Glacial Cycles, part 1b

This is just a “quickie” to show the results of Fourier analysis of a stack of delta-oxygen-18 records from benthic (i.e., ocean floor) sediment cores, which is not orbitally tuned.

The data are from the supplemental online material to Huybers 2006, Early Pleistocene Glacial Cycles and the Integrated Summer Insolation Forcing, Science, 313, 508, doi:10.1126/science.1125249. They cover about the last two million years:

Here’s the Fourier periodogram

Note: this is a modified version of the “date-compensated discrete Fourier transform” or “DCDFT.” But in fact any respectable Fourier transform which compensates for uneven time sampling will give essentially the same results.

The 41,000-year obliquity period is clearly present. The frequencies associated with precession are possibly present, which is why I’ve indicated them with a question mark after “Precession”.

We can make the precession influence clearer by recognizing that its impact is stronger in the most recent million years than it was previously. In fact the Fourier spectrum of the first (older) million years of this data set doesn’t really show the precession band at all (while showing the obliquity cycle much more prominently):

The second (most recent) million years, however, does show the precession frequency band:

especially if we zoom in:

The response in this frequency range for the sediment core data is more “broadband” than that of the precession signal itself, as well as more so than we noted in the (orbitally tuned) LR04 stack. This is to be expected, if the age model fails to capture the timing of fluctuations perfectly. That will “smear out” the power in the precession band, while orbital tuning will re-focus it. This argues not only for the validity, but for the usefulness, of the orbital tuning.

NOTE: the age model for these data, older than about 1 million years is influenced in a very small way by orbital assumptions, but I think we’re on safe ground, since the precession cycle is really only present in the most recent million years, and the obliquity cycle is so strong and persistent as to be beyond all doubt. Huybers describes it thus:

A Pleistocene age-model is estimated by extending the depth-derived age-model of (S1) from 0.8 myr to 2.4 myr ago. Geomagnetic age-control comes from the Brunhes-Matuyama transition (0.78 myr), the Olduvai (1.95-1.77 myr) and Jaramillo (10.7-9.9 myr) subchrons, and the Matuyama-Gauss transition (2.58 myr). These geomagnetic ages are derived from a combination of radiometric dating techniques, sea-floor spreading rates, and for ages older than the Brunhes-Matuyama, astronomically derived age estimates (S2,S3). Thus, geomagnetic ages older than the Brunhes-Matuyama magnetic reversal (0.78 myr) do contain limited orbital assumptions, but constitute only five ages in the course of 2 myr of which orbital assumptions are but one constraint. Geomagnetic events are assumed to occur in the same isotopic stages identified by (S4) for core DSDP607.

There’s certainly no “wiggle matching” involved.

5 responses to “Glacial Cycles, part 1b

  1. Like!

    I’m intrigued by the slightly split peak on the 100,000 year signal. I wonder if it will play into your upcoming attribution post??

    • Well, its seems to the be the 100 ooo year orbital eccentricity cycle….

      But its influence is relatively weak in the first 1M years and instead dominant in the last 1M years … why it is so I have no idea…

  2. David B. Benson


  3. I don’t want to be a pain, but I will be and point out that Huyber’s said of his data:

    “Thus, geomagnetic ages older than the Brunhes-Matuyama magnetic
    reversal (0.78 myr) do contain limited orbital assumptions, but constitute only five ages in the course of 2 myr of which orbital assumptions are but one constraint.”

    I would not worry about this except that there is a clear change in periodicity at around -0.8 million in the data, and your analysis of the most recent million years (the period without orbital assumptions) shows much weaker Milankovitch periods than does the first million years (in which orbital assumptions were made).

    Do you know the nature of Huyber’s “orbital assumptions”, and how they may have influenced the data?

    Of course, the individual data series used by Huyber over the period are continuos, and have strong contraints on synchronicity at five points in time timed by geomagetic reversals. Therefore the effects of the “orbital assumptions” must be relatively small, so I am only asking you to dot “i”s and cross “t”s. I perfectly understand if you consider the geomagnetic constraints on the chronology sufficient for your purpose.

    I have to say I have thought very hard about this post. I want to increase my understanding, but I do not want to feed denier trolls. In the end, however, I set a very high premium on understanding things.

    [Response: I noted the orbital influence on the age model (quoting from Huybers) at the end of the post. You may find some of what you want to know about his methods there.

    But the changes about 800,000 years or so ago (the “mid-Pleistocene transition” simply cannot be due to orbital tuning (of any kind) on the age model. For one thing, the most obvious change is the dramatic increase in the physical amplitude (visually evident from the graph), which orbital tuning can’t affect. For another thing, all age models (including “wiggle-matched” models) are determined by numerous factors (e.g. sedimentation rates), and you can’t orbitally tune the age model enough to change a 41,000-year fluctuation into a 100,000-year fluctuation without introducing so much conflict with those other factors as to set off lots of very loud alarm bells. There’s really not enough “wiggle room” to do that. And, as Huybers makes clear, the impact on the *entire* record (not just prior to the Brunhes-Matuyama transition) is really very small.

    Also, although the obliquity cycle is weaker in the most recent million years (untainted by orbital timing) than prior to that, it’s still plenty strong enough to be simply undeniable. As for the precession cycle, it’s stronger during that period.

    Keep setting a very high premium on understanding things.]

  4. I’ll have fun here and report that “The first solid evidence of human use of fire in Eurasia as early as 790,000 years ago has been found “, near the start of the unexplained 100,000 year cycle, and push the beginning of the Anthropocene a bit more back… ;-)