satellite measurements. One of which is more regular, more accurate, more precise, and more disbursed.
Shift, then try to find a calibration for individual surface stations to the nearest measurement in the satellite record.
That should indicate the actual UHI effect, is it possible to do as Alan has suggested.
What? WHere did they tell us this?
Did you notice as you circled that plot that the runs were for 20th century historical forcings? Did you notice that the solar forcing for the 20th century is weak compared to GHGs? Did you notice that panel A–even with a weak solar forcing–shows a trop/trop hotspot?
You’ve misread the plot. I think it was very clear what the plot shows: 20th century forcings. And they definitely do not use “fingerprint” in their description of the plot. So I don’t see how they could be “telling” us about this fingerprint.
I guess its more glamorous and politically useful to disprove GHG warming rather than the wet adiabatic rate.
Oh, well, back to vacation for me…
Boris– the IPCC uses the word “pattern”. Later, they use the term “fingerprints” in scarequotes. But the way fthe term “fingerprint” is used, is synomymous with pattern. Why do you think it’s silly to use a word so vaguely defined in the way that the IPCC does use it when they do use it?
How have I misread hte plot? I never said the plots don’t show 20th century forcings. What does the discussion the fact that the IPCC discusses the hot spot in the section on patterns– which they also call fingerprints– have to do with whether or not the mid-tropospheric hotspot is expected as a result of the level of ghg’s that we have had? And that this pattern, with the hot spot, is due to the ghgs?
And what does any of this have to do with disproving GHG warming or the wet adiabatic rate? The claim the it’s silly to call the tropospheric warming a fingerprint, and then resorting to discussions of the adiabatic lapse rates are red herrings. Steve is using a term that the IPCC does use– but in “scare quotes”, and he is using it the way they use it to describe something they say is predicted to be the result of GHG warming the earth has experienced in the 20th century. That’s a perfectly fine use.
Have fun on vacation!
You say tomato and I’ll say tahmato.
In the meantime, can we please get back to the essentials in discussing the ratios of the surface to troposphere temperature trends in the tropics vis a vis the climate model and observed results.
#92 Matt,
I think my point is valid. Did you look at the TSI graph? It shows a decline in TSI since 1950. This size and intensity of the cycles declines also. If you also look at the Met graph is shows a flat of increasing trend since the early 1990’s. So your point is? If the models say there should be cooling and for 15+ years its been flat or increasing then? Models are great at hind-casting but not forecasting. Being at solar minimum right now and the fact that the ionosphere is at the lowest levels measured we should be seeing a great amount of cooling but we don’t.
92, 98 – solar bugs, tread lightly here; otherwise, go to a solar thread.
#92 Matt
These papers show not only how UV has declined but also ozone.
Click to access acpd-8-453-2008.pdf
https://oa.doria.fi/bitstream/handle/10024/29172/reconstr.pdf?sequence=1
#99
I was trying to show that Stratospheric cooling can be explained from a different view. Sorry if it drifted.
#101 – I understand. The editorial problem that I’m faced with on solar arguments is that I haven’t spent the time yet to understand the potential problems in this data and don’t want to be put in a position where the site is seen as advocating this theory which, for better or worse, is decidedly non-mainstream. The premise of this site is verification not the promulgation of alternative theories – as, editorially, promotion of personal theories easily gets out of hand from my perspective. Tread lightly, that’s all (or go to a solar thread – they are sort of distinct from the other threads)
Steve McIntyre #81
Any chemical engineer can show you that the lapse rate is indeed not fixed. It is a function of the absolute humidity which is a function of SST. The lapse rate reduces as humidity increases raising tropospheric temperatures and increasing OLR. In this way increasing humidity is a negative feedback not a positive feedback. The lapse rate effect is greater than GHG effect. Any model that uses a fixed lapse rate will show warming as moisture increases.
Re: Gary (#103),
You’re spot on about the cooling effect of increasing evaporation from the surface. Entropy operates invariably, cheerfully ignoring the premises of most GCMs.
jim arndt RE #92
Thanks for the clarification.
Under Ross McKitrick’s T3 tax proposal, the CO2 price (tax) would be pegged at 20 times the 3 year moving average of the tropical tropospheric temperature anomaly. In June, 2007, this was estimated at $4.70 per ton of CO2. Any idea what the price would be in December 2008 using the latest temperature data?
A fingerprint isn’t just a pattern. It’s a unique pattern. If everyone’s fingers had the same pattern, they wouldn’t be used for identification. This idea that enhanced tropical troposphere warming is a “fingerprint” of GHG warming has been used by some (David Evans and Lord Monckton come to mind) to say that the warming isn’t caused by GHG, but something else.
The existence or nonexistence of this hotspot depends on selective choice of highly uncertain temperature analyses. Even if these measurements were perfect, you’d still have to contend with huge fluctuations (e.g. ENSO, volcanic eruptions) which obfuscates any signal, especially in the tropics. Look at Santer figure 1. 5 different runs from the same climate model produce trends anywhere between no warming and 0.37 degrees per decade. Same model, different initial values. In comparison, UAH shows no warming over the tropics. RSS shows about 0.1 degree per decade. UMD shows about twice that. And let’s not even get into the radiosondes. Yet we keep hearing about how models are inconsistent with observations, even though the observations themselves are inconsistent, and even then, fall within the model results!
cce:
You are absolutley right. the fluctuations in the model forcasts shows that there is no pattern. therefore they are not forcasting anything. They are saying that anything can happen in the future so we should disregard model results.
