Auditing the end of the Wisconsin glaciation.

[Andre]

If I google 'Wisconsin glaciation' I see periods mentioned of

-from about 80,000 years ago until about 12,000 years ago-
- lasting from about 100,000 to 10,000 years ago-
-The time interval between 35000 and 11150 calendar years ago is referred to informally as "late Wisconsin time".-
- The glaciers receded into Canada by 9000 years ago-

Clearly it seems desireable to find the real termination time of the glacial period and see if it matches other ideas on the termination of the ice ages. So let's duck into the formal scientific publications, hunting for dates associated with glacier activity receding.

However, first we need some background on dating methods, nowadays there is a plethora of them, roughly in three groups, radio-activity series, chemical alterations due to exposure to atmospheric conditions, and the third: counting any kind of annual layers.

A special method is radiocarbon dating, which is the most widely used due to the abundance of dateable records. There are several old threads on carbon dating, so I only briefly resume the essence of the method. Radioactive 14C with the half time of 5568±30 (Libby) or 5730±40 years (Cambridge), is generated in the atmosphere and taken up in the biologic carbon cycle. Life tissues constantly refresh carbon, keeping their relative rate of radioactive 14C (delta14C) about constant, but after death the interaction stops and the delta14C decreases logaritmically due to the radioactivity. So the delta14C is a measure of time after death.

It soon showed however that carbon dating did not compare wuth other dating methods because was much more complicated due to many different processes. The most important being variation in the atmospheric delta14C due to variation in pCO2, intensity of the cosmic activity and changing interactions with the oceans. However large (around 1000-3000 years) those differences are constant and robust calibration tables could be made by comparing counted records (annual coral rings, tree rings, lake sediment layers) with the carbon dates. The most current is Intcal04 Reimer et al, Radiocarbon 2004. These tables also circumnavigate the halftime deviation problem as long as the same value is used. Therefore, although wrong, Libby is still in use to avoid recalibrating all the old publications.

This brings us at the main question: how much older publications, relying on uncalibrated carbon dates, are still used to define borders between periods. Have all the dates been calibrated and updated?

We'll see.

 

[Andre]

The first publication that deserves some interest is:

Forman, Steven L. , E. Arthur Bettis III, Timothy J. Kemmis, and Barry B. Miller, 1992; Chronologic evidence for multiple periods of loess deposition during the Late Pleistocene in the Missouri and Mississippi River Valley, United States: Implications for the activity of the Laurentide Ice Sheet , Palaeogeography, Palaeoclimatology, Palaeoecology, 93 (1992): 71-83 71 Elsevier Science Publishers B.V., Amsterdam

The abstract reads:

The loess stratigraphy of the mid-continental U.S. is an important proxy record for the activity of the Laurentide Ice Sheet in North America. One of the most outstanding problems is deciphering the age of loess deposits in this area during the late
Pleistocene. Radiocarbon dating of snails and thermoluminescence dating of the fine-silt fraction (4-11 ~tm) from loess at the Loveland Loess type section, Loveland, Iowa and a recent excavation at the Pleasant Grove School section, Madison County, Illinois provide new chronologic control on loess deposition in the Mississippi/Missouri River Valley, U.S. The thermoluminescence chronology indicates that the Loveland Loess is Illinoian in age (135 + 20 ka) but is not correlative with the Teneriffe Silt which is dated to 77 + 8 ka. Concordant radiocarbon and thermoluminescence age estimates demonstrate that the Roxana Silt and a correlative loess in Iowa, the Pisgah Formation, is probably 40-30 ka old. These age estimates in conjunction with previous results indicate that there were four periods of loess deposition during the last 150 ka at 25-12 ka, 45-30 ka, 85-70 ka and at ca. 135+20 ka.

This chronology of loess deposition supports the presence of both a late Illinoian and early Wisconsinan loess and associated soils. Thus, there may be more than one soil in the loess stratigraphy of the mid-continental U.S. with morphologies similar to the Sangamon Soil. The last three periods of loess deposition may be correlative with periods of elevated dust concentrations recorded in the Dye 3 ice core from southern Greenland. This is particularly significant because both areas possibly had the same source for eolian particles. Reconstructions of atmospheric circulation for glacial periods show a southerly deflected jet stream that could have transported dust from the mid-continental USA to southern Greenland. Lastly, the inferred record of loess deposition is parallel to a chronology for deglaciation of the Laurentide Ice Sheet deciphered from chronologic and
stratigraphic studies of raised glacial and marine sediments in the Hudson Bay Lowlands, Canada. These chronologies indicate that the Laurentide Ice Sheet was quite dynamic during the late Pleistocene, advancing and retreating across North America at least four times during the last 150 ka.

(Loess is sort of ultra fine rock particles chafed off from bed rock by glacial activity, it's also known as glacial flour, making glacier lakes milky opaque.)

So there is "12 ka" for the end of the Loess deposits, hence end of the glacial period. So what about dates in the publication? We find in table 1 and 2 youngest carbon date of 20,540 BP and youngest thermo/(opto)luminescence (TL) date of 16000 BP. They also observe:

However, recent dating of corals by UTh and AMS ~4C indicates that the two time scales are not concordant for the interval 9-30 ka and thus the radiocarbon time-scale probably deviates significantly from calendar-year time (Bard et al., 1990). This comparison demonstrated that U-Th ages are significantly older than radiocarbon ages, with the maximum difference between these chronometers
of 3.5 ka at a radiocarbon age of ca. 20 ka. Unfortunately, there is insufficient data to
correct radiocarbon ages >20 ka for nonlinearities in the production of 14C, but the available analyses indicate that the magnitude of the underestimate in radiocarbon ages for this period is between 1 to 3 ka. Thus, radiocarbon ages reported in this paper may be underestimates by a few thousand years when compared to TL age estimates,

That's a most accurate observation as we know now, but it doesn't answer where that "12 Ka BP" comes from. Seeing the various graphs, nothing is indicating dating confines for the upper boundary of the loess. But then we read:

These results and previous determinations demonstrate that this loess was deposited during the Late Wisconsinan glacial/deglacial cycle between 25 to 12 ka (Ruhe, 1983).

So it appears that the "12ka" in the abstract is not the result of own research but merely passing on of older information. Mind that in 1983 the carbon dating problems were not really sorted out yet, so these dates may just be uncalibrated carbon dates as well (haven't located that paper yet) and the researchers knew that there was something fishy about carbon dates. So why not discuss this in relation to that earlier research? Note that 12000 radiocarbon years BP (Before present, present = 1950) is 13,830 calibrated calendar years BP on INTCAL04.

looking for Ruhe 1983 (Ruhe, R.V., 1983. Depositional environment of late Wisconsin
loess in the midcontinental United States. In: S.C. Porter (Editor), Late-Quaternary Environments of the United States. 1. The Pleistocene. Univ. Minnesota Press, Minneapolis, pp. 130- I37.)