Climate Science Glossary

Term Lookup

Enter a term in the search box to find its definition.


Use the controls in the far right panel to increase or decrease the number of terms automatically displayed (or to completely turn that feature off).

Term Lookup


All IPCC definitions taken from Climate Change 2007: The Physical Science Basis. Working Group I Contribution to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Annex I, Glossary, pp. 941-954. Cambridge University Press.

Home Arguments Software Resources Comments The Consensus Project Translations About Donate

Twitter Facebook YouTube Pinterest

RSS Posts RSS Comments Email Subscribe

Climate's changed before
It's the sun
It's not bad
There is no consensus
It's cooling
Models are unreliable
Temp record is unreliable
Animals and plants can adapt
It hasn't warmed since 1998
Antarctica is gaining ice
View All Arguments...

Keep me logged in
New? Register here
Forgot your password?

Latest Posts


Esper Millennial Cooling in Context

Posted on 21 July 2012 by dana1981

A new paper in Nature Climate Change from Esper et al. (2012) develops a 2,000-year local summer temperature reconstruction based on 587 high-precision maximum latewood density (MXD) tree ring series from northern Scandinavia, using living and subfossil pine trees from 14 lakes and 3 lakeshore sites in the region (>65°N).  The results are shown in Figure 1.

Esper fig 2

Figure 1a: The Scandanavian summer temperature reconstruction extends back to 138 BC highlighting extreme cool and warm summers (blue curve), cool and warm periods on decadal to centennial scales (black curve, 100-year spline filter) and a long-term cooling trend (dashed red curve; linear regression fit to the reconstruction over the 138 BCAD 1900 period). Estimation of uncertainty of the reconstruction (grey area) integrates the validation standard error (±2 × root mean square error) and bootstrap confidence estimates. Figure 1b: Regression of the MXD chronology (blue curve) against summer temperatures (red curve) over the 1876–2006 common period. Correlations between MXD and instrumental data are 0.77 (full period), 0.78 (1876–1941 period), and 0.75 (1942–2006 period).  Figure 2 from Esper et al. (2012)

The reaction from climate contrarians in the popular media has been predictable; for example, they have made various outlandish claims about the study somehow debunking global warming and suggested that other climate scientists will have to "poo-poo" the results of Esper et al.  Unfortunately, Jan Esper himself fueled this climate contrarian confusion by exaggerating the impact of his study's results in a press release, where he proclaimed:

"Our results suggest that the large-scale climate reconstruction shown by the Intergovernmental Panel on Climate Change (IPCC) likely underestimate this long-term cooling trend over the past few millennia."

However, as we discuss below, this conclusion does not follow from Esper et al. (2012) for several key reasons.

Putting Esper et al. in Proper Context

Most importantly, as noted above, the Esper et al. temperature reconstruction is a local one (from northern Scandanavia), and of summer temperatures.  The IPCC on the other hand focused on annual Northern Hemisphere temperature reconstructions; thus the Esper et al. results are not directly comparable to those discussed in the IPCC report

In another Arctic regional temperature study, Kaufmann et al. (2009) discussed the same Arctic summer cooling over the past several millennia noted by Esper et al., which they find is primarily due to the Earth's elliptical orbit:

"The millennial-scale cooling trend in our temperature reconstruction correlates with the reduction in summer insolation, which was primarily driven by the precession of the solstices around Earth’s elliptical orbit. Over the past 2000 years, summer (JJA) insolation at the top of the atmosphere decreased by about 6 W m−2 at 65°N. The forcing was weaker at lower latitude, especially for the early summer."

Kaufmann et al. suggested that a decrease in incoming solar radiation in the Arctic due to the Earth's orbital changes was amplified by various feedbacks like expanding Arctic sea ice and increased snowfall, both of which increase the local surface albedo, reflecting more sunlight and causing additional cooling.

In fact, Esper et al. note that the orbital forcing and feedbacks identified by Kaufmann et al. are a key difference between Arctic and Northern Hemisphere temperatures:

"the [orbital] forcing and radiative feedbacks decrease towards equatorial regions."

As noted by the climate scientists at RealClimate,

"insolation forcing is near zero at tropical latitudes, and long-term cooling trends are not seen in non-tree ring, tropical terrestrial proxy records such as the Lake Tanganyika (tropical East Africa) record (Tierney et al, 2010)"

Lake Tanganyika lake surface temperature
Figure 2: Lake Surface Temperature from Lake Tanganyika palaeorecord for the past 1,500 years, measured in core KH1 (red line) and MC1 (dark red line) with 95% error bars (orange shading).

Thus we should expect annual Northern Hemisphere reconstructions - which include data from both the Arctic and equatorial regions - to exhibit less of an orbitally-caused cooling trend than reconstructions of summer Arctic temperatures like those in the Kaufmann and Esper studies.  In an interview with Bud Ward, one of the co-authors of Esper et al., Robert Wilson noted:

"Our paper is for northern Scandinavian summer temperatures so extrapolating to large scale annual temperatures is not really correct."