Mike,
No, it shows us that no “forcast” can be made, nor have they been made, for the tropical troposphere over a period as short as 30 years. And it shows us that we should disregard anyone claiming to falsify model results based on criteria and observations that are so inconsistent they’d falsify each other.
A correction to 106: The Santer results are for the model equivalent of TLT/T2LT, while the satellite numbers are for “TMT” or equivalent.
#100-101
The reduction in the concentration of ozone will cool the stratosphere, troposphere and surface layers of the Earth.
cce:
I tnought you told us that there were no model results and that the models falsified each other. Must be a language problem!
I only see 3 uses of “fingerprint” in chapter 9–the chapter containing the plot in question. Twice it is used to describe, vaguely, studies/and or techniques. Once it is used figuratively to describe the cooling stratosphere.
Further, the glossary states:
So, by definition, a fingerprint is an entire climate response.
Also note that it is not a red herring to show plots like those in comment #46. In fact. that is exactly what fingerprinting is, according to the IPCC.
As cce says, the important part of the fingerprint analogy is that we are looking for the unique aspects of a forcing–just as a detective would look for distinctive whorls in a suspect’s fingerprint.
It’s not a red herring at all. Why do the models show a trop/trop hotspot? The answer is the wet adiabatic lapse rate–which is in play any time there is a warming and regardless of the cause.
True. Enhanced greenhouse effects will occur in the upper stratosphere, however.
114
How? and at what height.
About 40km-50km.
http://www.atmosphere.mpg.de/enid/20c.html
116
Interesting however at this height the catalytic properties of the odd nitrogen outweigh lwr.No2 blocks solar radiation in the blue green spectrum by a factor of 2
Maybe the author of you link should have consulted his colleague Nobel Laureate Paul Crutzen.
That aside we now have a plurality of causal mechanisms and different CCM and no qualitative matrices,which as Vladimir Alexandrovich Smirnov suggested falsifies the premis.
Re: maksimovich (#117),
It’s not clear to me what you’re referring to here? Above 40km enhanced CO2 radiative loss is the principle cooling agent.
The model results we are talking about are not “forecasts”, but single realizations, just as is the real climate. And it is the “high quality satellite and radiosonde observations” that falsify each other given their vast differences, assuming they are held to the same standard as the models.
Re: cce (#118),
There is still no “rabbit” (warming) It does not matter if the rabbit was bred in the wild,(natural) or bred in captivity, (man made), the rabbit is missing.
Wll cce Let me try to get this right. The model realizations falsify each other because they can not arrive at useful results individualy.
No Mike, the observations “falsify” each other (based on the same criteria used to criticize the models) because they are not consistent, which is a crucial requirement for them to be “useful”. So when people such as David say that the “Rabbit” is missing, we know that he is conveniently picking observations that he likes and ignoring those that he doesn’t while simultaneously mischaracterizing model results. Since variability is especially strong in the tropics (that’s where ENSO “lives”), which is both confirmed by observations and by models, no one would (should) characterize these results as “forcasts” because the signal is comparatively weak. The actual forcast is “expect highly variable weather in the tropics.”
Re: cce (#122),
The actual observations are either a very small rabbit, or his name is Harvey, as no one can see him. The small “observed” warming does not match the IPCC pattern. To then extrapolate that the observations are bad and “Harvey” is real, is seeing what you wish, but unconvincing.
#95: Boris, of course we noticed the Figure 9.1 is in response to historical forcings. That’s the whole point! The model says that if the climate has such-and-such sensitivity, given 20th century changes this is the pattern we’d observe, and the hotspot would be almost all due to GHG’s since the sun didn’t change enough to do it. The data that we have available however says no such hotspot is there. So either GHG levels didn’t rise or they did rise but the sensitivity embedded in Fig 9.1 is too high. But you can’t have the high sensitivity and the observed increase in GHG’s and the observational data. They add up to a contradiction.
#122: Where’s all this uncertainty and anything-could-happen shoulder shrugging coming from? The IPCC sure didn’t admit to any such indeterminacy. This is what the AR4 says when talking about the response to increased GHG’s (pp.764-765):
You need to rewrite your paragraphs using the phrases “good consistency among the models” and “rapid adjustment of the atmosphere to the forcing” somewhere, but I don’t think they’ll end up saying what you want them to say once you’re done. Trying to argue that in tropospheric analysis the models allow for anything and everything is just special pleading, in light of the repeated claims by modelers that the surface predictions and signal detection analysis yields such high certainty. Show me the GCM runs with increased GHG levels that don’t yield a tropical hotspot. The CCSP commented (p,. 11)
In other words when they ran models that yielded results that looked like the data they did so by dialling down the GHG sensitivity.
If you are aware of a published study that has high GHG sensitivity and the historically-observed increase in GHG’s and yet doesn’t show a tropical hotspot then post the citation. And before you cite the Santer scattershot of models you’d better check how they get the tail of the model runs to overlap with the satellite and balloon data. As far as I can tell they do it by putting in models that yield little warming at the surface or aloft, and by dilating the standard errors. In any other context, IPCC authors fiercely resist the idea that error bars might need to be drawn wider and GHG sensitivity might need to be dialled down.
Boris and CCE are beginning to sound just like the Associated Press, when it states that the current cooling trend is a sure sign of global warming. All of a sudden there is too much variability in the tropics to sort out any trends?? Any sort of wiggling nonsensical statement is OK, if you are saving the planet, I guess.
Boris
The hotspot is a response to the specific forcing of GHG.
They don’t say the response must be unique to the forcing. They say it is a response to that specific forcing.
