14C as a proxy, but for what exactly?

Posted: July 18, 2011 by tchannon in climate, Solar physics

Figure 1

How is 14C data turned into a solar or radiative proxy?

Didn’t know that, is a trivial question which has been troubling me for some time. By chance the key to understanding turned up in a 200 page tome, “Stable isotope mass spectrometry and AMS dating applied to a multi-proxy climate record from the Bliden Lake, Denmark”, doctorial thesis of Jesper Olsen.

He drew a picture.


Figure 2 (Olsen Figure 4-5)


Page 64 (page 76 of PDF) of a fascinating thesis.

Perhaps he was passing on what had been shown to him, never mind, he did it. As a result the handwaving by others, not actually saying, turned into my understanding what they omitted to say.

Want to really understand something and use it fully, take it apart, put it together, explore it.

Reproducing the 14C flux curve

The data I used is very recent IntCal09


Figure 3

That tells the whole story, there is nothing magical, simply split the data into long and short term, and invert the Y axis so it looks like solar data. According to Fig4-5 in Olsen has shown this done for recent times by lsqr fitting a sine and then subtacting, but not inverted.

I point out this could be done using filtering, except that would need to address the end of dataset problem, avoided by most people in science.

For the moment I am looking at the most recent 9,000 years.

I used the synthesiser software here to extract the four largest long terms, thought, switched off two of them, the above is the result.


Figure 4

The makes reasonable sense yet raises many questions.

Added image 20th July, marine data version


Figure 4a

Last datapoint is at 1950 when the exploding of nuclear weapons killed 14C as a useful natural proxy, later data needs extreme compensation.


Figure 5

In my mind the straight line from 1900 is bizarre.

A part explanation might be human mining, such as coal extraction and combustion: radioactive gases are emitted, whereas naturally these leak out slowly. That still makes little sense, the line is too straight, and why 1900? I have no answer.

Turning to the longer term

The late Jack Eddy is cited, the long term change is part of solar activity change. My reaction is more than surprise. That variation is huge relative to the supposed solar proxy when you consider what we know about observed activity change.

According to the data there has been orders of magnitude change roughly linearly during the past 30,000 years.

Radiation, presumably cosmic rays, show as a huge flux.

Next I turned to splitting the dataset using a filter, quick and easy.


Figure 6

What is that? Well, I am throwing a spanner in the works of 14C as a solar proxy. Four different low pass filters applied to the input data, subtract and there is the high pass, what remains without the longer term wiggles.

If you find it doesn’t look credible, I have allowed the plot software to rescale the Y-axis. It really is the same data.

Why 200, 700, 2000, 7000 year? A guess at what would show the effect without too many plots. Note the filter is end compensated, have to trust me. (otherwise walk away and do your own)

That result points to 14C being hugely influenced by something else, perhaps vegetation activity, so could be be water availability, temperature, how much of the earth is environmentally acceptable to plants?

A quick explore suggests a signature of orbital insolation variation but that will have to wait for one day maybe.

  1. Dennis Cox says:

    I have been told by a well recognized authority that, when used for dating of any given stratigraphic layer, 14C data is at best, only acurite to +/-100 years.

    Does that lack of resolution in the data apply here as well?

    Also, the idea of 14C being “hugely influenced by something else” resonates well with something I’ve been digging in, and trying to make sense of, for a couple of years now.

    In their work on the Taurid Complex, astronomers Victor Clube, and Bill Napier, have pointed out the very real probability that, in the past 20,000 years or so, the Earth has been hit by periodic meteoroid swarms, of sub-kilometer comet fragments which were some of the debris left over from the breakup of the Taurid progenitor.

    The worst of them may have triggered the Younger Dryas Cooling, the demise of the Clovis culture, and the mega faunal extinctions 12,900 YA.

    The energies of the multiple airburst impact storms they propose would’ve played hell with the radiometric data.

    I realize you are probably talking about something completely different here. But can you explain what you mean by “Orbital insolation variation”?

  2. Tim Channon says:

    Dating using 14C (sic) which is a single spot value will be a very different thing from a carefully constructed reference dataset. It’s purpose is date correction.

