- JoNova - https://www.joannenova.com.au -

Dessler 2010: How to call vast amounts of data “spurious”

This is part of the big PR game of publishing “papers.”

In the climate models, the critical hot spot is supposed to occur because (specific) humidity rises in the upper troposphere about 10km above the tropics. The weather balloons clearly show that temperatures are not rising as predicted, so it was not altogether surprising that when Garth Paltridge analyzed weather balloon results for humidity, and found that humidity was not rising as predicted either.

Indeed, he found specific humidity was falling, which was the opposite of what all the major climate models predicted and posed yet a another problem for the theory that a carbon-caused disaster is coming. He had a great deal of trouble getting published in the first place, but once he finally did get published and skeptics were starting to quote “Paltridge 2009”, clearly, Team AGW needed an answer.  “Dessler 2010” is transparently supposed to be that answer.

To start by putting things into perspective, lets consider just how “spuriously” small, patchy and insubstantial the radiosonde measurements have been. According to NOAA The integrated Global Radiosonde Archive contains more than 28 million soundings, from roughly 1250 stations.

Worldwide radiosonde stations (NOAA)

Or how about the data from just one month.

Radiosondes kept in 200911
ARSA : Location and number of radiosonde reports
selected for the month of November 2009
(Click the figure to enlarge it or click here)

The ARSA site estimates there are around 40,000 radiosondes released each month. (While the NOAA site above suggests it’s been closer to 50,000 per month. And it’s been something roughly like that every month since 1958).

The radiosonde results are uncertain, but they suggest specific humidity is falling (and the error bars all fall in the negative range). So how do you dismiss that wall of data going back for half a century?

Satellites record humidity too, and they suggest specific humidity hasn’t changed much, or has been slightly positive. So if you collect studies of satellite data and stick those results on a graph, lo and behold, the only study that exclusively uses radiosonde data suddenly looks like an “outlier”.

There are problems with radiosonde data — especially humidity measurements. But satellite channels have equally large (if not larger) uncertainties. They can’t pick out humidity specifically, and they can’t resolve say, 10 km from 11km. They cover thick slabs of the atmosphere (Channel 12 gets results from the region that is 8 -12 km up).  In the end the only thing we know for sure is that it’s hard to measure humidity 12 km off the ground.


The Dessler 2010 paper in Journal of Geophysical Research (JGR) describes Paltridge’s results as “spurious”. JGR did ask Paltridge to comment, which he did, but Dessler did not resolve Paltridge’s criticisms (see some of them below), and JGR published anyway.

JGR let some decidedly unscientific things slip into that Dessler paper. One of the reasons provided is nothing more than a form of argument from ignorance: “there’s no theory that explains why the short term might be different to the long term”. Why would any serious scientist admit that they don’t have the creativity or knowledge to come up with some reasons, and worse, why would they think we’d find that ignorance convincing? (Though it does appear to have impressed John Cook of the not so skepticalscience).

It’s not that difficult to think of reasons why it’s possible that humidity might rise in the short run, but then circulation patterns or other slower compensatory effects shift and the long run pattern is different. Indeed they didn’t even have to look further than the Paltridge paper they were supposedly trying to rebut (see Garth’s writing below). In any case, even if someone couldn’t think of a mechanism in a complex unknown system like our climate, that’s not “a reason” worth mentioning in a scientific paper.

Andy Dessler is so convinced the models are right, he flatly declared it back in January on Piekle’s blog, even going so far as to say water vapor feedback was strong, positive and announcing that it was “unequivocal”. Ponder what “unequivocal” means when millions of weather balloons  travelled up through the air measuring humidity and repeatedly found something that was the exact opposite of what he says. Only a religious fanatic would call that “unequivocal”.

Instead, in the real world, most of the data and observations suggest that the feedbacks are not strongly positive. In the ice cores there’s a long lag (where CO2 rises and falls after temperatures) and there no definitive evidence coming out of the best ice core data of “amplification”. The climate models (which Dessler claims are independent) are programmed to have positive feedback, so they could hardly show anything else. That’s circular reasoning.

Furthermore, both satellites and weather-balloons observe the tropical upper troposphere is not warming faster than the surface (the climate models predict warming — the hotspot). If specific humidity was increasing significantly in the upper troposphere, wouldn’t that produce a long term warming trend?

The peak effects of the water vapor amplification are supposed to occur between 200-300hpa, and even the satellite data doesn’t show much support for that.

Guest Post by Garth Paltridge

Re The Dessler and Davis Paper

Its primary aim is to make the point that, in the authors’ opinion, the negative NCEP trends (reported in Paltridge et al – i.e. in PO9) in water vapour concentration in the middle and upper troposphere are spurious – and therefore that overall water vapour feedback in the climate system is indeed positive and is an amplifier of global warming.  It supports this opinion with:

(a) A simple comparison of the NCEP trends with the trends from four other re-analysis data sets.  The comparison indicates that the three most recently developed re-analyses have positive trends in the middle and upper troposphere.

(b) Correlations (from each of the re-analyses) between middle and upper level water vapour concentration qmu and surface temperature T.  These show that, at least for overall short-term change on time scales less than 10 years, all the correlations are positive and thereby indicate positive feedback.

(c ) Comment to the effect that the concept of long-term positive feedback is in accord with virtually all of the independent lines of evidence (models, observations, theory, and newer re-analyses).