See dictionry example:
http://www.answers.com/topic/specific
“a specific remedy for warts.” is a use of the term specific. This means the remedy specifically treats warts. You won’t need any additional treatment to get read if the warts. Use this spedific remedy by itself, your wart will go away.
The word doesn’t mean remedy treats nothing other than warts but that it will treat warts without the need to use any additional remedy.
A response to the specific forcing of ghg is whatever response arises as a result of ghg even if all other forcing were absent. It doesn’t mean it can’t arise from something else– it means it’s the resonse to ghg.
If the IPCC authors wish to define fingerprint differently and to capture the “if and only if” aspect’ of the analogy to fingerprints,, they need to get better copy editors. In the meantime, people who read the definition of “fingerprints” given by the IPCC will read it to mean what it says: It is the the pattern that arises when we increase GHGs even if we increase no other forcings. That’s what they show in figure 9.1 C of chapter 9.
You can also read the disease part of the devinition. That’s the useage.
Ross,
Ensembles consistently show enhanced warming of the tropical troposphere whether they are combining multiple runs by the same model or all of the models together. That is why there is “strong agreement.” Ensembles eliminate the considerable variability that is easily seen in both the observations and in single realizations. You can look at the Santer et al examples of tropical temperature change based on 5 model runs. Large and small (flat) amounts of warming depended on the shape of ENSO. Same forcings. Same sensitivity. Different inital values. Different results. The uncertainty surrounding any trend is therefore going to be huge. Some might find that convenient, but to quote The Shawshank Redemption, “It is decidedly inconvenient.”
jae,
There is no cooling “trend”.
As for a trend in the tropics, you can take the same satellite data, examining the same area, over the same time period, and get ~0 degrees per decade, ~0.1 degrees per decade, and ~0.2 degrees per decade based on the study (UAH, RSS, and UMD — TMT or equivalent). If you want an example of a nonsensical statement say, “These are high qualtiy observations from which we can disprove model results!”
David,
I think it’s a case of people who don’t like certain rabbits so they claim they don’t exist.
Re: cce (#127),
I don’t want to debate here what the trend actually is; I’m just relaying what the AP said.
Re: cce (#127),
Another red herring. The absolute value of the trend in TLT or TMT is unimportant and quite uncertain to boot. It’s the ratio of TMT to TLT for RSS or UAH data that is the signature, fingerprint or whatever. The ratio should be greater than one according to the IPCC even if the pressure range over which TMT is measured is lower than we would like if we wanted to see the maximum effect.
So let’s look at the data. For UAH if we regress the t2 tropical land anomaly to the t2lt tropical land anomaly we get a slope of 0.936 +/- 0.027 (95% confidence). Doing the same thing for RSS gives a slope of 0.912 +/- 0.025. In other words, the satellite data shows that the middle troposphere cools when the lower troposphere warms with a high degree of certainty. The F statistic for the RSS regression is 5123 and for UAH it’s 4856 so the correlation is indeed quite good. I’m just using Excel here so I haven’t tested for autocorrelation or other problems that might reduce the significance of the test. I may even be doing the wrong test. Comments appreciated.
Re: DeWitt Payne (#137),
T2/TMT is 5-15% stratosphere. (T3/TTS, although centered on the “hotspot”, has a boatload of stratosphere). T2LT/TLT attempts to remove almost all of the stratospheric influence. Thus, you are not going to get a meaningful ratio because TMT warming is muted by the cooling stratosphere. The fact that these channels are so wide is one of the reasons why it’s so hard to compare observations to models. You have to recreate the weighted profile of the satellite channels with model results.
Furthermore:
Click to access VinnikovEtAlTempTrends2005JD006392.pdf
Again, that conclusion is based on the same underlying satellite data.
Re: cce (#138),
I don’t consider Vinnikov a reliable source considering the hash he made of converting MSU data to temperature a few years back. IIRC, and I’m going from memory here, according to his algorithm, diurnal temperature had two peaks at about 11 AM and 5 PM local time. Of course, his OLS trend was about 0.24 C/decade so a lot of people were very pleased with him at the time.
So let’s restrict the data to 1994 on where t4 data shows the lower stratosphere temperature has been nearly constant. UAH tmt vs t2lt tropics (land) still has a slope of 0.926 +/- 0.038 and an F statistic for the regression of 2344. The anomaly range is from -0.51 to 1.19 for t2lt and -0.45 to 1.22 for tmt. Remember this is a pairwise comparison of the data from the same year and month for each level. Even though a lot of correlation is expected because of the overlap in the measurement range, if the troposphere warms faster with altitude, the ratio should be greater than one. It isn’t for whatever reason.
Since you claim the stratosphere temperature is a confounding factor, I’ll regress tlt and tmt on tls. The slope for tlt vs tls is -0.158 +/- 0.067 with an R squared of 0.10 and an F statistic of 22. For tmt on tls the slope is -0.098 +/-0.068 with an R squared of 0.04 and an F statistic of 8.4. Both are significant but not very large. If I had the skills and or the software package I’d do a multiple regression. I’m betting, though, that the result would be the same.
Re: cce (#138),
But that weighted profile is exactly what Santer et al. (2008) did in their paper and what was discussed at length here at CA. In fact, it is the weighted profile from the climate models (to match the T2LT and T2 observed temperatures from RSS and UAH) that was requested from Santer so that we could move the model and observed comparison further toward the present.
From my notes I have:
It is not a distinguishing feature as Douglass would have you believe. Fingerprint analyses are interested in the differences between forcings, not in what is the same. You guys seemed to be obsessed with what is the same for all forcings–and then call this a fingerprint. I don’t get it. If “fingerprint” is getting in the way, let’s use another term.
