Guest Post by Willis Eschenbach
I chanced to plot up the lower tropospheric temperatures by broad latitude zones today. This is based on the data from the satellite microwave sounding unit (MSU), as analyzed by the good folks at the University of Alabama at Huntsville. Here are the results, divided into tropical, extratropical, and polar. I’ve divided them at the Arctic and Antarctic Circles at 67° North and South, and at the Tropics of Capricorn and Cancer at 23° N & S.
Figure 1. Satellite-based microwave sounding unit temperatures (red line) from the University of Alabama Huntsville. Blue line shows a loess smooth, span=0.4. Data from KNMI (NCDF file, 17 Mb)
So … is this something to worry about?
Well, let’s take a look. To start with, the tropics have no trend, that’s 40% of the planet. So all you folks who have been forecasting doom and gloom for the billions of poor people in the tropics? Sorry … no apparent threat there in the slightest. Well, actually there is a threat, which is the threat of increased energy prices from the futile war on carbon—rising energy prices hit the poor the hardest. But I digress …
What else. Southern Extratropics? No trend. South of the Antarctic Circle? No trend, it cooled slightly then warmed slightly back to where it started.
So that’s 70% of the planet with no appreciable temperature trend over the last third of a century …
What else. Northern Extratropics? A barely visible trend, and no trend since 2000.
And that means that 96% of the planet is basically going nowhere …
Now, that leaves the 4% of the planet north of the Arctic Circle. It cooled slightly over the first decade and a half. Then it warmed for a decade, and it has stayed even for a decade …
My conclusion? I don’t see anything at all that is worrisome there. To me the surprising thing once again is the amazing stability of the planet’s temperature. A third of a century, and the temperature of the tropics hasn’t budged even the width of a hairline. That is an extremely stable system.
I explain that as being the result of the thermoregulatory effect of emergent climate phenomena … you have a better explanation?
My best regards to everyone,
w.
PLEASE! If you disagree with what I or anyone says, QUOTE THE WORDS that you disagree with, and say why you disagree with them. That way we can understand each other. Vague statements and handwaving opinions are not appreciated.
DATA: All data and R code as used are here in a zip file.
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
timetochooseagain says, January 29, 2014 at 12:46 pm:
“The same thing happened around 1992 globally, except it was RSS warming relative to UAH. And RSS was wrong there and they are wrong now.
http://devoidofnulls.wordpress.com/2014/01/13/the-curious-case-of-noaa-12/“
Ok, so it is you. No, in 1992 UAH is clearly wrong (compared to every other global dataset, land and ocean) in bringing temperatures post 1992 too far down. No question about it. I urge everybody to just look at the data. After 2005/6 RSS definitely goes a bit low, so should be adjusted up accordingly. But UAH timeseries needs to be lifted en bloc between 1992 and 2005/6. That will effectively wipe out the discrepancy in trend 2001/2-13 between UAH and the rest of the pack.
RichardLH says:
January 29, 2014 at 12:15 pm
HAIYEEE … no, RichardLH, changes in temperature can no more be averaged than temperatures can, despite the fact that we do it all the time. If we were smart we’d convert it all to W/m2 fluxes, do our averaging, and convert back to temperature.
The problem is that temperature is not conserved, while W/m2, being a measure of energy, IS conserved.
w.
BarryW says:
January 29, 2014 at 1:17 pm
Thanks, Barry. I went ahead and wrote my own … found out that the difference between V5.5 and V5.6 wasn’t visible to the naked eye in Figure 1, as I’d suspected.
Appreciated,
w.
@Kristian-No, you continue to spread completely inaccurate nonsense.
“What the data shows us” is something you haven’t sufficient braincells to actually understand, as evidenced by your continuing to make invalid comparisons.
I have *shown* that you are dead *FLAT WRONG* and you continue to spew ignorant, incorrect nonsense. You should get lost.
timetochooseagain says:
January 29, 2014 at 1:39 pm
“@RichardLH-You can’t just apply scale factors to attempt to force the long term trends to agree with one another. You end up getting backwards results.”
Interesting. You can do this with proxies but not with measurements. Even though it aligns them all to a common standard in the process.
RSS and UAG are purely a 0.1c offset over their overlap period. And there is a very tiny adjustment needed to bring their range scale into line as well.
Takes care of the different methodologies quite well without having to go line by line through them.
As to thermometers and satellites, there seems no reason not to do the same.
They ARE supposed to be representing the same basic underlying property after all. As these are all anomalies rather than absolutes it should work just fine. After all rate of change = rate of change.
How else do you suggest we calibrate them?
