The MD99-2275 core offshore Iceland is a very high-resolution ocean sediment core, results of which over the past millennium have been discussed here from time to time. Alkenone and diatom results for the last millennium have been available for about 10 years. MD99-2275 results were used in PAGES2K Arctic and Hanhijarvi 2013, also Trouet et al 2009, but not Marcott et al 2013. Because of their high resolution and because of their extension through box cores into the 20th century, it’s gratifying that high resolution results are now becoming available through the Holocene. In 2012, alkenone results were extended back at high resolution to 4500 BP and last week, Jiang et al archived their diatom results back to 9271 BP in the early Holocene. In today’s post, I’ll show the new results from Jiang et al and show an interesting comparison to Marcott.
First, here are the new results from Jiang et al 2015, Solar forcing of Holocene summer sea-surface temperatures in the northern North Atlantic, published in Geology rather than Nature or Science (data here). On the left is the data back to the early Holocene; on the right, a blowup of the last millennium. The diatom data represents summer temperatures and shows the familiar Milankovitch decline over the Holocene. The data for the last millennium also shows a marked MWP, relative to the Little Ice Age, though in a Holocene context, its the Little Ice Age rather than the MWP that is distinctive. In Hvitarvatn, Iceland lake sediment data, discussed here on several occasions, the 19th century was exceptionally cold on a Holocene basis (marked by ice-rafted debris), thus requiring identification of thick 19th century varves with cold rather than warmth (PAGES2K using this data upside down.)
Because the Jiang 2015 dataset ends in 1875, it seems to leave the modern comparison frustratingly unanswered. however, Ran et al 2011 had combined two contiguous box cores with MD99-2275 to construct a continuous diatom record from AD1000 to AD1999. In the figure below, the Ran et al 2011 diatom series is shown in the right panel in red: its 20th century values are essentially unchanged from 19th century values. (In the left panel, I’ve only shown post-1875 Ran et al 2011 values (red).)
In the next figure, I’ve compared the high-resolution MD99-2275 record to the Marcott NXT reconstruction. For comparison, I matched the mean of the MD99-2275 record to Marcott over the 5000-1000 BP period. The absolute values of the decline in temperature from the early Holocene to the last millennium (~1.8 deg) is surprisingly similar in the two series. The two main differences between the series are the very strong warmth in the MD99-2275 record around 7500 BP not shown in the Marcott record and the absence in the MD99-2275 record of the very erratic Marcott results over the past 200 years.
During the Holocene, high-latitude summer insolation has decreased dramatically (over 40 wm-2), while changes in annual insolation have been very modest. Nearly all biological proxies are responsive to summer temperature, rather than annual temperature. The MD99-2275 temperatures are summer temperatures. The Marcott dataset is dominated by alkenone series, which are also responsive to summer temperatures and should therefore be interpreted as a reconstruction of summer temperatures, rather than annual temperatures – see useful discussion by Richard Telford here here other than his pointless snipe at me about criticizing the fake Marcott uptick ). While it is an article of faith in Mannian paleoclimate that annual and summer temperatures are correlated, this definitely ceases to be the case on a millennial and Holocene scale, as precession and obliquity change.
In geophysics, explorationists have considerable experience in working with datasets of very different quality. Their practice is to work out from the “best” high resolution data rather than contaminate high-resolution data with poor data. Reasonable people can disagree on precisely how to do this in paleoclimate. However, it is possible for methods to be objectively bad and even wrong – Mannian principal components is one example, the Marcott uptick artifact is another. As noticed at CA, the Marcott uptick was a botched artifact of station dropout, exacerbated by Marcott’s unwitting reallocation of core top dates. Marcott et al ought to have issued a corrigendum, but unfortunately preferred to pretend infallibility. As readers are aware, D.C. judges have ruled that it is a tort in D.C. to question Mann’s reasoning; it is presently unknown whether it is also a tort in D.C. to question the reasoning of Marcott and other coauthors of Marcott et al 2013 or climate scientists in general, or whether it is only a tort to “question” Mann’s “reasoning”.