Do you agree that the troptrop hotspot is not a distinguishing feature of GHG warming?
I just stopped there. Do you have some sort of compulsion that makes you post inane comments no one is the least bit interested in?
Re: Boris (#129),
Heh, heh. Irony. And dodge.
A “fingerprint” means unambiguous attribution of an observed effect to a specific cause. Theft and thief. Warming and warming agent.
Ross, this has squat to do with sensitivity. The hotspot isn’t related to sensitivity, but to the actual warming that has already occurred. They didn’t dial down the sensitivity-they dialed down the warming, as they clearly state.
This is why it’s essential to understand the reason why the hotspot is present–what Lucia called a “red herring” earlier. The hotspot is not merely a result of models. It is inputted into the models and is based on solid observations of the wet (AKA moist, AKA saturated) adiabatic lapse rate.
Re: Boris (#130),
The hotspot is not merely a result of models. It is inputted into the models and is based on solid observations of the wet (AKA moist, AKA saturated) adiabatic lapse rate.
Re: Boris (#130),
The hot spot is inputted into the model?! Supposedly, conservations equations and parameterizations are coded, and these features emerge. The adiabatic lapse rate is a solution based on our knowledge of conservation equations, thermo and some simple assumptions.
You now want to explain that understanding the reason for the hot spot is “essential”, and put my red herring in scare quotes. But when you make a claim something is “essential”, you should say it’s essential to some particular argument. Understanding why the hot spot occurs is certainly essential to some discussion somewhere– but shoved into a criticism of Douglas it was a red herring.
The reason the RC discussion is a red herring has to do with where the discussion appears. The discussion was wedged into an article where it was utterly irrelevant. The authors essentially changed the subject from what Douglas said or claimed, went off on a little tangent and then came back to discussing aspects of Douglas.
Changing the subject to something irrelevant to the topic at hand is called a red herring. See wikipedia. The fact that the irrelevant topic might be interesting and even correct doesn’t change it’s color or fishiness. The fact that it might be important in some other context doesn’t turn it into delicious lamb chops.
The feature of red herrings is they are irrelevant to what is being discussed. Oddly, enough, because audiences are generally only fooled by red herrings that are both interesting and correct in some other context, most red herrings are both. They are also irrelevant in the context in which they appear.
Boris–
You specifically criticized people for using the term fingerprint. Now that you have supplied the definition of fingerprint from the IPCC, and I pointed to the meaning of the word in the definitoin, you want to change to whatever you mean by distinguising feature? Could you define what you mean by “distinguishing feature”? And then, tell us how this is relevant to either a) your claim that steve misused figerprint in a silly way, b) Douglas calling the hot spot a fingerprint, which it is.
Boris, I fail to see what the issue is here. As far as I can see, observations and models do not match up. Nature is the final arbitor, is she not? If they don’t match then either:
1) models are wrong and/or incomplete or
2) the data has issues with resolution and/or accuracy that preclude definative evidence on this issue.
If 1 is true, well, no need to go there – besides, it’d be snipped anyway. 🙂 If 2 is true, then we are talking about changes that we cannot measure. If we cannot measure them, they can hardly be evidence of disaster being near to hand (or evidence of it not, for that matter). Either way, they most certainly cannot be evidence that the models are correct. Suggesting that the data is wrong because it doesn’t match the models is… well… somewhat odd, to put it mildly. Would you accept that the data is wrong because it doesn’t match the models for say, gravity? Perhaps if this was a single observational series, there would be pause, but as far as I know we have at the least hundreds of observations by different groups in different conditions and locations, using different equipment. I am no stats expert, but the odds of them *all* failing on this particular issue seem somewhat remote to me, so I feel the data cannot be discarded a priori as faulty unless you have some plausable explaination of what is happening and can then demonstrate a repeatable failure under lab conditions.
I did the multiple regression in Excel. TLS contributes very little compared to TLT. The line fit plot for TLS looks very much like a target hit by a shotgun with a narrow choke and number 8 shot. Of course, the residuals are highly autocorrelated so the significance of the fit is nowhere near as good as indicated and the true confidence limits on the slope are higher. How much higher I don’t know. Someone who understands this better than I could certainly do a better job.
Tell that to Lucia. Then research upper stratospheric temperatures. Then come back and say something useful. I know you’re capable of cloying your way through the BS and spin when you want to.
And Lucia, why do you think the IPCC conducts fingerprint analyses? I wonder if it could be to find distinguishing features. The TT hotspot is just another dneialist talking point.
Re: Boris (#141),
And a very good talking-point, too. The AGW crowd is running out of “proofs” for any significant warming, and meanwhile it’s getting colder…
Re: Boris (#141),
Pretty ironic words considering that your argument boils down to “no warming” = “anthropogenic global warming.”
Re: Not sure (#145),
I’m “not sure” you have understood my argument.
Meh, I know. But seriously, when did I say anything like that?
#130: Boris, you’ve lost me.
Maybe there’s some deep idea here I’m totally missing. First of all, the problem with “the actual warming that has already occurred” is that it hasn’t occurred in the tropical troposphere on any scale remotely close to what the backcasts predict, or the A1B scenarios in IPCC Fig 10.7 predict we should now be seeing. Yet the increase in GHG’s has occurred. To me this has a lot to do with sensitivity.
Second, how did they dial down the warming in the presence of so much additional CO2 without dialing down the sensitivity?