Willis Eschenbach says:
January 29, 2014 at 2:03 pm
“no, RichardLH, changes in temperature can no more be averaged than temperatures can, despite the fact that we do it all the time.”
If I put a thermometer is each room in my house, I can derive a perfectly good ‘average air temperature’ out of them.
If I track the changes in those thermometers I will be pretty accurate in the energy I supply to the house as well, if losses are being kept approximately constant.
Kristian says:
January 29, 2014 at 1:55 pm
If you think you are correct, I urge you to make your case to Spenser and Christie. After all, they’ve spent thousands and thousands of hours contemplating those exact questions in painful mathematical detail, and you’ve spent … well, hours on it at least, and that without even looking at the underlying equations they’ve used and the issues they’ve identified.
So you may be right … but I’m not buying your claims until S&C do.
It’s easy to say “three people say yes and one person says no, so the one person must be wrong” as you have done … but we all know that’s not the case in general. Far too often, the one guy is right and all the rest are wrong.
In addition, you can bet that S&C have looked long and hard at the discrepancies you mention between their work and that of RSS … as have the RSS guys. So for you to come along and blithely announce that you’ve solved the conundrum, and that the “UAH timeseries needs to be lifted en bloc between 1992 and 2005/6” … well, I don’t think that is what passes for science in the world of RSS and UAH.
But heck, Kristian, please get back to us and let us know what S&C have to say about your theory …
w.
@RichardLH-“Interesting. You can do this with proxies but not with measurements. Even though it aligns them all to a common standard in the process.”
No, can’t do that either, not if there may be spurious biases in the data. Mannian method, Mannian result.
“They ARE supposed to be representing the same basic underlying property after all.”
Not quite. See comments above, I’ve been pretty clear on this. Did you notice that you *increased* the extent to which the satellites have greater variability than the surface?
“How else do you suggest we calibrate them?”
If you *really* want to infer surface trends from satellites, you need to first take them over the same period (1979-2013) take annual averages for all, detrend each, and then average the like kinds of detrended data, and then do a regression-you should get a coefficient for detrended surface data as a predictor of satellite data of about 1.44. Divide the satellite data by this factor. Here’s your comparison:
http://www.woodfortrees.org/plot/gistemp/from:1979/offset:-0.125051/mean:12/plot/hadcrut4gl/from:1979/offset:-0.021648/mean:12/plot/uah/from:1979/offset:0.229543/scale:0.69526565634609873044979098651821/mean:12/plot/rss/from:1979/offset:0.122699/scale:0.69526565634609873044979098651821/mean:12
Huh. The trends differ!
timetochooseagain says:
January 29, 2014 at 2:26 pm
“No, can’t do that either, not if there may be spurious biases in the data.”
Well lets say you have a new measuring instrument with unknown properties. How would you go about calibrating it so that it produces reliable figures?
Lets say you cant touch it or place known ice/water melting points around it, just take its output and compare it to your reference set.
Apply common heating/cooling to both and….. OLS align the two sets of readings by range and scale and voila you have a calibrated instrument.
I’m just doing the same 🙂
RichardLH says:
January 29, 2014 at 2:16 pm
Nobody said that thermometers are not useful for approximations. However, the idea that one temperature per room gives a “perfectly good” average air temperature brings up the question, “good for what”?
Now, people are claiming an average temperature for the entire globe good to a tenth of a degree … do you think one thermometer per room will give you that accuracy for your house?
However, that’s not the main problem. What you seem to be missing is the difference between an extensive property and an intensive property. For example, mass is an extensive property. If you double the amount of something (the extent), then the value of all extensive properties will double.
Temperature, in the other hand, is an intensive property. If I get one glass of water from the tap, it has the same temperature as two glasses of water from the tap, despite the fact that the mass has doubled.
The problem with intensive properties is that inherently they can’t be averaged. Or rather, they can be, but the answer may be meaningless. Color is an intensive property, but the question “what is the average color of the American flag” doesn’t yield meaningful information.
The reason that intensive properties can’t be averaged is that we quickly leave reality. Let’s average 4 masses, of 20, 30, 30, and 40 kg, To do that, we first find the total mass, which is 120 kg. So if we put them all together, that’s their mass … and then we divide it by 4.
Now lets try the same with temperatures of 10°C, 20°C 30°C and 40°C. We add them up and proudly say “The total temperature of the four items is over boiling, its one hundred and twenty degrees.” So if we put them all together, that’s their temperature … you see the problem? YOU CAN’T ADD TEMPERATURES TOGETHER because the result is meaningless, it says nothing about the real world to say their total temperature is over boiling … which means that you can’t average temperatures either.
