Arguments
New study finds a hot spot in the atmosphere
Posted on 15 May 2015 by John Abraham
A new study, just published in Environmental Research Letters by Steven Sherwood and Nidhi Nashant, has answered a number of questions about the rate at which the Earth is warming. Once again, the mainstream science regarding warming of the atmosphere is shown to be correct. This new study also helps to answer a debate amongst a number of scientists about temperature variations throughout different parts of the atmosphere.
When someone says “The Earth is warming”, the first questions to ask are (1) what parts of the Earth? and (2) over what time period? The Earth’s climate system is large; it includes oceans, the atmosphere, land surface, ice areas, etc.
When scientists use the phrase “global warming” they are often talking about increases to the amount of energy stored in oceans or increases to the temperature of the atmosphere closest to the ground. By either of these measures, climate change has led to a progressive increase in temperatures over the past four decades. But what about other parts of the climate system? What is happening to them?
One important area to consider is the troposphere. It is the bottom portion of the atmosphere where most weather occurs. Tropospheric temperatures can be taken by satellites, by weather balloons, or other instruments. In the past, both satellites and weather balloons reported no warming or even a cooling.
However, that original work was shown to be faulty and now even the most strident sceptics admit that the troposphere is warming. But obtaining an accurate estimate of the rate of warming is difficult. Changes to instruments, errors in measurements, short term fluctuations all can conspire to hide the “real” temperature.
This is where the new study comes in. The authors develop a new method to account for natural variability, long-term trends, and instruments in the temperature measurement. They make three conclusions.
First, warming of the atmosphere in the tropical regions of the globe hasn’t changed much since the late 1950s. Temperatures have increased smoothly and follow what is called the moist-adiabatic rate (temperature decrease of humid air with elevation). This result is in very close agreement with climate computer models and it contradicts the view that there is a slowdown in climate change.
Second, the vertical height of the tropics that has warmed is a bit smaller than the models predict. Finally, there is a change in observed cooling in the stratosphere – the layer of the atmosphere above the troposphere.
Taken together, these results show that the tropospheric warming has continued as predicted by scientists years ago.
Embedded in this research is a conclusion about the so-called “tropospheric hot spot”. This “hot spot” refers to expectations that as global warming progresses, the troposphere will warm faster than the Earth surface. The hot spot is really hard to detect; it requires high quality measurements at both the surface and throughout the troposphere. Past studies which could not detect a hot spot were often used by climate contrarians to call into question our simulation models and even our basic understanding of the atmosphere.
But this new study finds a clear signal of the hotspot. In fact, the temperature in the troposphere is rising roughly 80% faster than the temperature at the Earth’s surface (within the tropics region). This finding agrees very well with climate models which predicted a 64% difference.
And this is exactly how models are supposed to work. Models can be used to predict changes that will occur in the future. Once we make measurements, we can compare them with the models. If the two disagree, it either means our models are wrong, our measurements are wrong, or both are wrong. More often than not, the models have been found to be vindicated.
Mike Mann has an excellent description of this new study's relevance to AGW. In brief,
Interesting point in the comments at HotWhopper, that the term "tropospheric hot spot" seems not to be used scientifically. Implications of failure to use that term are more than merely semantic.
Tom Dayton @1, in this case Michael Mann is wrong. As the RealClimate team have explained, the tropospheric hotspot is a result of the lapse rate feedback, ie, the change in decrease in temperature with altitude due to the increased heat capacity of the atmosphere resulting from a higher absolute humidity with increased warming. The RealClimate team illustrate this by comparing expected warming and cooling by latitude and altitude for a doubling of CO2, and an equivalent increase in solar forcing:
The patterns are broadly similar, although the CO2 forcing results in a stronger arctic warming (but note the difference in scales).
Unfortunately, it is not possible to similarly compare the result of CO2 forcing to a loss of background volcanic aerosols as the background level of aerosols is not that large. We can however compare the cooling due to a loss of 50% of CO2, and due to a 1/3 rd Pinatubo level volcanic forcing:
As you can see, for these negative forcings, the patterns in the tropics are very similar. At the poles, however, the increased volcanic aerosols result in stratospheric cooling, unlike the stratospheric warming resulting from reduced CO2. Further, the peak arctic cooling is further south with volcanic aerosols. Again the scales are different, but the forcings are not SFAIK of equivalent strength. The patterns, however, are the important feature.
