The Western Geologist

I should probably start this post by pointing out that I'm not a rock magnetist (although I've been friends and office mates with them), nor am I a paleoclimatologist or a sedimentologist. I don't usually think about those three subjects being related. Paleoclimatology and sedimentology, sure. Same thing for rock magnetism/paleomagnetism and sedimentology, I can see how those could sometimes overlap. I think it's pretty cool when a study can combine all three.

Several parameters of Earth's orbit vary cyclically (precession, axial tilt, and eccentricity – see this article about Milankovitch cycles (go Wikipedia!). These variations result in changes in climate, which cause changes in sedimentation, which result in changes in the geologic record. These changes are particularly clear in materials that have layers deposited annually. The classic example is varves. Changes in climate during the summer and winter result in changes in the types of sediments being deposited in lakes. For example, runoff during the spring and summer could carry more silt into a lake, while during the winter, when the runoff is low, clay would settle out. In that example a pair of silt and clay layers represents one year's worth of deposition. There can be changes in things like pollen concentration and microfossils too. By counting the pairs (called couplets) and tracking changes like couplet thickness and chemistry, it's possible to recognize cycles, and those cycles can be related to climate.

Varves occur in both modern and ancient lakes. For example, at Lake Suigetsu, (where the varves are defined by a couplet with a clay layer and a white fossiliferous layer) not only have varves been counted back 40,000 years (like counting tree rings), but the ages calculated from counting varves were also compared against radiocarbon dates from organic material plucked from the varves. (see Figure 2 at the link). In the Lake Suigetsu study the authors counted 80,000 thin layers of sediments (40,000 couplets). That's an impressive amount of work, but while 40,000 years is a long time from a human perspective, it's not really enough to record Milankovitch cycles (the timings of the cycles range from around 19 to 405 thousand years). If you want to see Milankovitch cycles you need to look at sequences of rock that record very long periods of time. People have done this for varved deposits (see here for example). Very cool stuff, definitely, but it gets better. The article that prompted this entry is about people who found a way to measure Milankovitch cycles in non-varved deposits.

"Magnetic record of Milankovitch rhythms in lithologically noncyclic marine carbonates", by a group of scientists from Lehigh University, Johns Hopkins, and the University of New Mexico came out in the December 2005 issue of Geology (to go off on a tangent, Geology is a nice journal to subscribe to – it's composed of short (4 page or so) articles in all areas of geology). "Lithologically noncyclic" means that there weren't any varves in the rocks they looked at 110 m of a "… thick-bedded lime mudstones with rare chert nodules, nannofossils, and planktonic foraminifera." To me that sounds like they looked at a thick, boring mass of limestone. They took 367 samples, crushed them, and measured their magnetic properties, in particular something called ARM (anhysteretic remanent magnetization – see this site for more information). ARM is a good way to measure the concentration of minerals like magnetite (the greater the ARM signal, the greater the concentration of magnetite). They plotted the strength of the ARM over the 110 m of limestone they measured, and then looked for cycles in the variations of magnetic strength. They found several – the largest at almost 29 m (which they report as 0.035 cycles/meter), and 5 more at closer spacings.

Converting that 29 m (and the other more narrowly spaced cycles) to time is the tricky part. As I mentioned above, the timing of the longest Milankovitch cycle is around 405 thousand years. They assigned this age to the 29 m cycle, assumed a constant deposition rate, and then inferred the time that would correspond to the rest of the cycles, coming up with 405, 123, 51.2, 39.4, 22.5, and 18.6 thousand years (ka). Those ages are about what is expected for Milankovitch cycles in the Cretaceous, which is when the limestone they were looking at was deposited. This does increase my confidence that they really are measuring something Milankovitch-related, but I do wish they had an independent control on the ages. They do have an independent verification that there should be cycles in the limestone they looked at (the San Angel limestone). The San Angel was deposited at the same time as the Cupido formation, which was deposited closer to shore, and has lithologic variations that record Milankovitch cycles. So, it seems reasonable to me to expect that there should be Milankovitch cycles in the San Angel too.

I mentioned above that the cycles are defined by variations in the concentration of magnetite. The authors of the paper used scanning electron microscopy (SEM) to image the magnetite. The grains were small, around 3 microns (3 millionths of a meter), which led the authors to infer that they originated as dust. Dust can make it pretty far out into the ocean. When there are dust storms in the Sahara for example, they can be detected far out into the Atlantic.

The idea is that cycles in the parameters of Earth's orbit (Milankokvitch cycles) lead to cycles in climate, which cause cycles in erosion, which cause cycles in the amount of dust carried out into the ocean, which cause cycles in the amount of magnetite deposited in marine limestones (even though there isn't a change in the lithology of the limestone). I was very impressed that the authors could take a massive limestone formation, just the opposite of the sort of formation that has typically been used to recognize Milankovitch cycles, and recognize cycles that correspond to variations in climate in the Cretaceous.