Balascio et al and the Baffin Island Inconsistency

There was some publicity this week on a paper by Young et al (Science Advances, 2015), which, according to Gifford Miller , whose work has been frequently discussed at CA (see  tag), had supposedly put the “coup de grace on the Medieval Warm Period”, that had been so long wished for by the Team.   I will discuss (and dissect) this article in a forthcoming post, but first wish to report on some related developments in an article (Balascio et al, Climate of the Past, 2015) which I had discussed last summer while it was in open review, which I will subsequently connect to the discussion of Young et al.

Balascio et al (coauthor Raymond Bradley) had reported the (amazingly) strong similarity between the varve thickness series at Hvitarvatn, Iceland (Miller et al 2012) – see panel D below –  and at Big Round Lake, Baffin Island (Thomas and Briner, 2009) -see Panel B below. This similarity was originally reported at Climate Audit in 2012  and subsequently discussed in many Climate Audit articles, including the Climate Audit criticism of PAGES2K’s upside-down use of the Hvitarvatn series, a criticism that was subsequently plagiarized by Darrel Kaufman and Nick McKay (Nature Climate Change, 2014).

The Balascio et al 2015 figure showing the similarity (shown as an excerpt below) is unchanged from the article submitted for open review.


Figure 1. Excerpt from Balascio et al, 2015 Figure 5.  Original caption: …  (b) Big Round Lake, Baffin Island, varve thickness and magnetic susceptibility (Thomas and Briner, 2009; Thomas et al., 2010). (c) Baffin Island ice cap activity reconstructed using vegetation kill dates with text showing original interpretations (Miller et al., 2012). (d) Langjökull ice cap, Iceland-based on varve thickness from Lake Hvítárvatn (Larsen et al., 2011). Blue shading marks periods of increased glacier size (sustained above average PC1 values).

For reference, here is the figure that I had used in my comparison of these two series in February 2012.

Figure 2. Varve thickness (mm). Purple – Hvítárvatn from Miller et al 2012 Figure 2D; blue – Big Round Lake (NCDC sheet 7 column 4) 30-year running mean. Both shown on same mm scale.

However, in the open review article, Balascio et al did not confront the opposite interpretations of the varve data by Miller et al and by Thomas and Briner. Miller et al had (plausibly) interpreted thick varves as evidence of glacier proximity – see CA here for support of this interpretation from IRD (ice-rafted debris) data, while Thomas and Briner 2009 (followed by numerous multiproxy authors including PAGES2K) interpreted thick varves as evidence of warming.  This created an inconsistency for PAGES2K, which originally used Hvitarvatn varve data upside-down to Miller’s interpretation – an assumption sharply criticized at Climate Audit here.

A year later, PAGES2K lead author Darrell Kaufman, in in McKay and Kaufman 2014, acknowledged that the Hvitarvan data had been used upside-down in PAGES2K, but did not credit (i.e. plagiarized) Climate Audit -see CA discussion here.  Kaufman also refused to issue a corrigendum to the erroneous PAGES2K reconstruction, which remains widely used.  As I observed at the time, PAGES2K’s correction of Hvitarvatn resulted in the ironic situation that they now used the very similar Hvitarvatn and Big Round Lake series in opposite orientations.

In the original (open review) version of Balascio et al 2015 (discussed at CA here), Balascio and coauthors presented the similarity of the Hvitarvatn and Big Round series, but did not discuss the  inconsistency latent in the academic literature, as follows:

Varve thickness and magnetic susceptibility data from a proglacial lake [Big Round Lake] on Baffin Island, if interpreted to reflect glacier size, also reveal two similar distinct glacier advances at these times (Fig. 4) as well as the earlier advance ca. AD 1250–1300 also observed in the Kulusuk record (Thomas and Briner, 2009; Thomas et al., 2010).

In the final version, Balascio et al were aware of the inconsistency, but refused to grasp the nettle to comment on the inconsistency.  First, they modified the above sentence to only refer to magnetic susceptibility:

Magnetic susceptibility trends, linked to glacier size changes, from another high-resolution proglacial lake record on Baffin Island (Big Round Lake) reveal two similar distinct glacier advances at these times (Fig. 5) as well as an earlier advance ca. AD 1250-1300, which is also observed in the Kulusuk record (Thomas et al., 2010).

They then observe (as I had commented at CA on numerous occasions) that varve thickness at Big Round Lake, Baffin Island has been “previously interpreted” as a proxy for warmth, rather than glacier proximity, but refused to grasp the nettle. Rather than requiring consistency of interpretation, they airily speculated that the opposite interpretations could be saved, if the opposite interpretations “can possibly be attributed” to “different sedimentary processes operating over different timescales”.  They evaded reconciliation by saying that they “cannot account for this apparent contradiction without further analysis of those records”:

However, varve thickness data from Big Round Lake, which has previously been interpreted to represent summer temperature, resemble trends in magnetic susceptibility (Thomas and Briner, 2009). This discrepancy can possibly be attributed to how the two proxies track different sedimentary processes operating over different timescales (annual vs. centennial), but without further analysis of those records we cannot account for this apparent contradiction.