The polar stratospheric cooling with increased volcanic forcing may be partly due to the location of the aerosol cloud, which is not global (and was tropical in Pinatubo's case) and ergo will have different regional effects. It is, however, at least partly due to a decreased warming of the stratosphere from upwelling IR radiation due to the cooler troposphere not being compensated by increases short wave radiation reflected of the stratospheric aerosols due to the low insolation at the poles. Ergo, it is at least in part a feature we would expect of any volcanic cooling, although the exact boundaries of warming and cooling zones will vary depending on the latitude of the volcano causing the cooling.
The important thing about this comparison for our discussion is that there is definitely a tropospheric cool spot for both reduced CO2 and the volcanic forcing. As can be seen by comparing the doubled CO2 and halved CO2, the patterns of change in temperature are the same for both. It is just that the signs of the change are different. The same would be true for a reduced volcanic forcing. That is, it would show a similar pattern to the increased volcanic forcing, only with warming where the volcanic forcing shows cooling, and vice versa. Ergo a reduction of background volcanic aerosols would definitely result in a tropospheric hotspot, contra Mann.
I believe Mann has been decieved on this because:
1) He fails to note that the observed heating for Pinatubo through radiative absorption and re-emission of the aerosols is primarilly above the 100 mb line, ie, the tropopause:
2) And because he focuses only on radiative heat transfer, ignoring the convective heat transfer effects that drive the change in lapse rate.
Tom Curtis, I asked Mike Mann to reply to your comment. Here's what he wrote over on his FaceBook post:
Mike Mann on FaceBook followed up in his reply to Tom Curtis's above comment:
Tom Curtis, I'm being go-between you and Mike Mann, because though he is popping over here to read your comments, he can't post here because he does not have an SkS account and doesn't want to get one because he already is too busy. (I don't buy that excuse, because I know the only way he can accomplish all the things he already accomplishes is by having an army of clones of himself. What's one more clone?) He wrote on FaceBook:
Tom Curtis: Mike Mann continued at FaceBook:
Tom Dayton @4-7:
1) The model used was the GISS Model E vIII, as can be found out by tracing back the links through the Real Climate link I provided. As such I suspect it is the same model that was used for the second figure in Mann's original article, with Mann's figure differing in being rapped around the hemicircle, and not being to scale relative to the real thickness of the atmosphere. Importantly, Mann's figure also differs in showing the effects of historic forcings (which are very different in magnitude between the various forcings) rather than the effects of forcings of approximately equivalent magnitude:
2) Santer et al (2013) shows historical forcings from Jan 1979 to Dec 2012. Santer's graph shows an overall positive volcanic forcing, with a negative forcing in southern polar latitudes, and no forcing in northern polar latitudes. That is, it shows a very different regional pattern (and a far smaller forcing) than does Pinatubo. Consequently I would hesitate to draw any inference from the differences in pattern between GISS model E and the CMIP5 mean. Such differences are as likely to be due to the different geographical pattern of the forcing as to any difference in model physics.
3) Each of the TMT (Channel T2), TTS (Channel T3), TLS (Channel T4) include a portion of the lower stratosphere in their weighting functions:
The result is that all are downgraded by opposite signed signals contaminating the data. This is particularly the case for the TMT (T3) channel which obtains approximately a third of its data from the stratosphere, and hence with opposite sign to the upper tropospheric signal. (This is also why certain deniers prefer to use the TMT channel as representative of tropospheric temperature trends.) As Santer et al graph by anomalies for these channels, I presume they have taken a weighted temperature function by altitude to match the weighting of the satellite channels, and will suffer the same problem. This by itself would lead us to expect a reduced TMT temperature response relative to the TLT response as shown by Santer et al, even with the GISS Model E temperature response.