It’s all ocean currents, particularly the AMO. I see a rise in the smoothed TLT anomaly in the NoPol (UAH) or 60-82.5 (RSS) starting in about 1995 that peaked at over 1 C 2005 or 2006. Plug a rise of 1 C into MODTRAN for sub-arctic summer or winter and you get an increase in OLR at 100 km of about 3 W/m2. Heat radiates much more efficiently from high latitudes, where the tropopause is at a lower altitude and higher temperature, than from the tropics. Factor out ENSO and there hasn’t been much of a global temperature increase in the satellite data since about 1995. That also explains flat to down temperature time series and small increasing trend in sea ice in the high southern latitudes as well. When the AMO finally slows down again, which may be starting to happen now, Arctic ice will recover, but the tropics should warm. That will be the time to test whether the troposphere temperature profile in the tropics follows the model predictions or not.
Ross,
Models with different senitivities will give the same result for surface wamring and TT warming. Only the models that return an unrealistic surface warming show a match to the TT. But then those models don’t match the surface–so they tell us nothing. Now, the models that show little surface warming might tend to have lower sensitivity–that seems plausible. So I see where you’re coming from here. But those models are wrong–they cannot hindcast the known warming.
The upshot is that there is a discrepancy:
Could the models be wrong in the way they handle the wet adiabat? Absolutely.
Could the data be too poor to draw conclusions? Absolutely.
Could the TT variability over 30 years be such that we cannot compute the true climate trend? Absolutely.
Could there have been no real surface warming in the tropics? Extremely doubtful.
But the TT hotspot is not a function of the model’s sensitivity. It is a function of the surface warming.
Re: Boris (#147),
Boris: We seem to be going in circles here. I understand that TT should warm faster than surface per models used by IPCC. I under stand that IPCC states that warming is proof of AGW caused by CO2. The TT is not warming faster than the surface. The TT hot spot is only created in the models not in reality in any provable way.
I’m a tad confused. The troposphere is supposed to get warmer than the surface and then I assume at the stratosphere it gets colder?
The simple planetary radiative balance equations say there should be a decrease in the optical depth of the atmosphere which is balanced by an increase in the effective temperature which I assume is in the part/layer of the troposphere that got measured and everyone is talking about (IPCC included)? That bit is going to move about a bit in height though I expect.
If so the MSU tlt3 data appears to show that the troposphere is not warming as much as it should compared to the surface and hence the ‘forcing’ of CO2 appears not to be as strong as modelled.
And I am purposefully ignoring the massive uncertainties just so the discussion isn’t moot.
Okay I think I get it. So the hope is that the next IPCC report might incorporate this knowledge then?
Re: MC (#148),
If I remember correctly, there’s not just the various altitude ranges of the atmosphere that participate in balance, I’m remembering land masses and bodies of water. I understand some of these bodies of water are very deep.
On the other hand, when William and Meriwether duplicated the efforts of Sir Alexander in reaching the Pacific coast of North America, nobody was even sure exactly how much Thomas had bought from Napoleon. Or actually in person, Robert and François. A group, really. But in any case, it turned out to be quite the sale! Or at least a precursor to buying things from the actual owners. In a way. Different times.
Best,
SU
Re: Sam Urbinto (#150), Yeah there’s a bit more too it. I’ve had a look deeper into it. Still its quite clear what the models have predicted and then to what end and within what uncertainties the measured data compares. It looks like there is a discrepancy. And then comes the inevitable conclusion that we don’t really know enough yet to say with the certainty what people are saying currently. If the uncertainty in measured data is low, the models are incorrect: go do some more work; if the uncertainty is high then there’s no use comparing the models to data as there’s such a low signal in the noise: go do some more work and get better data. I’m not hearing this inevitable and obvious conclusion in the mainstream and so its a bit sad from a scientist’s point of view
Re: Ross McKitrick (#156), There’s also the assumption about a mean trend and weather noise in models so the definition of climate being only defined over a 20+ year period appears completely arbitrary. Surely if you are sampling every month then 2 years is a big enough sample to get a mean that represents some temporal trend. But then a better understanding of the underlying physics would normally give you an indication of this.
The following are initial calculations (small E&OE) for the maximum and minimum annual average tmperatures from 13 isolated, rural Australian sites from the Hi Quality Bureau of meteorology reference network rural group. They are tropical or near-tropical, (between 0 and 26.6 deg south lat). Some are coastal, some are inland. They are stated to be raw data, though I have infilled some missing values and these are preliminary, but unlikely to change in the qualitative view.
year Tmax Tmin
1968 29.69 17.89
1969 30.55 17.94
1970 30.81 18.34
1971 30.29 18.27
1972 30.90 18.48
1973 30.57 19.04
1974 29.55 17.75
1975 30.16 18.03
1976 30.28 17.41
1977 30.59 18.09
1978 30.06 18.18
1979 30.83 18.62
1980 31.18 18.90
1981 30.72 18.54
1982 30.39 18.05
1983 30.65 18.56
1984 30.20 18.14
1985 30.91 18.65
1986 30.71 18.89
1987 30.65 18.75
1988 31.45 19.33
1989 30.41 18.51
1990 30.89 18.81
1991 31.01 18.85
1992 30.68 18.90
1993 30.31 18.83
1994 31.11 18.15
1995 30.56 18.86
1996 31.21 19.14
1997 30.35 18.49
1998 31.21 19.67
1999 30.44 18.37
2000 29.71 17.97
2001 30.09 17.91
2002 31.57 18.39
2003 31.00 18.80
2004 30.66 18.60
2005 31.41 19.35
2006 30.57 18.42
I suspect that you will find that these results do not conform with the general shape so often shown for SH or global temps.