The referees did not require them to carry out the “further analysis”.  But why not?  This is a fundamental discrepancy between important proxies that Balascio et al had themselves cited in their article. And if not now, when?

Thus, despite the obvious and striking coherence of the varve thickness data between Iceland and Baffin Island – one of the most striking similarities in the entire field, Balascio et al only cited magnetic susceptibility as support for “regionally coherent cooling phases” in the North Atlantic from Baffin Island to Greenland and to Iceland (when the coherence between varve thickness series was, in my opinion, even more impressive):

We argue that the magnetic susceptibility data from Big Round Lake are consistent with other data from around Greenland, indicating that the most extensive glacier advances since the early Holocene occurred between AD 1250 and 1900, and provide evidence for regionally coherent cooling phases during the Little Ice Age (Grove, 2001).


It’s too bad that Balascio et al didn’t confront the obvious answer:  the astonishing similarity between the varve thickness series at Hvitarvatn, Iceland and Big Round Lake, Baffin Island – both in the North Atlantic and both proximate to glaciers with appreciable Little Ice Age variation – points to “regionally coherent” changes and the need to interpret both series consistently.  The inconsistency between interpretations of the Hvitarvatn and Big Round Lake series remains a rebuke to the “community”.

Even aside from its similarity to Hvitarvatn, the “community” interpretation of Big Round Lake makes no sense.  Under the Thomas and Briner interpretation, the 19th century appears as the “warmest” time in the past millennium, when we know that IRD (ice-rafted debris) re-occurred for virtually the first time since the LGM.  Under the Thomas and Briner interpretation, Baffin Island supposed experienced a Medieval Cold Period and Little Warm Age – a history that is directly contracted by other indicators, not least of which are Miller’s radiocarbon dates for moss killed by ice cap expansion (panel C in first figure above.)

The community reticence in calling out the incorrect interpretation of Big Round Lake series may very well come from its widespread use in IPCC multiproxy studies, including the Ljungqvist variations:  Kaufman et al 2009, Ljungqvist 2010, Christiansen and Ljungqvist 2011, Christiansen and Ljungqvist 2012, Ljungqvist etl 2012, Shi et al 2013, Tingley and Huybers 2014,  PAGES2K – see CA here.  Flipping the Hvitarvatn series had a dramatic impact on the PAGES2K Arctic reconstruction; while flipping the Big Round Lake series would not have as much impact, it would impact medieval-modern comparisons.

Also, if the community were to admit an error on Big Round Lake, then this would require re-examination of the numerous other varve thickness series that are a staple of the IPCC AR5 millennium spaghetti graph.



  1. Posted Dec 10, 2015 at 1:41 AM | Permalink

    Around the other side of Greenland is Kulusuk Island about the same latitude as northern Iceland. The analysis in a recent paper does not provide a lot to work with, but the result does suggest the existent of both the Medieval Warm Period (MWP) and the Little Ice Age and a lot more besides.

    I don’t have a lot of trouble accepting that the Roman Warm Period may have been substantially longer and warmer than the MWP.

  2. Patagon
    Posted Dec 10, 2015 at 3:35 AM | Permalink

    Kulusuk glaciers are tiny for Greenland standards. The average temperature is above freezing point in summer, even for the highest section, which is about 600 m a.s.l. (e.g. In these conditions glacier mass balance is probably determined by winter accumulation, including wind drift, which is rather important here. Balascio et al. study, if correct, indicates more a decrease in precipitation than an increase in temperature with respect to the MWP. Figure 5 clearly shows a smaller glacier about year 1200, although with a faster recuperation (again, rapid oscillations are indicative of changes in accumulation – if varve resolution were the same).

    • Posted Dec 10, 2015 at 4:16 AM | Permalink

      Thanks for this information.

      I noticed an odd thing, a correlation between the pattern in change in total mass during the period 4000 to 2000 BP and the period 2000 BP to the present.

      I estimate R as 0.78 for 21 pairs of accumulated observations, which is significant at the one per cent level of confidence.

      Do you know whether anything like this is reported for other Arctic glaciers?

  3. ianl8888
    Posted Dec 10, 2015 at 3:51 AM | Permalink

    > that they now used the very similar Hvitarvatn and Big Round Lake series in opposite orientations

    Oh dear … words are inadequate

  4. Posted Dec 10, 2015 at 10:05 AM | Permalink

    From the Balascio, et al., abstract:

    …in many Northern Hemisphere regions glacier advances of the past few hundred years were the most extensive and destroyed the geomorphic evidence of ice growth and retreat during the past several thousand years.