All the best,
w.
@Willis Eschenbach-Yup, John Christy *has* written on the spurious shift of RSS relative to UAH. Here is the relevant paper:
http://journals.ametsoc.org/doi/abs/10.1175/JTECH1937.1?prevSearch=christy&searchHistoryKey=
Note what they say about the 1992-1994 period (the transition of the addition of NOAA-12 into the data.
@RichardLH-“Well lets say you have a new measuring instrument with unknown properties. How would you go about calibrating it so that it produces reliable figures?”
I sure as heck wouldn’t try to make it agree with another instrument, measuring a different thing, that might be wrong.
I explained something you could do instead. I showed what that comparison looks like.
timetochooseagain says:
January 29, 2014 at 2:26 pm
OLS aligned data sets from the various sources over their overlap period.
http://i29.photobucket.com/albums/c274/richardlinsleyhood/OLSTracesaligned_zps6d702275.png
timetochooseagain says:
January 29, 2014 at 2:38 pm
“I sure as heck wouldn’t try to make it agree with another instrument, measuring a different thing, that might be wrong.”
They are supposed to tracking the same thing, the change in Global Temperature since 1979. Even if they are using different parts of the same system, unless you want to suggest some part is warming/cooling at a different rate than the other…. OLS seems like the way to go.
TimC says:
January 29, 2014 at 3:15 am
– – – –
But shouldn’t it have been “What me worry”? …!
===
I skipped a few score posts – but I, too, recognise the Alfred E. Neumann [sic] thread.
Auto
@RichardLH-Look at your own chart. Look how much larger the swings in LT data are relative to surface data. Do you see you’ve done something wrong here?
“They are supposed to tracking the same thing, the change in Global Temperature since 1979.”
No! They aren’t! For the like, nine billionth time!
“unless you want to suggest some part is warming/cooling at a different rate than the other”
You mean like what happens *every* time there is an El Nino, *every* time there is a La Nina, *every* time there is a volcanic eruption? No, gosh, assuming that they would change at different rates is just silly, even though they literally change at different rates all the time.
And guess what? Turns out that theory and models also agree that they should change at the same rate. Gosh, who’d a thunk it, huh?
And theory and models agree with the comparison of interannual variations: the troposphere is more variable-and should warm and cool more-than the surface. Wow, *really?* I’d have never guessed that should happen, even though it happens all the time.
And then if I take the fact that that happens all the time, and use *that* to cross calibrate them:
http://www.woodfortrees.org/plot/gistemp/from:1979/offset:-0.125051/plot/hadcrut4gl/from:1979/offset:-0.021648/plot/uah/from:1979/offset:0.229543/scale:0.69526565634609873044979098651821/plot/rss/from:1979/offset:0.122699/scale:0.69526565634609873044979098651821
Gosh, how *completely different* the picture looks!
RichardLH-”Well lets say you have a new measuring instrument with unknown properties. ”
Well if it has “unknown properties” it can hardly be used to measure anything !
Willis Eschenbach says:
January 29, 2014 at 2:34 pm
“The reason that intensive properties can’t be averaged is that we quickly leave reality. Let’s average 4 masses, of 20, 30, 30, and 40 kg, To do that, we first find the total mass, which is 120 kg. So if we put them all together, that’s their mass … and then we divide it by 4.”
Mass is an extensive property!
Willis writes:
“Now lets try the same with temperatures of 10°C, 20°C 30°C and 40°C. We add them up and proudly say “The total temperature of the four items is over boiling, its one hundred and twenty degrees.””
10+20+30+40=100, but that’s the sum, the average is 25:
http://wiki.answers.com/Q/What_is_the_average_of_10_20_30_and_40?#slide=1
timetochooseagain:
I make technical points and arguments and you have replied at January 29, 2014 at 1:07 pm with bluster and self-refuted assertion.
I wrote
and you have replied with this misrepresentation
I said the the determinations are not of A DEFINED METRIC. I did not say the weighting profiles are not well defined. And I DID say the the weightings are arbitrary. Simply what I said is true, and what you pretended I said is bollocks.
I said
and you have replied
It means there is no possibility of calibration for the indicated results of temperatures of atmospheric layers. The accuracy and precision of a measurement cannot be known in the absence of a calibration standard. Instrument error may be assessed on the satellites but that is only one source of error. Indeed, you stated this in a previous post at January 29, 2014 at 12:46 pm where you wrote
I made no mention of the trends: your mention of them is a red herring.