Re: Geoff Sherrington (#152),
Typo again. There are 9 sites in this average calcuation, not 13.
But then, tropical mainland Australia is only about 2% of the area of the whole SH.
Re: Geoff Sherrington (#152),
I make the linear trend 1.4 deg C per 100 yrs for Tmax and 1.65 deg C per 100 yrs for Tmin. These are much greater than the Douglass and Christy calculation on UAH data since 1979 before they took out the impact of the ENSO and volcanoes which reduced it to about 0.6 deg C per 100 yrs.
However taking out your pre-1979 data points also reduced your trends to around 0.11 deg C per 100 yrs – flat lines!
Re: Alan Wilkinson (#154),
Re: John S. (#155),
Thank you both for comments and to John S for the extra data via Craig.
I chose the period 1968-2006 because (a) Australia went decimal in 1966 and thermometry might have changed scales (b) a number of the BOM designated rural reference sites start their official capacity about then (c) there was the stready introduction of automated data gathering often at more frequent intervals and (d) I did not want to be accused of cherry picking sites. There is no agenda, I have just put the available data into easy visual form. Soon I hope to add 2007 and 2008.
There is nothing much new in this work, except questions like (a) why did Australia go into that 1950-60s trough and (b) if these isolated, selected definitely rural sites mostly show no 1998-2006 “hump” when the rest of the world does, what is wrong with the data from rest of the world, excluding 30 years of satellites?
I just happen to think that these are critical questions for the world scene. Steve has answered many.
Re: Geoff Sherrington (#152),
Geoff,
It’s always fun to compare “official” annual average temperature series for a region with cursory compilations from unadjusted station records at locations little affected by UHI or by land-use changes. The latter results invariably are flatter over the long run. I did the same with a handful of Australian records a few years ago and discovered that the BOM series, based on scores of records, differed from mine primarily before 1940, where BOM shows considerably lower values. This discrepancy obscured the deep trough that apparently developed in Australian temperatures in the 1950’s and 1960’s and produced a “trend” five times greater than my data show. Your compilation for tropical Australia begins at a time of recovery from that trough, a full decade before the rest of the world–a point to keep in mind when considering subsequent “trends” in your data. It seems that Australia led the world in something besides the jaunty machismo everyone admires.
P. S. I hope you received my 1897-2005 compilation via Craig Loehle.
#147: Boris, where you place your fixed point determines where the trouble arises.
You are taking the surface warming, especially in the tropics, as a fixed point and assessing plausibility by the match thereto. I take the surface temperature data, especially in the tropics, to be poor quality and contaminated with an upward bias. Actually the tropics is so badly sampled in the post-1979 interval that most cells have too little continuous data to identify a trend, but the ones we do have data for show pretty consistent positive bias in Figure 4 of my 2007 JGR paper. Remember these are poor countries with few resources for data collection. Not much of a basis for building hypotheses on.
30 years of continuous monthly satellite data is sufficient to identify a 30-year trend. If a 30-year trend is too short to be climatically meaningful then we cannot talk about the post-1980 interval at all. But it’s the post-1980 temperature data that carries the promotional freight in the AR4. Look at the way the instrumental data gets tacked on to the end of the paleo spaghetti graph, for instance, and look at the way the model-observation lines diverge in the “signal detection” diagram for 5 continents in the SPM (sorry I don’t have a page # handy). In each case the visual effects are in the post-1980 portion.
More formally, in time series analysis using persistency modeling, there is no scale break at 30 years, or 20 years, that would indicate emergence of a climatological structure not observable at shorter timescales. The heuristic argument that climate is defined as an average over ~30 years (say) assumes that some kind of stationarity emerges in averages defined over that interval or longer. The interval is called an integral scale, but despite numerous people looking for it there is no evidence for it (of the papers I’ve seen). There is an integral scale break somewhere between 14-30 days, but no new structure emerges thereafter. I’m not explaining this very well: Karner’s papers are the ones to read.
Ross:
When I go to the NOAA CDC site they provide regional 90 day to 2 year climate “projections” (What if’s). Also NCDC, NASA, ad naseum compare monthly and yearly to a fixed 30 year “mean”. Right or wrong it seems to be set in stone if these groups use these time periods and call them climate Who are we to think otherwise? I think Boris has confused himself over this situation.
Regarding #152, Geoff Sherrington’s data, if you tentatively accept that a simple linear model may perhaps not be appropriate for this length of time series it is interesting to look at some alternatives. A very simple one that I propose is to split the data into two segments at 1985. Now carry out simple regressions on the two segments, and you will find that none of the four has a significant slope. However, the mean values are (with rounding) 30.5 and 30.7 for TMax and 18.3 and 18.7 for TMin. Plots of the regressions with confidence intervals (95% seems to be the conventional choice for many practitioners) are very consistent with a tentative hypothesis that no temperature changes are statistically significant for the proposed segments. From the mean values it looks as if upward step changes for Tmax and Tmin occurred at about 1985, of magnitudes estimated by the differences between the means given above. Both prior to and after these steps the temperatures cannot be shown to have changed. (It is easy to put forward a slightly more elaborate model, but this woud take too much space to explain).
“How did you choose 1985 for the step point?” you may ask. Simply by using an industrial quality control technique which readily indicates possible points of level changes.
As an aside, If you apply this technique to Arctic (NW Atlantic, Greenland and Iceland) data you will find a very large, obvious and consistent step at about Sept 1922, of the order of 2 deg C, which seems somehow to have been overlooked by climatologists, as far as I can ascertain.
