Humlum et al: Identifying natural contributions to late Holocene climate change

Posted: December 10, 2011 by tallbloke in climate, solar system dynamics

New contributor ‘Jostemikk’ has alerted us to an interesting recent paper from Norwegian scientists Humlum et al.
This paper examines natural climate cycles in the GISP2 Greenland ice core record and the Svalbaard temperature series.

Ole Humlum a,b, Jan-Erik Solheim c, Kjell Stordahl d
a Department of Geosciences, University of Oslo, Oslo, Norway
b Department of Geology, University Centre in Svalbard (UNIS), Svalbard, Norway
c Institute of Theoretical Astrophysics, University of Oslo, Oslo, Norway
d Telenor Norway, Finance, Fornebu, Oslo, Norway
a r t i c l e i n f o a b s t r a c t
Article history:
Received 16 June 2011
Accepted 9 September 2011
Available online 18 September 2011

A couple of  key passages I spotted were these:

The causes of long (millennial scale) climate changes are
generally poorly understood, and the issue is important for understanding
the natural climate variability, as illustrated by ice and
ocean cores. Also the lack of a CO2 cycle at millennial time scales underscores
the independence of such long climate variability from anthropogenic
greenhouse enhancement.

The warming following the Little Ice Age is generally perceived as a
natural recovery from the previously cold period, and the effect of anthropogenic
greenhouse enhancement is assumed to become important
only after 1975 (IPCC, 2007). Our simple cyclic model (Fig. 8.) is
able to forecast the main features of this recorded warming until
2010, underlining that a significant part of the 20th century warming
may be interpreted as the result of natural climatic variations, known
to characterise at least the previous 4000 years.

The full paper is available here:

  1. Stephen Wilde says:

    That is a sound starting point for a bottom up (oceanic) approach but they need to incorporate a top down (solar) approach and study the interaction.

    Anyway, the upshot is that top down or bottom up, the climate outturn is a redistribution of the permanent climate zones as the system responds in a negative manner to any forcing that tries to alter the global equilibrium energy content (not temperature) set by atmospheric pressure, solar input and the physical properties of water especially the forces required to break the bonds between water molecules (which are themselves pressure dependent).

    I’m not expecting anyone to suddenly see what I mean and agree with me. This post (and the rest of my work) is for future reference when I believe it will become obvious to all.

    God, I hope that doesn’t sound TOO arrogant 🙂

  2. tallbloke says:

    Humlum et al’s fig 8 with their prediction based on 3 longterm cycles is interesting. I think they’ve done a good job in general but it doesn’t take acount of the top down solar effect Stephen is highlighting.
    Nor does it take into account the bicentennial crashes in solar activity such as those seen in the Maunder and Dalton Minima, and perhaps just around now. This will affect their prediction for the rest of the C20th if it comes to pass.

    Fig. 8. Central Greenland (GISP2) surface temperature the past 4000 years (blue line). Natural cycle modelled and forecasted data are shown by the green line. The coefficient of
    determination (r2) for the hindcasting period is 0.63. The timing of certain historical warm periods in Europe is shown by red text, and the timing of Greenland cultures are
    shown in grey. The overall declining linear temperature trend during the last 4000 years is −0.0052 °C per decade, and is assumed to continue when plotting the forecasted
    data beyond 1855.

  3. Roger Andrews says:

    Haven’t looked at the GISP2 results except to confirm that the ice core temperatures are correct.

    However, it’s necessary to treat the Svalbard results with considerable caution because the Svalbard temperature record, to put it as charitably as possible, is not of the standard required to support detailed spectral analysis (details on request if anyone’s interested). Humlum et al. would have done a lot better to use the record from Jan Mayen, which is only nine years shorter and which is a lot more robust.

    . .

  4. tallbloke says:

    Thanks for the data analyst expertise Roger. I’m primarily interested in the long term cycles detected in the GISP2 series. I like their review of the literature in that section. Useful.

    Here’s a juicy quote from email #5111

    >2. About the past 1000 years. The very best boreholes are in the ice in
    >Greenland and Antarctica (low noise big signal environments). The
    >geothermal reconstructions for Greenland continue to show a temperature
    >maximum around 1000 AD (see the paper by Dahl-Jensen et al, Science 282,
    >268-271, 1998) that is about 1K above present day temperatures, and the
    >temperature around 1600 as the coldest (not by much) of the past
    >millennium. The 19th century cooling is also apparent.
    >We will review carefully the borehole database for deep high quality data
    >that may shed some additional light on the full millennium in different
    >regions. We might even think about different parameterizations for
    >some of the best data. But it will be very difficult to make the MWP go
    >away in Greenland.