Esper's Divergence Non-Problem

As the RealClimate post also notes, perhaps the most useful contribution of Esper et al. lies in the fact that they may have resolved the so-called "divergence problem."  This refers to the fact that in some tree rings, mostly from high northern latitudes like those examined by Esper et al., temperatures appear to have declined since about 1960, whereas we know from thermometer measurements that real-world temperatures have actually increased over this period.  However, the MXD data in Esper et al. appear to track instrumental temperatures quite well even in recent decades, as illustrated in Figure 1b above.  Note the strong correlations between MXD and instrumental temperature data in the Figure 1b caption.


Esper's claims that he has identified a shortcoming in Northern Hemisphere millennial temperature reconstructions stem from the fact that tree ring width (TRW) temperature reconstructions exhibit less of a long-term cooling trend than their maximum latewood density (MXD) reconstruction.  Thus Esper suggests that including TRW data will lead to an underestimate of the long-term Northern Hemisphere cooling trend.

However, the RealClimate post notes that this hypothesis is not borne out in the data.  For example, Moberg et al. (2005) created a Northern Hemisphere millennial temperature reconstruction without using any tree ring data, and it has among the smallest long-term cooling trends of all such reconstructions (just -0.06°C per millennium from 0 to 1900 AD, as opposed to the -0.31°C per millennium trend in Esper et al.).  The long-term annual Northern Hemisphere cooling trends in Ljungqvist et al. (2010) (-0.18°C per millennium from 0 to 1900 AD) and Mann et al. (2008) (-0.19°C per millennium from 300 to 1900 AD) are similar to but less than the summer northern latitude trend in Esper et al., as we would expect.

Another Hockey Own Goal

There is a key point that often gets lost in discussions about 'hockey sticks' - larger long-term natural variability is inconsistent with a low climate sensitivity.  Low climate sensitivity is the lynchpin for all climate contrarianism; therefore, arguing for a highly variable climate is a contrarian own goal (meaning they accidentally score against their own team).

Climate contrarians tend to tie themselves up in knots trying to argue that the hockey stick is broken and the Medieval Warm Period was warmer than today.  We'd be better off if temperatures were relatively stable and invariable as in the original 1999 'hockey stick,' which would indicate that the climate is relatively insensitive to CO2 and other forcings, which in turn would suggest that we have relatively more time to reduce our greenhouse gas emissions before we're hit by the really nasty consequences of climate change.

We also need to bear in mind that the relative absolute global temperatures of Medieval and Roman times as compared to current temperatures are not all that important.  In all likelihood we have already matched or exceeded the peak average global temperature during those prior periods, but far more important is the current rate of warming, which greatly exceeds anything in the prior several thousand years.  Even if we hadn't quite yet matched peak Medieval temperatures, we would certainly blow past them very soon, with several degrees more global warming likely to come this century.  Current temperatures are not the concern - everybody involved would be happy if we could maintain the current average global temperature.  It's the rapid rate of warming that concerns climate scientists.

Esper et al. - A Valuable but Overstated Contribution

In the end, Esper et al. did produce a valuable result by potentially resolving the divergence problem and identifying the apparent shortcomings of TRW data.  However, it's also important not to overstate the relevance of the study's results, which do not appear to have identified any significant shortcomings in annual Northern Hemisphere millennial temperature reconstructions, and certainly have not undermined the robust human-caused global warming theory.

0 0

Bookmark and Share Printable Version  |  Link to this page | Repost this Article Repost This


Comments 1 to 18:

  1. As far as I can tell, the Earth has been cooling slowly since the end of the Holocene climatic optimum, with some ups and downs along the way... until the Industrial Revolution.

    This paper does not seem to alter this impression in any meaningful way.
    0 0
  2. Generally true Composer. The climate contrarian spin from the paper was "Roman and Medieval times were as warm as present", which aside from being rather irrelevant (as discussed in the post), is not something this paper is capable of showing, except for temps specifically in northern Scandanavia.
    0 0
  3. I know that the MWP is a very slippery beast but I thought it was supposed to run from about 900 to 1300. The period of high temperatures in Fig 1 appears to be much earlier than this.
    0 0
  4. Sceptical Wombat @3:
    "I know that the MWP is a very slippery beast but I thought it was supposed to run from about 900 to 1300. The period of high temperatures in Fig 1 appears to be much earlier than this."