    It would appear that they confirm the existence of the Little Ice Age.

  5. Craig Loehle
    Posted Dec 10, 2015 at 10:57 AM | Permalink

    It is ok for inconsistencies to exist, but if so don’t use that data in a reconstruction. Any reconstruction that does use it should show with and without as a sensitivity test.
    Since glaciers can be affected by both temperature and precipitation (like some other proxy we might have heard of), there is an essential indeterminacy in the signal. In particular, some of the lake sediment indices seem to me to have essential ambiguities that have yet to be resolved. I personally prefer not to build my house on quicksand.

  6. kaganovitch
    Posted Dec 10, 2015 at 11:03 AM | Permalink

    “a history that is directly contracted” should be “a history that is directly contradicted”

  7. Follow the Money
    Posted Dec 10, 2015 at 3:38 PM | Permalink

    Figure 5a, not shown above, is the Kulusuk Glacier Size record. There is a noticeable gap between the endpoint of the data depicted and the end of the time line, unlike for Figures 5b, c, and d.

    At sec. 3.2 it is stated, “Ages are presented in calendar years before AD 1950 (BP) unless otherwise presented.”

    The 5a data does look like it ends around 1950.

    Why 1950? And why should this glacier study have any relevance for findings about climate after 1950?

    • Posted Dec 10, 2015 at 5:52 PM | Permalink

      Using 1950 is just a research convention used to establish the starting date (since years ago “before present” would literally change its start-date daily).

      • Follow the Money
        Posted Dec 10, 2015 at 9:53 PM | Permalink

        Why is “1950” the research convention? Why not 2000 or 2010? Why omit the data at all?

        The paper’s abstract makes a couple of claims for “20th century warming.” Did the paper’s authors write the abstract? Should not the authors issue a demand for the abstract to be corrected or withdrawn?

        Check out Fig. 5a. it looks like the “warming” flattened out in the early twentieth, and may have been decreasing even, until… … don’t know, the data ends.

        • Follow the Money
          Posted Dec 12, 2015 at 4:49 PM | Permalink

          Following up, the main body of the article very briefly mentions 20th century warming at p. 1594: The overall trend reveals a small and very gradual glacier expansion after 0.7 ka followed by 20th century retreat, which resembles the overall trend in Arctic temperatures over the past 2ka (Kaufman et al., 2009). Compare two paragraphs later: Both the Kulusuk and Langjokull glaciers appear to have advanced in at last two phases, at ca. AD 1450-1630 and ca. AD 1700-1930. So the glacier was advancing up to 1930…that’s in the twentieth century. More, the data at fig. 5a shows a big drop around 1930, indicating the end of the glacier broke off or some other explanation rather than the drop was caused by increasing temps a few hundreds of a degree per annum. After this quick drop the post-~1930 ’20th century’ trend shows no possible correlative evidence of steady warming. The data ends about, say, 1975–I previously said the depicted data line in the graph stops about 1950. There still is substantial missing data (the paper’s coring was done in AD 2010) and I don’t see how the data, or the paper itself, can correctly claim a “20th century retreat” in this glacier consistent with a posited, actual or fake, upward trend in 20th c. temps.

        • tty
          Posted Dec 15, 2015 at 10:23 AM | Permalink

          “Why is “1950” the research convention? Why not 2000 or 2010?”

          Historical reasons. Radiocarbon dating was invented c. 1950 AD, so to avoid inconsistencies between older and newer determinations the “present” has stayed put at 1950 AD. And it is likely to remain so. Going back and changing hundreds of thousands of dates would be a daunting task.

        • tty
          Posted Dec 15, 2015 at 4:08 PM | Permalink

          Thanks. For the carbon isotopes I know there is a ~1950 concern about above ground nuclear weapons testing contamination. I suppose annular periodicity effects can be eyeballed in the subject relatively easily for recent decades. Still, the paper makes claims about “20th century warming” and depicts data after 1950. I don’t think the post-1950 data is an instrumental record attached..

  8. Posted Dec 14, 2015 at 12:15 PM | Permalink

    Reblogged this on I Didn't Ask To Be a Blog.

  9. Caligula Jones
    Posted Dec 14, 2015 at 4:20 PM | Permalink

    Well, this has reached the Toronto Star:

    feel free to go an “debate” the warmunists there. I’m tired of it.

    • Posted Dec 20, 2015 at 12:11 PM | Permalink

      I think the total picture of Vikings in Greenland includes some theories that if correct would add up to an unviable society, but researchers and news media like hype not perspective.