Robin
Ross,
I understand that a possible argument could be that there has been no surface warming in the tropics. That is a more cogent position than some I’ve seen on some blogs recently. However, whatever problems there are with the surface record, the upper tropospheric record is worse. It’s hard to see how an inferior dataset can negate a superior one.
Re: Boris (#161),
Another argument could be that warming exists but is overstated. In fact, of all the hypotheses, I think that is the one that best explains the (admittedly flawed) observations.
The evidence that Ross and others have generated indicates that the surface record is likely biased upwards. Is there any evidence that the upper tropical troposphere record is biased one way or another? Why do you say it is worse?
Boris–
YOu are correct that an inferior data set can’t negate the better one. But whichever set is better or worse, the tropospher record doesn’t confirm the models. This could be because data are poor. But it could be because the predictions are inaccurate.. But we also can’t be absolutely sure the surface record is excellent. So, even if the surface record is better than the troposphere, it could be off somewhat. (In principle, it could be off in either direction.)
If we are goint to use a preponderance of the evidence test, we need to consider a range of possible theories, and see how well (or poorly) the data support those theories.
This is definitely not the case for a typical atmospheric variable(such as temperature) where large deviations around the mean are comparable to the mean itself. More generally, this should hold true in any nonlinear dynamical system in which the evolution equations admit multiple solutions or give rise to chaotic dynamics.
The divergence in the temporal horizon suggest the predictive values are of limited qualities.ie there is a temporal boundary.
Boris, in addition to the problems with land-based surface data, there is the problem noted back in 2001 with the use of sea surface data in the tropics. Christy et al found that where buoys measured both sea surface and air temperatures, the air was cooling relative to the sea surface, so SST data overstated the atmospheric warming in the tropical region. I hope this study will be updated at some point.
As for ranking the surface and mid-T data, I wish the CCSP would subject the surface temperature data to as much critical scrutiny as they did the MSU data. The GISS and CRU products get what appears to me to be a free pass from groups like the CCSP that should be all over it with a Red Team-type assessment.
Ross (#156),
How true! There is little structural basis for the conventional 30-year climatic “norms.” They are a relic from the pre-computer era and constitute little more than a convenient sample size for hand-calculating the customary climate statistics suitable for tour-guides. Such customary norms are wholly inadequate for the scientific purpose of tracking long-term climate variations measureable in tenths of a degree or establishing any secular trends.
In time-series analysis of real world data, I prefer to work with model-free methods. The power spectra of yearly average temperatures invariably show that the highest power densities occur at frequencies below that of the Hale cycle (~22yrs). Throughout the higher frequencies, spectral densities are comparitively flat, but by no means featureless. Significant peaks are virtually ubiqitous near periods of ~5.5yrs and are often found also near ~3.7yrs. While the 30-yr norm does an adequate job of averaging over a handful of the shorter cycles and produces fairly stable aggregate statistics, the corresponding “trend” is very highly variable. It can be shown analytically that the regressional trend computed on a moving basis is nothing more than a crude band-pass filter, with rather undesireable response characteristics. Projecting such conventional “trends” into the future is patently foolish.
Re: John S. (#166),
It appears you are a working scientist. If so, I take much gratification (and vindication!) from your skepticism regarding “conventional 30-year climatic “norms.”,” which have always seemed arbitrarily short to me. I am in no sense a scientist, but still suspect that 100 years would be the minimum required to see a definite trend.
——————————–
Starting to stray OT a little, but could someone point me to a good primer on trop warming/strat cooling, which, unlike the controversial tropical hotspot, seems an agreed-upon & unambiguous AGW “fingerprint.” I get the basic theory about ozone depletion and tropospheric heat-hoarding, but in reviewing the relevant (I think) sections of AR4 & other sources, I find no discussion of potentially confounding variables & alternate explanations. IOW, what makes the AGW crowd so confident that this combination is a smoking gun?
Anyone? Thanks.
Re: PaddikJ (#167),
The short answer based on a very simple two layer model of radiative transfer with all other things like clouds and aerosols being equal is that the troposphere is relatively opaque to long wave radiation and relatively transparent to short wave radiation while the stratosphere, while being relatively transparent overall, is more opaque to short wavelength (less than 350 nanometers wavelength) than to long wavelength radiation. Using a simple one dimensional gray atmosphere model it’s easy to show that if you increase the long wave optical density of the troposphere by adding CO2, long wave emission goes down and the temperature of the troposphere and the surface has to go up to restore balance.
For the stratosphere, an increase in CO2 increases long wavelength emissivity and absorptivity significantly and has almost no effect on short wavelength absorptivity, but there isn’t much long wave to absorb compared to incoming solar radiation (which we are assuming isn’t changing), so the increase in emissivity at long wavelength causes a net loss in energy, particularly since the temperature increases with altitude in the stratosphere due to short wavelength absorption by oxygen and ozone and emission increases with temperature. So the stratosphere should cool until energy balance is again restored. With very little convective heat transfer in the stratosphere and very low heat capacity, the stratosphere equilibrates rapidly (weeks or months).
There is little question that this pattern of low altitude warming and high altitude cooling, all other things being equal, distinguishes ghg induced tropospheric warming from warming due to absorption of short wavelength radiation from an increase in solar radiation, which would warm the stratosphere as well as the troposphere and the surface.
Re: DeWitt Payne (#170),
There is no more efficient greenhouse gas than ozone. This makes the lower stratosphere relatively opaque to long wave radiation. The evidence for this lies in the strong temperature peak at 200hPa and above (yes, in the troposphere) in mid year at 30-40° S latitude in the south East Pacific.
