Sunday, October 22, 2023

Something left unsaid about Koutsoyiannis et al.

The latest in here is a shiny statistical model that disproves everything you thought about greenhouse gases causing climate change is Koutsoyiannis Demetris, Onof Christian, Christofides Antonis and Kundzewicz Zbigniew W.2022  Revisiting causality using stochastics: 1. Theory Proc. R. Soc. A 4782021083520210835 and Revisiting causality using stochastics: 2. Applications Proc. R. Soc. A.4782021083620210836

Of course, the classic was the unit root and Bart Verheggen's weight loss program but recently we have Statistics Norway trying it on, and just a year ago another ground breaking article

Demetris Koutsoyiannis is a well established, tho emeritus, professor of hydrology at the National Technical University in Athens, Christian Onof a rainfall guy at Imperial College, Antonis Christofides, also at NTU Athens and Zbigniew W. Kundzewicz, a pretty well known rainfall/hydrology person. Add to this that the Proc. R. Soc. A, ain't MDPI and a bunny knows the sharks will gather

So no surprise to find them at Curry's, with the birth announcement (well, first year anniversary) on Twitter, and the usual pile in. 

Naomi Oreskes and John Cook have a lot to answer for, if it was not for them pointing out that there is very little serious literature disagreeing with the IPCC consensus poor Eli would not have to wade throught this stuff 

The TL:DR 
The remaining real-world case study led to an important side product of the current research. This is the surprising finding that, while in general the causal relationship of atmospheric T and CO₂ concentration, as obtained by proxy data, appears to be of hen-oregg type with principal direction 𝑇 → [CO₂], in the recent decades the more accurate modern data support a conclusion that this principal direction has become exclusive
and the problem is the usual one. As Eli, and others pointed out (there are some useful references)

 We know from measurements that more CO2 is being emitted by burning fossil fuels than remains in the atmosphere. We also have measurements showing that ~105 _+ 8 Pg C from these emissions has been absorbed in the oceans* and the biosphere has i** and we have measurements and theory on how temperature and salinity affect CO2 partial pressure in the gas phase above sea water*** where a difference of 1C corresponds at best to a few ppm.

Of course we know that the atmospheric concentration of CO2 has increased from ~290 to ~430 ppm over the last 150 years.

This poses some obvious problems for any statistical treatment which claims that CO2 concentration follows changes in temperature

*Feely, R. A., Sabine, C. L., Takahashi, T., & Wanninkhof, R. (2001). Uptake and storage of carbon dioxide in the ocean: The global co~ 2 survey. OCEANOGRAPHY-WASHINGTON DC-OCEANOGRAPHY SOCIETY-, 14(4), 18-32.

**Amthor, Jeffrey S. “Terrestrial higher‐plant response to increasing atmospheric [CO2] in relation to the global carbon cycle.” Global Change Biology 1.4 (1995): 243-274 and especially the annual cycle

***Weiss, R. F., Jahnke, R. A., & Keeling, C. D. (1982). Seasonal effects of temperature and salinity on the partial pressure of CO2 in seawater. Nature, 300(5892), 511-513

Poking about and tossed up this comment on Koutsoyiannis et al. by Leif Åsbrink in the january 11 2023 issue of PRSA. It says pretty much what everybunny has been saying
From the information in [1] we can therefore draw a second conclusion: For timescales in multiple decades, the results in figure 14 have no meaning since the data span a far too short time. The differentiation has also suppressed slow changes. Common perception that [CO2] causes an increase in the temperature can thus neither be verified nor falsified. It is self-evident that the effect ΔT → [CO2], which is seen on a short timescale, will also be present on all longer timescales. A temperature change in the order of 0.4°C will cause a change in [CO2] of about 2 ppm, as measured on the derivatives.3 The temperature change over the 42 years is about 0.5°C while [CO2] has changed by about 80 ppm. Some other mechanism is likely to dominate on timescales of multiple decades. Hence, the common perception that increasing [CO2] causes increased T seems likely. . . . . 
Settled science’ is supported by many lines of evidence and it includes many more factors than [CO2] as causes of the increased temperature since pre-industrial times. CO2 is however considered to be the largest factor. ‘Settled science’ includes the phenomenon that increased T causes a modest increase of [CO2] (outgassing from the sea), which is detected in [1]. The statement ‘in the recent decades the more accurate modern data support a conclusion that this principal direction has become exclusive. In other words, it is the increase of temperature that caused increased CO2 concentration’ that appears in Discussion and conclusions in [1] is not supported by proper arguments in the paper.

This comment is notable for two things, all the references are to websites (inc. Roy Spencer’s blog) and there is no reply by Koutsoyiannis. 

Also, Eli can't find it being discussed most anywhere at all and certainly not at Curry's. The editors at Proc. T. Soc. A seem to have tossed Koutsoyiannis under the bus.

1 comment:

  1. Who needs a bus when you can comment at Judy's

    demetriskoutsoyiannis | September 26, 2023 at 3:23 pm | Reply
    Good question… Difficult answer…. So, we preferred different phraseology (and different foundation of our math) as you may see in the paper.

    Russell | September 30, 2023 at 4:39 pm |
    Argumentum ad ignorantiam is a bad opening move. You aver that the standard Kilo in Paris

    ” has been found that its mass changes over time by something like 0.000005% per century, and no-one knows why exactly.”

    The loss is due to two mechanisms, one physical and one chemical. Despite care in handling, the platinum -iridium cylinder is not immune to mechanical and atmospheric erosion. At the time of the metric standards manufacture, zone refined high purity ( 99.9999+%) platinum and iridium did not yet exist.

    The Victorian era mass cylinders and standard meters accordingly contain amongst other impurities ,hundreds of parts per million of osmium, which slowly oxidizes in air to form highly volatile OsO4.

    As a general rule, reviewers who check facts rigorously stop short at the first counterfactual they encounter. Yours is on Page 1 Para3.

    climategrog | October 6, 2023 at 6:34 pm |
    I have to agree with Russell on this, plus the long rambling intro about M&M etc. is a tedious way to make a point which can be easily stated clearly.

    I don’t see why you choose different averaging periods for your two variable. Use Jan-Dec for both. If you want to find the lag with the highest correlation, do a cross-correlation analysis.

    Also, dividing a temperature by “Kelvin” to pretend it is a dimensionless quantity you can subtract from ln(CO2) is not credible physics. What is “temperature per kelvin” supposed to be physically? Sorry, this is silly. This question of causality is crucially important and needs to be done competently. Please try again.


Dear Anonymous,

UPDATE: The spambots got clever so the verification is back. Apologies

Some of the regulars here are having trouble telling the anonymice apart. Please add some distinguishing name to your comment such as Mickey, Minnie, Mighty, or Fred.

You can stretch the comment box for more space

The management.