4) Beyond that, it is possible that the CMIP5 models show a different response to volcanic forcing, and that consequently the Real Climate article and GISS model E data are out of date. If Mann could get Gavin Schmidt to comment on that point, I would be very interested. As it stands, however, I do not find his counter argument convincing, and nor is this topic close enough to Mann's core areas of expertise that I would take him as an authority on this aspect of AGW.
Tom Curtis: Mike Mann read your most recent response, and replied:
Mike Mann updated his post on the hot spot. This is the last time I'm going to copy over his posts and comments. Please go to his FaceBook page if you want to read and comment more.
Figure 2:
Tom Curtis @3,
Looks like Mike Mann did not even change the details of his hypothesis based on your input. And as it turns out MM hypothesis is backed by Ben Santer, your inquiry to the authority of Gavin Schmidt on the subject is unlikely to change anything. Meanwhile Chris Colose offered to run GISS model you claim about volcanic hot spot is based on, to run it and check the apparent discrepancy: we are all eager to see the results, thanks Chris!
Meanwhile, pardon my punt, I don't see the extraordinary evidence you need to provide in support of your extraordinary claim taking on the top scientists. If particular, your third image is just a hot spot cooling due to a loss of 50% of CO2, as seen in the NASA link you cite. Where is the "1/3 rd Pinatubo level volcanic forcing" picture where you claim "the patterns in the tropics are very similar"? Maybe you've seen but missed that picture. Please provide it for the benefit of us better understanding your point.
Tom Dayton @9 & 10, Michael Mann writes:
My emphasis.
Below is a figure showing the trend warming rates from 1979-2012 in the troposphere by altitude for all CMIP5 RCP8.5 models:
The source of the diagram is John Christy, which is dubious in some respects, but I presume he has not misrepresented the model outputs. That being the case, the CMIP5 models show a clear upper tropospheric hotspot with a multimodel mean peak warming at 250 mb. Despite this the trends from 1979-current are:
GISS 0.158 C/decade
RSS TLT 0.121 C/decade
RSS TMT 0.077 C/decade
RSS TTS 0.011 C/decade (from 1987)
RSS TLS -0.269 C/decade.
Clearly the trends decline as the mean altitude of measurement rises, with the TMT trend being less than either the TLT or surface trends, and greater than any trends above it. It follows that if Mann's claim highlighted above is true, then there is no tropospheric hotspot contrary to the model predictions. In other words, if Mann's highlighted claim is true, the models have been falsified on a fundamental issue that is vital to estimates of climate sensitivity and future warming. Michael Mann may want to claim in response that his claim is only applicable to volcanic warming, and not warming in general. If so that is just the rankest special pleading. If not he either has to accept my claim @8, point (3), above, or radically revise his reliance on models, and his view as to the dangers of AGW.
Further, the GISS Model E vIII 1880-2000 all forcing response by altitude is shown below:
As you can see, the peak warming is around 350 mb (337.5 mb). That is in fact lower than the CMIP5 multimodel mean, and also lower than the CMIP5 versions of the GISS Model E (six versions, shown with dotted lines of various shades in the first figure above). That discrepancy may be simply due to the different forcing periods, but if anything the evidence is that the altitude of peak warming has shifted upwards from CMIP3 to CMIP5, not downwards as would be required for the GISS model E images I have been using to be in need of significant correction.
As an aside, there is certainly no reason to prefer the data from cone CMIP3 era model, ie the Parallel Climate Model (PCM) as shown in the IPCC reports and used by Mann over another model of the same vintage (GISS Model E vIII). As a further aside, the peak cooling for the one third Pinatubo forcing shown @3 is also at 337.5 mb). This point should be irrelevant in that I originally rebutted Mann's claim that, "if global warming really *were* due to a (natural) decrease in volcanic activity over time ... then we would expect to see an increase in global surface temperatures WITHOUT any mid-tropospheric "hot spot"". Mann now appears to be trying to make the issue about the exact altitude of peak warming, which shifts the goal posts. He also misrepresents the altitude of the peak warming (or cooling) in the GISS model, claiming (@10) it is about 140 mb higher in the atmosphere than actually shown by the model.