      Climate deterioration is a well-supported factor.

      But changes in trading opportunities were likely a challenge (alternative source of ivory, fewer fish in the north Atlantic nearby reducing visits by fishing boats for supplies and repairs).

      Changes in societies in Greenland and the motherland due to religion and societal structure like clans may have reversed attractiveness – remember the founders of the Greenland society were outcasts and settlers were attracted by opportunity. It was probably a tough life at the best of times. There’s also a question of whether or not they had serviceable ships hundreds of years after they came to Greenland, they may have simply died out or assimilated with Inuit.

      Keep in mind this is archaeology which, while there have been extensive excavations, is not an exact science. (The book Across Atlantic Ice explains reasons why.)

      Given that eco-activists like to claim that degradation in one factor for non-human species will end the species, it would be contradictory of them to claim that one factor did or did not lead to disappearance of the Viking society in Greenland.

      • Caligula Jones
        Posted Dec 21, 2015 at 12:34 PM | Permalink

        Yes, I’ll continue to re-read Kirsten Seaver’s “The Frozen Echo” when I can. I find that actual research and science makes more sense to me than this political stuff posing as science does.

  10. EdeF
    Posted Dec 16, 2015 at 12:39 PM | Permalink

    Here is a link to varve thickness data at Iceburg Lake, Alaska over the last 1500 yrs
    (one of the above mentioned reference glacier-fed lakes).
    It has a prounounced MWP and LIA before very steep warming in the mid- to late 1900s.
    The glacier-dammed lake suddenly drained in 1999 and scientists were able to get detailed varve thickness measurments and other day over the next 6 yrs. Their interpretation is that varve thickness correlates to summer temperature, matched by
    nearby tree-ring data. Man, I’m confused.

    Steve: there have been numerous critical discussions at CA of Loso’s Iceberg Lake and other varve series. Most varvologists including PAGES2K say that thick varves mean warm, but Miller took the opposite view in Iceland. So PAGES2K is both upside-up and upside-down. It’s incoherent.

    • Howard
      Posted Dec 16, 2015 at 12:44 PM | Permalink

      Both the varves and tree rings look like hockey sticks with a late 20th century divergence problem

    • joe
      Posted Dec 31, 2015 at 10:33 AM | Permalink

      EdeF – thanks for the link to the article

      As a layman on the topic of AWG – I find the unoffical proxies to be good indicators of whether the official proxies reasonably reflect the relative changes in temps.
      The article cited by EdeF mentions that the lake was still dammed by the glacier during the mwp indicating that is was cooler during that period since the glacier had not melted whereas it is melted today.
      The receding Mendenhal glacier has exposed tree trunks that were carbon dated circa 1000ad which would indicate that area was warmer during that period vs today. As a rule, I would think proxies which are reasonably close should reflect similar values (granted the two locations are 300 miles or so apart)

      Any thoughts on what would appear to be somewhat inconsistent proxies – a frozen lake during the mwp vs trees growing where a glacier was until a few years ago. (unofficial proxies as they may be)

  11. Posted Jan 9, 2016 at 8:00 AM | Permalink

    Important analysis, thanks Steve. I am a geologist and am well aware that not all proxies are straight forward. This kind of discussion is absolutely needed.

    Obviously, it is important to work with as many proxy types as possible to compare the results with each other. I am currently mapping the MWP globally and have found 10 datasets for Baffin Island, most of them showing the MWP being present. Click on the interactive dots on the map to see the summary panel and key figure. Red=warm MWP present, blue=cold MWP

    More on the MWP mapping project here:

    • joe
      Posted Jan 9, 2016 at 11:31 AM | Permalink

      Sebastian – I appreciate both your work and Steve’s work on the subject. As a layman, I cant comment on whether the mwp was warmer or cooler in comparison to today, primarily because I question whether the proxies have a high enough resolution to provide sufficient insight to the question.

      My confidence in the science is not enhanced by A) the claim of the regional nature of the mwp for the 300+ years and B) the “scientific ” explanation of why the regional weather event lasted 300+ years over the same spot of the globe. Neither the claim that it occurred or the scientific reason seems remotely plausible.

      • Posted Jan 9, 2016 at 12:52 PM | Permalink

        Proxies are surely not easy and need a lot of know how to interpret. But it is far better to have this data than no data. There have been significant climate fluctuations in the past which we first need to explain before modelling the future of the climate. Models need to be calibrated before they are fit for prognoses.


2 Trackbacks

  1. […] […]

  2. […] may have been actually associated with cooler summer temperatures, as Steve McIntyre documents on Climate Audit-Blog. Meanwhile also Balascio et al. 2015 appear to have adopted the new varve model for Big Round Lake […]

%d bloggers like this: