Only few records are going back to 1850 or earlier:

• monthly temperature (HadCRUT4)
• monthly Atlantic multi-decadal oscillations (AMO),
• daily and monthly counts of solar spots (Spots),
• monthly global sea level,
• yearly and then monthly CO₂ atmospheric concentration, with monthly interpolation of earlier yearly data (CO2),
• Methane and nitrous oxide on 5 year intervals since 1750 and yearly since 1978. But for large parts this data series is a calculated reconstruction since no such chemical analysis was be performed over the whole time period.

And others are only available since the 1950’s:

• MEI, the El Niño-La Niña index in the Pacific Ocean (MEI),
• Transmittance of the solar irradiation that shows significant dips following volcanic eruptions (Trans).

Long enough data series of other significant parameters are not available. Other similar data series have been constructed, but are variations, or precisions, or small alterations of what has been selected here. Therefore, if valid correlations can be found they could also be obtained with other data series presenting similar patterns. The data used in this evaluation is already the result of intense massaging, although it can be considered as valid input, with (hopefully) no inter-correlation already built in the massaging process.

Considering the billions of dollars poured into climate research this is a rather poor harvest for a time period much shorter than a geological era.

It is however quite helpful to use these data since each series represent an essential element of the system:

• T as the parameter to follow,
• AMO and MEI, as on-going disturbances,
• Solar spots and Transmittance as variation of energy input,
• Sea level (GSL), as an on-going natural ramping parameter initiated before BIE,
• CO₂ , as the smoking gun, representative of all radiative forcings, whatever their origin.

Calculated data are available for methane and N₂O, the source of which having both natural and artificial. Since a linear relationship exist with CO₂, it suffice to use the latter as a proxy[1] for all greenhouse gases.

All data used in correlation attempts.
Temperature anomaly, AMO, MEI, Transmittance on the left scale.
Solar spots, CO₂ , CH₄ , and N₂0 on the right scale.

No conclusive cause-effect relationship can be derived from the scarce climate data at hand. Although it is possible to correlate global temperature evolution with carbon dioxide concentration, well fitted regression formula can also be obtained when ignoring this parameter. Faced with such findings, prudence requires refraining from using any singularly preferred phenomena to explain and quantify climate mechanisms, while others may be as significant, if not more.

Obtaining reliable records is already a challenge because data massaging cannot be avoided. The time span over which they are available is at best 2 centuries. The rest is second hand reconstruction using proxies; in this case, uncertainties are not only in the exactness and precision of the physical values themselves, but also in the time scale estimates of the data points. In addition, entire domains are void of quantitative information, e.g. precipitation distribution over the globe and during past climatic events, soil composition, humidity and biomass accumulation, or solar activity, and many unknown etcetera.

Let’s remind ourselves that there is one only experiment running in one only laboratory: the experiment “Earth” in the solar system and its history. By the way: any model calculation shall never be called “experiment” (some “climate scientists” do so, they are wrong!).

When confronted to the question “what causes what?”, we can only look back at available time series to investigate if significant signals (a necessary pleonasm) can be distinguished from all other phenomena, known and unknown ones.  To my knowledge this has proven an impossible task.

Simplistic people will be content with a mere coincidence: is temperature increasing, as does the CO₂ concentration? Then the first is deemed a consequence of the second. They are wrong! Or they would believe that warming our houses in the Fall is the cause of falling leaves from the trees.

More inquisitive people look at regression analysis to estimate a degree of probability with which a variable may correlate with other ones, the famous R² goodness of fit. Absence of noticeable correlation is a strong indicator of the absence of a cause-effect relationship; but it is no proof because, in all logic, absence can never be proven. Neither does it falsify a hypothesis saying that such relationship should exist, because the signal being sought may remain hidden behind too strong perturbations that make it impossible to isolate from a noisy background. It’s impossible to listen to a pianissimo of an Erich Satie composition when Hard Metal Rock and Carnival samba are being played simultaneously at 110 decibels. On the other hand, neither will a good correlative fit demonstrate a cause–effect relationship; it may just be a circumstantial evidence, as in the example of falling leaves in autumn. Experimental confirmation is needed, under elimination of unwanted interferences or perturbations. And we can’t experiment with the Earth; just waiting requires a patience that is not the highest post-modern virtue.

Not right, not wrong, on the contrary!
This is the current situation with climate science: no evidence can be proven from physical measurements; no experimental confirmation can be performed. The anthropogenic hypothesis cannot be proven, nor can it be falsified absolutely. We simply don’t know.

Nevertheless, it is alleged that an overwhelming majority of scientists interested about the climate believes that climate change is mostly man-made. How do they jump from a hypothesis, however plausible it may sound, to a mere belief? I cannot answer for these believers.

But, being curious, I tried to find regression formula that reconstruct the observed temperature evolution. The multi-variable regression software Eureqa from Nutonian (www.nutonian.com) was used to this effect.

Taking HADCRUT4 global monthly temperature anomalies and CO₂ concentrations, one obtains an already fair correlation (R²=0.9564), albeit simplistic:

This result could be construed as an evidence that CO₂ alone determines the global temperature evolution. But to contemplate one only potential contributor is a too reductive approach, even if it is what has been the opinionated aim of all climate research over the past 3-4 decades.

When introducing other variables: Atlantic Multi-decadal Oscillations (AMO), Magnetic Field Declination (MFD), Solar Spot Numbers (SSN), and Global Sea Level GSL), the best regression formula (R²=0.9923) retains them all, except GSL:

From this formula, the sensitivity of Ta to CO₂, d(Ta)/d(CO₂), is diminishing as CO₂ increases, a not unexpected finding:

At 280 ppm (historic value):  +0.030 °C/ppm
At 400 ppm (current value):  +0.015 °C/ppm
Extrapolating to the doubling of the historic value, 560 ppm: 0.0078 °C/ppm

Anyone can feel free to interpret this diminishing return, in particular when making imprudent extrapolations about the effectiveness of future emission reduction scenarios.

Pacific oscillations or El Niño La Niña indexes could also have been used; but AMO has the unique advantage that it is available over a longer time period; and the results obtained are quite of the same kind, with lesser goodness of fit.

But to me, no proof is yet established with such regression. If for example the CO₂ concentration is taken out of the dataset, aiming at target formula like Ta=f(AMO, MFD, GSL), then an even better goodness of fit (R²= 0.9936) is obtained:

This means that the evolution of the climate could also be described without at all considering CO₂ as a contributor. In such case, the sensitivity of the climate to CO₂ would be zero, at any concentration. A quite disturbing finding, isn’t it?

Thus, this gas cannot be designated as the major culprit of warming as long as all other intervening parameters have not been taken carefully into account, which is an impossible task given the instrumental records at hand. Anthropogenic Global Warming (AGW), as claimed by the mainstream, is therefore greatly falsified, and remains a mere unquantified hypothesis, or an unfounded belief.

To obviate the futile and impossible exercise of attributing climate change to one preponderant cause, models were developed that cannot prove anything better since they make recurrent claims on themselves to establish a virtual theory. Such tautology is always formally right: it’s good because it’s good. But it’s wrong, because it remains without qualitative and quantitative validation.

In lieu of relying on invalidated models to save their research funding and to feed an apocalyptic view of the next century, modest climate scientists must then say: “I don’t know!”
Does such genus exist?

Data source

This chapter has been revised from a blog post written at: http://blog.mr-int.ch/?p=3943&lang=en