As you say, the RSS and UAH can be adjusted to agree with the radiosonde data and that gives some confidence in the MSU data from the satellites, but it does not afford calibration for the compiled layer temperatures for the globe and hemispheres because the balloons have limited coverage.
I wrote
and you have replied
OK. We agree they “measure something different” and, therefore, they aren’t *supposed* to be” indicative of the surface-derived data. So, why the explanation mark when we agree?
Importantly, how does one compare them “intelligently” when they “measure something different”, and there is no clear definition of the GASTA determinations, and those determinations often change? What possible “theory and models” enables that comparison, guesswork or prejudice?
And you conclude saying
Well, disagreement with your own misrepresentations does seem to be your forte.
Measured data need to be reliable and have known accuracy and known precision. The UAH and RSS data are reliable because they can be compared to the balloon data, but they have no known accuracy and precision because they are not calibrated and are known to drift.. Hence, the UAH and RSS data can be uuseful but are not a true measurement.
The various versions of GASTA are not reliable – they change because their definitions change almost monthly – and they have no known accuracy and no known precision have no because they are not calibration. They have potential use, but at present their use is likely to mislead.
You seem to think bluster is an alternative to rational argument and information. I could not disagree more.
Richard
OOPS!
I wrote
The various versions of GASTA are not reliable – they change because their definitions change almost monthly – and they have no known accuracy and no known precision have no because they are not calibration.
but I intended to write
The various versions of GASTA are not reliable – they change because their definitions change almost monthly – and they have no known accuracy and no known precision because they are not calibrated.
Sorry.
Richard
“Mass is an extensive property!”
I think that was his point.
Now why are we interested in these “averaging temperatures” ? Because we then look at how the “average temp” has changed over time and decide whether the world is getting warmer and whether we need to worry about it. Then we talk about radiative “forcing” of some gases and how sensitive said average temp is to radiative change.
So what we are really trying to do is measure energy content, not temperature, and that is extensive.
So the question becomes how good a proxy is temperature for energy content. The first thing that becomes apparent is the need to looking at similar masses for each sample. If we add two pots of water one at 10C and the other at 20C, we have no idea of the resultant temperature (the average temperature) unless we know the volumes. If we add two pints of water at those temps we can predict the answer.
The first thing to notice is the need to area weight the temperature grid cells
Then SST may tell us something about energy content outside of oceans that are ice buckets, where we need to look at ice area, or more correctly volume.
Air is more tricky because of pressure and humidity. Since meteorological measurements are done by pressure level not altitude , the pressure angle roughly taken care of. Humidity remains a confounding variable.
However, from that it quickly becomes obvious that surface air temperature where pressure and hence density varies does not work. Averaging air temps really does not work too well.
The flip side of the physics is that most people live in surface air temperature and it is what ultimately “matters”. Number of people affected is an extensible property and can be averaged.
“””””…..Willis Eschenbach says:
January 29, 2014 at 1:50 pm
george e. smith says:
January 29, 2014 at 11:14 am
… Howcome; on the same scales the north pole and the south pole have whacking great peak real signal amplitudes, and everywhere else has ho-hum real signals ??
Interesting question, George. I was surprised as well. Upon reflection I concluded that there were several reasons. 1) Size 2) Cohesions 3) Lack of GHGs……””””””
Well your theory is testable Willis; since you have all the data.
What if you re-divide your earth into five EQUAL AREA zones, instead of the unequal areas you used. You know how to do that; you just divide the polar diameter of the earth by five, and make each zone that tall (about 1584 miles). A cylinder circumscribing a sphere has the same area as the sphere (2pi.R x 2R =4pi.R^2) And any length segment of the cylinder, has the same area as the underlying surface of the sphere, even the polar caps.
I think that’s on page IIIL in Euclid; maybe it’s page XXIX. Too long since I did it.
Not that I want to load you up with work; but you’re not doing much of anything anyhow; are you ??!
But such a change might scramble the “local” climate of the zones compared to yours, and muck up the effect you were looking for. Well just a thought.
g
harrydhuffman:
I don’t believe your definition of “emergent phenomenon” is consistent with common usage. You say that an “emergent phenomenon” is one that is changing. In the article at http://redwood.berkeley.edu/w/images/5/5e/Scott2004-reductionism_revisited.pdf, though, an academic mathematician explains that an “emergent phenomenon” is one in which a portion of the cause and effect relationships are non-linear.
What is striking about figure 1 is the huge increase in variation amplitude at the poles.
Is part of this a limitation of satellite data? Is the satellite geostationary? If so, does the acute angle of observation of the poles and the longer path through the atmosphere lead to higher error variation?
Thus, is all this variation real?