    Your point is of course excellent.

    “Orbital insolation variation”: solar heat input change caused by solar system orbit variation primarily the earth is in a non-circular orbit, has an included spin axis and wobbles. The signature has been computed but is probably not quite right.

    Extinction events etc., is a wildcard.

  3. tallbloke says:

    Tim, fascinating post. Lots to think about here. Probably a lot of questions coming your way too – sorry. 🙂

    First one: Could the precession cycle of ~22Ky have something to do with the long term trend?

  4. Tim Channon says:

    The data suggests something like that. Be very wary.

    I suspect at least some if not all has been addressed and it is my ignorance showing.

  5. adolfogiurfa says:

    Funny indeed! 🙂 Nothing to do with “proximity”, perhaps “proxy” it´s a “proxy” for bias. BTW, perhaps all the atomic theory will be revisited in the following years.

  6. Tim Channon says:

    I’ve reproduced the result for the marine dataset.

    There is less variation, no major other difference except for the end hockey stick which is much more marked.

    Given I am _not_ using filtering there is no data end effect whatsoever, therefore is to do with the dataset.

  7. tallbloke says:

    Thanks Tim. I think we all need to do some reading around and report back here.

  8. Tim Channon says:

    What follows is very tentative, is unsafe.

    Whilst taking a new look at aa index data (someone asked), I noticed an echo of the strange end of 14C dataset shape. The datasets briefly overlap.

    On following this up a surprisingly good r2 appears (can go past 0.88) and perhaps even more interesting is aa leading 14C. ie. the correct phase.

    That said the processing to do this is extreme and very untrustworthy as meaningful. There are also known aa index data problems, a large subject in itself where I am not entirely in agreement with the papers published in the issue. (in part because of what they ignore, do not mention)

    There is supposed to be some kind of solar linkage with 14C and aa some kind of solar wind proxy.

    If we assume the above is correct the 14C deviation is abnormal in 14C history and suggests strange solar behaviour since 1900. I think that does not fit well with other data.

    Yet more questions…

  9. adolfogiurfa says:

    “AA index” is the geomagnetic index, and that “solar wind” remembers me of that old experiment in electricity where a sharp pinpointed electrode (out from a “Ramsden” conductor) “blows” a candle flame…
    Electricity and Magnetism are tied both at a 90°angle (the Oersted´s “right hand rule”).
    As birds on a high tension wire we do not feel the current and so we conclude that it doesn´t exist!

  10. bill says:

    I would require night vision spectacles to understand this.

    A few “daylight” observations:

    Roger pointed out july 13 on proxy fix-up post that there was a change in timing at 6000ybp which is about 1/2 of our present (late?) ~11500y cycle. we might be “changing sign” again.

    Your observation that the marine signature was different (esp. recently) suggests different polar deposition rates with observed averages being affected by polar ice melt (both sea and land) as well as increasing arctic circulation.

  11. Tim Channon says:

    bill, everyone,

    Marine data image has been added as Figure 4a to the article. Y axis scaling deliberately identical.

  12. bill says:


    looks like I drew two blanks. try for three.

    may correlate to ocean heat content (sort of like a long term, smoothed, temp chart)?

  13. tallbloke says:

    Regarding fig3
    If the curve were a correct reflection of solar activity,would Pytheas have made it to Iceland in 282BC in a boat if the pack ice reached nearly down to Scotland in the little ice age?

    I think there’s something else going on in the data.

    I doubt if big swings in solar activity cause the glacial/interglacial cycle. Either the orbital cycles have more profound effects on ocean circulation than we currently understand, or the C14 data is affected by an as yet unknown parameter. The rise from 3000BC is worrying, because it doesn’t fit with a gradual decline in temperature from the holocene climate optimum. Long term albedo changes?

    Having said that, it does seem to correlate reasonably well to known historical periods of cooling and warming over the last millennium or so

  14. bill says:

    Maybe similar mystery during other deglaciation period (s?)


  15. bill says:


    I think your work on this has uncovered the padlock that fits the Holocene’s coffin.

    Congratulations !!

    Make sure you get the credit due.