    Not only that, the LIA is gone in their graph, at least the LIA we're used to. Their LIA ends a little after 1500, then temps are mostly stable until the 19th century. I'm sure there's a lot of useful information in these new proxies, but they have to be integrated with many other regional studies in order to understand what was going on at a hemispheric level, let alone a global one.
    0 0
  5. Sceptical Wombat - that gets to another inconvenient fact about the MWP. It wasn't simultaneously global. In various geographic locations there were 'MWP' peaks at different times. China wasn't hot at the same time as northern Europe, for example. Some areas were relatively hot from 900-1000, others from 1000-1100, and so forth.
    0 0
  6. Could you explain how exactly they resolved the divergence problem?
    0 0
  7. It's just a result of their MXD reconstruction that it matches instrumental temperatures well, Martin. If you want a more technical answer, that's beyond my paygrade :-)
    0 0
  8. The apparent resolution of the Divergence Problem is what interests me about the study, but either I'm denser than the maximum latewood measurements or the solution wasn't discussed here or at RC. From what I could glean in the comments there, it seems like the problem simply didn't occur in their analysis, not that they've come up with a methodology that removes it.
    0 0
  9. Seems I was a little slow in posting!
    0 0
  10. Wheels OC -
    "it seems like the [divergence] problem simply didn't occur in their analysis, not that they've come up with a methodology that removes it.
    I believe that's correct, but I'm far from a dendrochronology expert.
    0 0
  11. Note the authors say
    "Calibration/verification with instrumental data is temporally robust and no evidence for divergence was noted."
    So it does seem that the divergence problem simply isn't an issue with their MXD analysis, not that they made a specific effort to remove the divergence.
    0 0
  12. WheelsOC at 04:21 AM on 21 July, 2012

    I understood it the same way as you did, like on Jim's inline response here:

    Relevant part below:

    They make nothing of that issue, barely even mentioning it in passing, and then never again. It's likely a fortuitous result, possibly related to use of density data and possibly not, but not one that is new--there have been numerous studies in which divergence at decadal scales was weak or absent.
    0 0
  13. Esper ran both MXD as well as TRW on the samples (as well as early width and density). They show not late diversion problem (post 1960). However, in figure S6 in the supplementary information, there is a huge divergence from the beginning of the data until after 900AD.

    This divergence is not discussed in the paper, unless I missed something.

    What might this indicate?

    Another potential nit is from figure S10, where the correlation between the instrument record and the tree rings breaks down completely circa 1910 (looks like 1911 to me, but these are 15 year running averages). Volcanic eruptions from that time seem unremarkable (see here), though 1911 was noted for an extreme weather event in November in the US midwest (the great blue norther). Esper notes the breakdown, but does not attempt to explain it or indicate what it means to the reliability of the proxy record.

    Anybody know a enough about dendroclimatology or tree biology to shed some light here?

    (I was unable to find an SkS article dealing with the reliability of tree-ring proxy techniques, beyond the divergence problem. It is being attacked on the typical fake skeptic sites, so an article may have value.)
    0 0
  14. Lambda 3.0,

    With "Figure S6" I assume you mean Figure S9, right?

    I'm not a treeringologist, but know a little about stats.

    You have to look carefully what is being plotted. The year number at the horizontal axis is the end year of an interval, of which the start year is 138 BC. Over this interval, a linear regression is performed.

    The core point is that the length of the regression period is very short at the left (538 years) and very long at the right (2138 years), and the data regressed is very "wiggly". Of course the trend variations due to the wiggling will then be considerably greater at the left end of the graph than at the right end. As we see.

    We also see that wood density gives different results then ring width. Due to the way this plot was made, also that difference has an effect on the plotted trends that grows going to the left (I suspect the tree-ring score differences themselves aren't any greater for the earlier times).

    About Figure 10, you have to look again at what is being plotted: 15-year moving-window correlations. So the "dip" at 1912 (I think) shows the correlations over the period 1905-1920. If you look at the upper part of the plot, you see that in this window, there happens to be only little variation in instrumental ("real") temperatures (black); these are almost constant over the window period. So, there is no temperature signal to cause corresponding variations in the proxy record, and no hope to extract good correlation values. The values we see (lower graph, colored curves) are more or less in the noise.

    The instrumental records for different stations show much better correlations among themselves (grey curves), because, well, purpose-built weather stations are more fit for purpose and less "noisy" than proxies-of-opportunity ;-)
    0 0
  15. what i find interesting is that while several deniers seem to accept this Esper paper, they didn't bring the usual, but they used Computer Models meme this time :D
    0 0
  16. Aanthanur - even more shocking, they didn't bring the usual "tree ring proxies are unreliable" meme this time.

    Climate contrarian skepticism is very selective, which is why they're not actual skeptics. They're only 'skeptical' until they find information to support their pre-determined conclusion. That's not skepticism, that's confirmation bias.
    0 0
  17. Dana-the link to Fig. 1 is broken.Heads up
    0 0
  18. Thanks tmac, image fixed.
    0 0

You need to be logged in to post a comment. Login via the left margin or if you're new, register here.

The Consensus Project Website



(free to republish)



The Scientific Guide to
Global Warming Skepticism

Smartphone Apps


© Copyright 2014 John Cook
Home | Links | Translations | About Us | Contact Us