3) Fairly abviously, the PCM models shown @10 do not use a period with significant volcanic forcing, with the consequent that any coldspot is too small to register for the volcanic forcing given scale. The same also applies with sign reversed for the solar forcing. That is telling in that solar forcing is definitely one of the forcings which does show a hotspot, but shows no hotspot in the PCM figure. As Mann accepts that solar forcing generates a hotspot, he must attribute the lack of a visible hotspot in the solar forcing panel to the small quantity of warming relative to the temperature scale used. He cannot therefore consistently argue that that is not also the case with respect to the volcanic forcing.
4) Finally, with respect to figure 2 from Santer et al (2013) (see your post @6 above), Santer et al write:
My emphasis.
It is not explained what "MSU Space" is, although it is almost certainly not a linear function of temperature with altitude. More likely, at each altitude, the value shown is the weighted average of the TMT, TLS and TLT trends, with the weighting determined by the relative weight of each channel at that altitude as shown in my post @8, although different algorithms with similar effect could also be used. As such, it loses vertical structure. That is because it first reduces the vertical structure to just three values, and then tries to recompose it from those three values. It is, in effect, a complex smoothing of the data. As such it will not more show the tropospheric hotspot than will the TMT channel (for reasons given above). What it will show is what will be found by attempts to reconstruct the vertical temperature signal from the MSU or AMSU channel outputs. That, of course, is very useful for comparison with satellite data, but renders the graphs positively misleading about the detailed vertical temperature structure of the atmosphere.
chriskoz @11, in my comment @3, the first two images are the images used in the RealClimate post to show that both CO2 and solar forcing result in a tropospheric hotspot. The third image is, as you show the effect of halving CO2 (after 100 years). The fourth image, for comparison, is the effect of maintaining 1/3rd Pinatubo forcing for 100 years.
I would certainly be interested in, and gratefull for, Chris' rerunning of the GISS model for the equivalent experiment.
I cannot comment on Ben Santer's input as I have not seen it. Could you provide a link. Further, if he is commenting on the topic somewhere, and you can respond, could you ask his opinion of my point 4 immediately above. Specifically, what is the algorithm for "MSU Space"? Does converting to it have the effect I postulate? Also does he have any graphs of absolute temperature values with altitude in MSU space as the test of the effect of using MSU space?
Yay, my first post. As most people here probably know, this isn't the first study to find the hot spot. The hot spot has been found since at least 2004 in the NOAA satellite data analysis. So I wanted to give a brief list of studies that found the hot spot. This list is by no means exhaustive, since it does not include at least 6 other papers that provided evidence of the hot spot. But it should be a helpful list nonetheless.
Here's the list, along with the data sources for the papers (I think your article is on paper #6):
In satellite data:
#1 : "Contribution of stratospheric cooling to satellite-inferred tropospheric temperature trends"
#2 : "Temperature trends at the surface and in the troposphere"
#3 : "Removing diurnal cycle contamination in satellite-derived tropospheric temperatures: understanding tropical tropospheric trend discrepancies", table 4
#4 : "Comparing tropospheric warming in climate models and satellite data", figure 9B
In radiosonde (weather balloon) data:
#5 : "Internal variability in simulated and observed tropical tropospheric temperature trends", figures 2c and 4c
#6 : "Atmospheric changes through 2012 as shown by iteratively homogenized radiosonde temperature and wind data (IUKv2)", figure 1 and 2
#7 : "New estimates of tropical mean temperature trend profiles from zonal mean historical radiosonde and pilot balloon wind shear observations", figure 9
#8 : "Reexamining the warming in the tropical upper troposphere: Models versus radiosonde observations", figure 3 and table 1
In re-analyses:
#9 : "Detection and analysis of an amplified warming of the Sahara Desert", figure 7
#10 : "Westward shift of western North Pacific tropical cyclogenesis", figure 4b
#11 : "Influence of tropical tropopause layer cooling on Atlantic hurricane activity", figure 4
#12 : "Estimating low-frequency variability and trends in atmospheric temperature using ERA-Interim", figure 23 and page 351