This is a zone of down-welling air where stratospheric ozone finds its way into the troposphere. A strong peak of outgoing long wave radiation that is associated with seasonal cloud loss in the southern tropics warms the air at 200hPa. Look at figure 5 in my post at http://climatechange1.wordpress.com/
In the southern tropics the tropopause experiences a thermal maximum in August right round the globe.
Theoretical physics is no substitute for a hard look at historical data. It can be downloaded at http://www.cdc.noaa.gov/cgi-bin/Timeseries/timeseries1.pl
From memory,(always doubtful) the tropopause warmed until 1978 and has cooled since, along with the rest of the lower stratosphere.
1978 marked the transition from a weak solar cycle 20 to a strong cycle 21 and the start of a run of El Nino events.
Over the period of record since 1948 specific and relative humidity has declined at all levels in the atmosphere. The implication is that cloud cover has declined. We need look no further for an explanation of the gradual warming or the cooling that is now in train.
Re: John S. (#166),
The unreliable eye often sees cyclicities in climate data and the more reliable mappers and geodesy people and mathematicians find observable effects like the gravitational influence of solar bodies on orbits of others. Do you know if anyone had ever compiled the list of periodicities say for the last 500,000 years or so? Dividied into reasonably assured and speculative? You mention the Hale cycle for starters.
Your comment on regressional trends drew me off on a tanget. My apologies. Let’s use plant growth as a dependent variable and fertilizers (including CO2) as a group of inputs. In a classical factorial trial, one uses (say) 20 different nutrients at (say) 5 levels of concentration each, in all possible combinations, then to be more sure, triplicates the study. In a conventional non-linear multiple regression analysis where it is assumed that the fertilizers are not independent of each other, one can do a conventional maths dissection to indicate the strongest influences. But, it has been long been discussed that if certain nutrients are low or absent, there will be no additional response caused by the addition of other fertilizers. Is there a name for this type of statistical analysis where the responses are not smooth but can be stepped or halted entirely? I ask because of the many past posts on regressions, where I have not seen an example used for “if no more v, then no more response to w, x, y or z”. In a hypothetical, if molybdenum was limited in some dendro areas, then water, CO2, SO2 and various other nutrients would not cause a growth response and the growth rings would be affected accordingly.
PaddikJ (#167),
The very question of linear temperature trends is moot. Given the strong multi-decadal components evident in the spectra of long series of yearly averages, even a century of data doesn’t establish a stable secular linear trend. If the interest is in tracking temperature variations on time-scales commensurate with human life-times, then linear-phase low-pass filters (Butterworth or Chebycheff) can be used to show those variations more clearly (see UC’s example here in #58). Linear regression tends toward quasi-stable results–decade to decade–only when computed over nearly two centuries of data. Thus, for example, the historic “De Bilt” series (from 1706) shows a 176-yr trend in the range of 0.2-0.3K/century throughout most of the 20th century. Being sensitive to start- and end-values, regressional methods are simply ill-suited for analyzing data with strong cyclical components.
The answer to your question about references on the tropospheric “hot spot” is, likewise, not simple. Inasmuch as this is a fiat of most models, the climate modelling literature would be the place to look. You will not find any explanation of this feature, however, in rigorous atmospheric physics tracts written by those who possess a mastery of the laws of thermodynamics. For a masterful account of how the climate system works, I would recommend John Dutton’s “The Ceaseless Wind.” Meanwhile, bear in mind that most GCM’s used for climate modelling do not even get the diurnal cycle right, because of over-damping by imputed “greenhouse” effects. And the one model (Russian) that does not conform with the rest, but agrees best with satellite observations, is dismissed by IPCC as “unrealistic.” Go figure!
Re: John S. (#169)
Thx for the info; much of the terminology is unfamiliar, but w/ the help of a statistician friend I was able to make it out, except: “even a century of data doesn’t establish a stable secular linear trend.” Assume that in this context, secular means century-to-century?
Did some research on your book suggestion, but was unable to determine if it’s a technical specialists’ book, or intended for the lay reader. Seems like that latter from the title, but before I shell out twenty bucks, it would be good to know . . .
—————————-
Re: DeWitt Payne (#170)
And also little question that this combination obtains strictly from either GHG’s or variations in the “solar constant”?
thx,
PJ
Geoff (#168),
I’m experiencing problems with my antiquated browser when the number of comments at CA grows large. Let’s discuss your purely statistical questions off-line. I believe you have received my e-mail address along with the Australian compilation.
Re: John S. (#171),
Ok and thanks Geoff.
Oh, this is just silly.
We all know that NORMAL temperature is the mean of 1960 to 1990. That’s what the temperature of the Earth is SUPPOSED to be! What it was always MEANT to be. That’s the RIGHT temperature.
(that was sarcastic)
I was just reading an article by Monckton and found he not only used the term “fingerprint” but also showed images of the fingerprints of different forcings. I have to admit that I do not understand this yet, but thought I would draw readers attention to the article because it may shed some light on the subject. The images can be found in Physics and Society on pages 10-11.
2 Trackbacks
[…] *2008 Tropical Temperatures“, Steve McIntyre, at Climate Audit, 18 December 2008 — “We now have a 30-year period of satellite records. Within that period, 2008 ranked 26th out of 30 (5th coldest), 23rd for RSS, 16th for CRU and 15th for NOAA and GISS.” […]
[…] on the it. For the more technically inclined, or for those wanting more, Steve McIntyre posted an interesting discussion at Climate Audit. – Anthony Models predict this heating in the […]