The title of this post is taken from my AGU Fall Meeting poster presentation on the afternoon of Tuesday Dec. 4 (tomorrow).
You can view the poster from the comfort (?) of your terminal at where you can click on the View-ePoster tab.
For full transparency of the number crunching the poster links to the
Excel spreadsheet from which the graphs were extracted, allowing the reader equipped with Excel (2000 or later) to audit all graphs directly and experiment with alternatives, some of which I’ve provided for with sliders. (Microsoft’s free viewer will allow all this except for the ability to experiment, but I would expect Excel 2000 to be pretty cheap these days. I had high hopes for OpenOffice’s CALC but it turned out to be a nonstarter for non-toy spreadsheets.) Please let me know of any glitches you encounter. Only macros signed by Microsoft are used so Excel 2007 onwards should be happy, though I don’t believe 2000 knows how to read macro signatures and hence lives in a perpetual state of paranoia concerning viruses.
Global warming of some kind is clearly visible in HadCRUT3 (Figure 1,
resistor-color-coded red in the poster) for the three decades 1970-2000. However the three decades 1910-1940 show a similar rate of global warming. This can’t all be due to CO2 since the emissions data from the Carbon Dioxide Information Analysis Center (CDIAC) show that human-emitted CO2 wasn’t rising anywhere near as fast then as during 1970-2000. Both the population and its per-capita technology increased enormously in the intervening 60 years, with WW2 technology capable of obliterating cities with both conventional and nuclear weapons that WW1 generals could only dream of, and with WW3 postponed as sheer MAD-ness.
Figure 1
It would seem therefore that rising climate since 1850 is a lot more complex than can be explained by our rising CO2 emissions.
My hypothesis is that there is less to HadCRUT3 than meets the eye. I
hypothesize the following.
1. I collect all the so-called multidecadal ocean oscillations into one
phenomenon I call a quasisawtooth, namely a sawtooth lacking its first
harmonic or fundamental. Sawtooth waves occur naturally as the result of a sudden perturbation away from equilibrium followed by a slow return to equilibrium. The several 100,000 year deglaciation cycles of the late Quaternary are one example; this might be another, albeit with its harmonics filtered differently due to whatever lies between the perturbations and our perception of them (speculated on at the right of the poster, where I forgot to point out that the effect is seen in the oceans because the crust beneath is only 20% the thickness under the continents and almost nothing along the mid-Atlantic ridge). I remove this from the data (Figure 1) by subtracting it, giving the orange curve in Figure 2 labeled DATA – SAW.
Figure 2
2. I further hypothesize that all remaining natural fluctuations in modern climate have as their slowest component the 21-year Hale or magnetic cycle. I filter out this and all higher frequencies with a low-pass filter designed to aggressively block such frequencies. Its frequency response is given as the curve F3 in Figure 5.
Figure 5
F3 is constructed as the convolution of three box or moving average filters of respective widths 21, 17, and 13 years. The first gives F1 which by itself takes out both the Hale cycle and the 11-year solar (TSI) cycle. The second then gives F2 by bearing up at the one-third point of the first side lobe, while the third pushes down on the two-thirds point to give F3. (This filter can be constructed at
woodfortrees.org as I noted in my previous post here early in 2012.)
The effect is to block essentially all frequencies with shorter periods than 22 years. At most 0.4% of any such frequency gets through. What remains is the green curve in Figure 2 labeled F3(DATA – SAW). This is clearly global warming, whatever the cause; the poster calls it Observed Global Warming, color-coded green.
These two operations on HadCRUT3 take place on the left or experimental side of the poster, which deals with observation (in this case HadCRUT3) and its analysis (in this case by subtracting low frequencies and filtering high frequencies).
The right or theory side of the poster deals with the rationalization of observation, the half of science that goes beyond the mere reporting of
experience by hypothesizing explanations for it. Figure 3, color-coded blue, graphs the formula immediately below it, both before (in orange) and after (in blue) applying F3. (It can be seen that the filter makes essentially no difference except for a one-decade end-effect artifact where the curve is changing rapidly. In that context the (literally) side point is made that all odd decades since 1870 have trended positively while all even ones have trended more weakly and often
negatively, verifiable at Wood For Trees.)
Figure 3
The formula is based on known ideas due to Arrhenius in 1896 and Hofmann in 2009 (that the portion of atmospheric CO2 above the preindustrial level is growing exponentially), with the added twist that the oceanic heat sink delays the impact of radiative forcing variations on HadCRUT3 by 15 years, analogously to the overheating of a CPU being delayed by the addition of a heatsink with no fan, what I refer to as the Hansen delay. I call this the Arrhenius-Hofmann-Hansen or AHH law.
The rest of the poster, color-coded violet (purple, whatever), deals with the similarity to the curve in Figure 2: they are equal to within millikelvins. For the period to 1950 the standard deviation of their difference is half a millikelvin. After 1950 there are two bumps that need explaining; the best I could think of was brown-cloud pollution from uncontrolled western civilization emissions abating in the 1970s and then resuming with the industrialization of nonwestern civilization, but there may well be other explanations. Whatever the explanation however the main point is that multidecadal residual (MRES) is inconsequential in the context of global warming.
The hypothesis then is that multidecadal climate has only two significant components: the sawtooth, whatever its origins, and warming that can be accounted for 99.98% by the AHH law as measured by the R2 of its fit to observed global warming (and could be brought even closer to 1 with a good story for MRES).
Because filter F3 rises slowly on the left of its cutoff there is the worry that some multidecadal phenomenon was overlooked by sneaking into the Hale octave. Something like this seems to have happened on the high-frequency side of the SOL band, perhaps some ENSO noise from DEC (decadal band) getting into the TSI band. However the HALE band looks much cleaner than the TSI band, suggesting that nothing entered from the MUL (multidecadal) side. And since MRES as shown in Figure 10 is essentially flat by comparison with SAW and AGW, it would appear that those multidecadal variations not accounted for here are either too small to matter or have been inadvertently lumped in with (i.e. are inseparable from) one of SAW, AGW, or HALE.
Figure 10
With enough parameters one can make any two curves equal to within any desired precision. The judgement to be made here is whether the nine parameters used here have any chance of achieving a comparably accurate fit on random data in place of HadCRUT3. This was an objection raised to Mann’s methodology, and I would be interested to see if it applies here.
After the comments responding to this post, this might not be the poster I want any more, but meanwhile you go to the conference with the poster you have, not the poster you want. :)
In any event this poster is offered as a preliminary report on work still in progress. One reason for advertising it here at an early stage is that the denizens of Climate Etc. seem well motivated to poke holes in theories of global warming, which I view as a positive benefit of skepticism, as distinguished from flat denial. There seems to be a lot of the latter here too but I haven’t found it as useful.
JC comments: This is an invited guest post. I think that this is an intriguing analysis, but I have not looked at this in any detail. This is a technical post, please keep your comments on topic.
I am at the AGU this week, I will have a post on the meeting later this week.