Miller et al (GRL 2012) url has attracted much recent attention for its argument that volcanism can account for the MWP-LIA transition. In my opinion, it is important for another reason, a reason not mentioned and apparently not noticed by the authors themselves. It offers a highly plausible re-interpretation of Arctic varve series, an interpretation that, in effect, stands the temperature interpretation of the important Big Round Lake, Baffin Island varve series on its head. Arctic varve series, including Big Round Lake, have become a mainstay of temperature reconstructions used in AR5 (FOD) and likely to be used in AR5 (e.g. Kaufman et al 2009) and Miller’s interpretation of varve data impacts multiple “new” AR5 studies. CA readers are familiar with climate scientists having trouble with the orientation of varve data e.g. the use of Tiljander’s varve data in Mann et al 2008-2009 (the latter frequently cited in AR5).
Over the past few years, Miller and associates have radiocarbon dated mosses at 90 sites revealed by receding Baffin Island glaciers, observing a concentration of kill dates in the late 13th century and again in the early 15th century. This is interesting and useful new data that is helpful to disentangling the climate history of the area. They interpret the lack of kill dates in the MWP (from ~950-1250) as due to relative warmth, resulting in recession and/or absence of the small Baffin Island glaciers:
Here we present precisely dated records of ice-cap growth from Arctic Canada and Iceland showing that LIA summer cold and ice growth began abruptly between 1275 and 1300 AD, followed by a substantial intensification 1430–1455 AD.
They interpreted the lack of kill dates from 1450 to the late 20th century as due to continuous ice cover until recent glacier recession. See their Figure 2C, excerpted in the top panel below:
From Miller et al 2012 Figure 2C-D. (c) Ice cap expansion dates based on a composite of 94 Arctic Canada calibrated 14C PDFs. (d) 30-year running mean varve thickness in Hvítárvatn sediment core HVT03-2 [Larsen et al., 2011].
Their Figure 2D (the bottom panel of the above graphic) is a smoothed version of varve thicknesses from Hvítárvatn, a proglacial lake in Iceland. Juxtaposing the information from kill dates with varve information, Miller et al concluded that the narrow varves from 950-1250 corresponded to glacier absence (or recession), indicating relative warmth, while the wider varves in the Little Ice Age showed the existence of active glaciers, indicating relative cold. Miller et al:
Baffin Island kill dates define abrupt and sustained summer cooling in the late 13th Century, which is matched by the start of a centennial trend of increasing Hvítárvatn varve thickness (Figure 2d), consistent with our predictions. A second abrupt increase in varve thickness in the 15th Century, and continuously thick varves through the following century, is consistent with persistent ice-cap expansion in the Canadian record at the same time. Hvítárvatn varves attain their maximum LIA thickness in the late 19th and early 20th centuries, decreasing again in the late 20th Century as Langjökull receded.
The picture, thus far, is consistent with the “traditional” perspective on the MWP and LIA – unsurprising since Baffin Island and Iceland are within the heart of the region conceded even by opponents. Miller et al argue (passim) that modern warmth is greater than medieval warmth. However, while the authors undoubtedly believe this, this point is made more as a genuflection than as a central focus of the article.
Varves of Baffin Island
However, this picture differs sharply from recent temperature reconstructions using Baffin Island varve data. For example, Thomas and Briner 2009, writing about the Big Round Lake (Baffin Island) varve series, reported that the period 1400-1575, rather than being a period of intensified cold with a “final pulse of ice-cap growth (as Miller et al concluded), had stated that it was relatively warm:
The warmest pre-twentieth century period in this 1000 year record, 1375–1575 AD,
Their temperature reconstruction, rather than interpreting narrow 11th century varves as evidence of warmth (as Miller et al 2012 did), concluded that the 11th century was exceptionally cold (despite the historical traditions.) This temperature reconstruction has been widely used in recent multiproxy reconstructions.
In the figure below, I’ve plotted the Big Round varve series on the same scale as the Hvitarvatn varve series, showing some important similarities.
For example, both the Big Round varve series and Hvitarvatn varve series show relatively thick varves in the 1400s and 1500s, with a local minimum in the late 1600s, then strong increases through the 18th, 19th and 20th century, perhaps tailing off a little towards the end. Despite this seeming similarity, opposite conclusions about the temperature implications were drawn by authors of the two series.
Thomas and Briner 2009 had interpreted the varve thickness increase at Big Round Lake in the 15th century as evidence of warmth, whereas Miller et al 2012 had interpreted a seemingly similar increase at Hvitarvatn as evidence of increasing cold.