RE (a):  A central issue is that the NCEP re-analysis relies on balloon observations of water vapour, whereas all the other re-analyses use satellite data of one form or another.  The other re-analyses may or may not use balloon data as well as satellite data.  The ERA40 re-analysis certainly includes balloon data, and that particular re-analysis also has a tendency toward negative trends of middle and upper level water vapour.

Raw balloon data (tropical data as examined in PO9 for instance) seem to show a negative long-term trend of water vapour at the upper levels.  Raw satellite data (as mentioned in PO9 and the Dessler paper for instance) seem to show a positive trend.

So the most likely and straightforward explanation of the difference between the outputs of the re-analysis schemes is that they (the schemes) are actually behaving as they were intended – namely, they are simply reflecting the behaviour of their different sources of input data.  Which, if so, leads not to the question of whether some re-analyses are ‘newer’ than others, but to the question of which source of input data (balloon or satellite) has the greater potential for error, and whether, in either case, those errors would or could lead to the trends that are observed.   Any significant attempt to resolve such a question would have to consider not only the potential errors of both the balloon and the satellite information, but also the possibility that different sorts of satellite data have been introduced into the re-analysis schemes at different times over the 30-year period since 1979.

The bottom line here is that it is a bit infra-dig to give the impression that the NCEP re-analysis is a single “outlier” pitched against a number of other independent re-analyses and can therefore probably be discarded.  If one ignores the question of which is the more reliable as input data (satellite or balloon data), the balance of ‘likelihood of verisimilitude’ between NCEP and the others has to be more like 50:50.

And therefore it is also a bit infra-dig to talk fairly extensively about the well-known problems associated with balloon measurements and make no reference to the many and various problems associated with satellite data.   (I have heard it said for instance that it is difficult enough to believe past satellite data on trends of total water vapour content – let alone of the water vapour concentration at any particular level).

Re (b): Much of the Dessler and Davis discussion is devoted to the correlation between changes in upper-level water vapour qmu with changes in surface temperature T.  In particular it is devoted to the possibility flagged in PO9 that the long-term correlation may be negative even though the short-term correlation is positive.

The authors say that there is no theory to explain such a difference in the sign of the correlations.  Suffice it to say that PO9 contains a diagram and discussion concerning two (admittedly only qualitative) theoretical suggestions as to how such an eventuality might occur.  The suggestions concern possible causes of long-term increase in the stability of the lower atmosphere – an event which, according to the NCEP data, indeed seems to have occurred over the last few decades, and which, if real and continued, could confine a long-term increase of water vapour concentration to the convective boundary layer.  (One of the possible causes is the relatively large increase in radiative heating in the middle troposphere associated with increasing CO2).  This is not to say that such theories are correct, but the absence of any reference to them suggests a reluctance even to contemplate arguments on the other side of the fence.

The authors make the point that there is poorer agreement among the re-analyses about the qmu vs T correlation on time scales longer than 10 years than there is about the correlation on time scales less than 10 years.  They attribute this to “handling data inhomogeneities” having more impact on long-term trends than short-term trends.  Fair enough.  But they could also have pointed out for the record that, at least at face value, the slopes of the long-term correlations displayed in their diagram are generally a lot less than those of the short-term correlations, and that some of them are indeed negative at certain levels.  The bottom line here is that even a reduction of slope (if it were verified) would be very significant in the overall water-vapour feedback story.

Re (c):  Superficially the comment is impressive.  One wonders however what is this theory which is supposed to be an independent line of evidence – apart, that is, from the individual bits of theory so far built into the models.   And in view of the discussion above about whether reference to the “newer re-analyses” is really germane to the issue, one wonders also about the significance of those analyses in the present context.  And in view of the fact that the veracity of models relies (among many other things) upon the observations on which they are based, it is pushing things a bit far to say that models are truly “independent” evidence.  (Perhaps more to the point in the context of models, their long-term trends of qmu depend among other things on simulation of vertical sub-grid-scale diffusion – the contribution of which is one of the most difficult characteristics of models to ‘nail down’ and verify independently).

The bottom line here is that “virtually all the independent lines of evidence” probably boil down only to the observations.  And at the moment, bearing in mind the discussion with regard to (a) above, there is still a lot of work to be done to establish just what the observations are telling us.


Of course Dessler and Davis are entitled to their opinion, and may even be proved right one day.  But at the moment it won’t do the discipline much good if people assume simply on the basis of the existence of their paper that the issue is now resolved.


Dessler, A. E., and S. M. Davis (2010), Trends in tropospheric humidity from reanalysis systems, J. Geophys. Res.,
115, D19127, doi:10.1029/2010JD014192 [PDF]

Paltridge, G., Arking, A., Pook, M., (2009) Trends in middle- and upper-level tropospheric humidity from NCEP reanalysis data. Theoretical and Applied Climatology, Volume 98, Numbers 3-4, pp. 351-35). [PDF]

Thanks to Garth for being so helpful with advice and information.

Thanks to the ineffable wisdom of Rod Smith for giving me some appreciation of the formidable database amassed from weather balloons.

UPDATE: The numbers of weather balloons each month differ between groups and datasets, I’ve updated those figures to refect the two sources I list.

Tiny url: http://bit.ly/9Rpafb

6 out of 10 based on 10 ratings