In the 11th century (critical for medieval-modern comparisons), Miller et al 2012 interpreted narrow Hvitarvatn varves as evidence of glacier recession and/or absence i.e. local warmth, whereas the Thomas and Briner reconstruction (applying a postulated linear relationship between varve thickness and temperature) deduced that the MWP was very cold in Baffin Island. (This conclusion was not articulated or discussed in Thomas and Briner but is inherent in their reconstruction.)
Varve series from Baffin Island (Big Round, Donard, Soper) have been used in recent temperature reconstructions (Mann et al 2008, Kaufman et al 2009, Kinnard et al 2011, Ljungqvist 2009, 2010, Ljungqvist et al 2011 (CPD), Christensen and Ljungqvist 2011 (J Clim; CPD).
Polissar et al 2006
Another example of using proglacial sediments to interpret glacier recession/advance is Polissar (Bradley) et al 2006, which studied sediments in a proglacial lake in the Venezuelan Andes (prior CA discussion here). Polissar et al Figure 2A,B shows increased magnetic susceptibility (more iron minerals) during periods of glacier activity (see below):
Figure 2. Excerpt from Polissar et al 2006 Figure 2: Glacial advances, indicated by increases of sediment MS in L. Mucubajı´ (A) (vertical gray shading), coincide with an increase in precipitation, shown by higher MS in L. Blanca (B).
Polissar concluded that local glaciers in the Venezuelan Andes had been absent in the MWP, but had reformed in the Little Ice Age (hence the different character of the sediments) prior to receding with increased warmth in the 20th century.
In this example, there was increased glacier activity in the Little Ice Age, resulting in more iron minerals in the sediments, a phenomenon seemingly analgous to Miller et al’s attribution of increased varve thickness in the LIA to increased glacier activity.
The recent articles on varves underpinning the AR5 citations (e.g. Kaufman et al 2009 and its underlying studies such as Thomas and Briner 2009) attempt to interpret varves solely in terms of temperature. Most of the articles show positive correlations between late 20th century and varve thickness. Narrow varves (of MWP type) are not observed in the calibration period. This requires extrapolation of a relationship established over thick varves to the perhaps different circumstances of thin varves – a rather adventurous extrapolation of the sort that is typically discouraged in statistical literature.
There is considerable older literature on varves, principally in connection with LGM deglaciation. There is a prominent varve outcrop (see Tufts varve webpage here) about half a mile from my house. (The varves were re-exposed last year during straightening of Pottery Road). In the older literature on the LGM e.g. Agterberg and Banerjee 1969, varve thickness was held to depend on proximity to the ice front:
Both the silt (summer) and the clay (winter) layers in a varve couplet show an approximately exponential decrease in thickness away from the icefront. This trend is more conspicuous in the sill
When the Laurentide ice sheet had sufficiently receded, so did the varve series. When Miller et al 2012 attribute the narrow 11th century varves to glacier recession/absence, it seems to me that they are observing more or less the same phenomenon, though on a much diminished scale.
Thus, as Miller et al 2012 imply, thin varves could result from either glacier recession/absence (cumulative warmth in the 11th century) or relative cold (in the late 17th century) – confounding efforts to reconstruct past temperatures using a linear relationship to varve thickness.
Efforts to reconstruct past temperature using simplistic linear relationships to varve thicknesses (as in the studies applied in Kaufman et al 2009) had already struck me as problematic though, prior to Miller et al 2012, the precise problem had not been diagnosed. In my opinion, Miller et al 2012 provides substantial support for rejecting the interpretation of narrow 11th century varves as evidence of medieval cold (the Big Round temperature reconstruction) and requires analysis of the effect of this (and similar data) on downstream multiproxy reconstructions.
Postscript: I’ve managed to write this post without referring to Kortajarvi sediments other than once in passing. Mia Tiljander had interpreted narrow varves as evidence of medieval warmth and wide varves as evidence of the LIA. The data was used in opposite orientations in the corrigendum to Kaufman et al 2009 and in Mann et al 2008-2009, with Raymond Bradley ironically being a coauthor of both studies. In Mann’s recent book, he argued, in effect, that it doesn’t matter whether the Tiljander data is used upside down or not in Mann et al 2008 and 2009. I disagree, but, regardless of its orientation “matters” in the Mann et al articles, it seems reasonable to expect scientists in the field to develop consistent scientific interpretations of narrow 11th century varves.