  5. P.G. Sharrow says:

    Stephen Wilde says:
    December 10, 2011 at 7:44 pm
    ” Anyway, the upshot is that top down or bottom up, the climate outturn is a redistribution of the permanent climate zones as the system responds in a negative manner to any forcing that tries to alter the global equilibrium energy content (not temperature) set by atmospheric pressure, solar input and the physical properties of water especially the forces required to break the bonds between water molecules (which are themselves pressure dependent).”

    And THAT says the whole thing! pg

  6. Willis Eschenbach says:

    There is no “Svalbard (78°N) surface air temperature series 1912–2010”, that’s a modeler’s fantasy. It is a pastiche of a variety of different records from different times, including stations on both sides of the island and thus in different climate zones.

    The authors say:

    The Svalbard (78°N) data series is unique by being the longest meteorological record from the High Arctic, a region usually considered very sensitive to global climate changes.

    Again, that’s total horse-derived waste products. The “Svalbard surface air temperature series” is actually a combination the Istfjord Radio dataset (1911-1976) and the Svalbard Luft dataset (1977-present). The Istfjord Radio station changed location three times. It was taken over by the Germans during WWII. this record is a spliced record from three Isfjord
    Radio locations in one area (with data collected by two different countries at different
    times), plus the Svalbard record from a location in a different climate zone 43
    kilometres away. Not only that, but there is no overlap between the Svalbard and Istfjord stations, so there is no way to compare them and adjust them.

    The authors of the paper somehow forgot to mention all of that. They also do a Fourier analysis and a wavelet analysis of a composite of four station locations in two climate zones without even mentioning that there might be, you know, a few problems with that.

    That’s very shabby work, tallbloke. Either they didn’t know the history of the Svalbard station, or they thought it was unimportant … they didn’t even mention Isfjord Radio station (in a different climate zone), or its station moves.

    Sorry, gets no points from me at all. People that do Fourier and wavelet analysis with a dataset spliced together from four different stations, with no overlap between the stations, should be banned from climate science until they mend their ways.


  7. tallbloke says:

    Hey, Willis, seasons greets.
    Thanks for the in depth summary, you and Roger Andrews up the thread concur on the Svalbaard dataset. I’ve pointed out some possible issues with their cycles methodology in relation to the long term GISP2 Greenland data as well.

    Please note I introduced the paper not as being correct, but as ‘interesting’, which in parts it is. A ‘curate’s egg’ of a paper is my overall impression so far. 🙂

  8. Jostemikk says:

    Thank’s for putting the spotlight on this paper, Tallbloke!

    Willis is completely on target regarding the Svalbard series. To make it even worse, there’s a lot of years missing, or in such bad shape that the Norwegian Met Office seems embarrassed to present it to the public, as can be seen here:,39035,73_39049&_dad=portal&_schema=PORTAL&6009_BATCHORDER_3197941

    Like several other countries during the second world war, Norway with Svalbard was occupied. Lots of Met stations moved, were destroyed etc.

    Since Jan Mayen is mentioned as a better alternative than the Svalbard series:

    1940, station destroyed by act of war.
    1941, station rebuild by norwegian forces on Iceland.
    1941, station moved 500 meters SSE of Nordlagunen.
    1941, station moved 3 km ENE to Jacobsdalen.
    1941, station moved 3 km SW to the vestside of Soerlagunen.
    1943, station moved 400 meters to the NW.
    1946, station moved to the north end of Nordlagunen.
    1949, station moved 200 meters to the NE.
    1962, station moved 11.3 km to the SW to Helenesanden.
    1969, station moved 24 meters to the SE.

    For some reason there’s no station history for the Jan Mayen Met station before 1940. For the bad translation you should blame me, not the Norwegian Met Office.

    Personally I’d preferred the Karasjok series from Finnmark in the northern Norway, but I’m far from being an expert.

  9. tallbloke says:

    Jostermikk: your input here is very welcome, thanks for your addition to our knowledge. In addition to surface air temperature, there are sst records from the surrounding sea. I wonder if Roger has looked at those.

    Some of this sst data was used by Harald Yndestad in his study highlighted two years ago here:

  10. Roger Andrews says:

    Thanks to Willis Eschenbach for detailing the problems with the Svalbard records, which I link to here for reference:

    Isfjord Radio:

    To enlarge on Willis’s comments, when we plot these records on the same graph we find that the 1977 station relocation from Isfjord to Svalbard was accompanied by a baseline shift of about -2C. But was it really -2C, or was it -1C, or was it -3C? There’s no way of knowing because, as Willis points out, there’s no overlap between the two records. In fact there’s about a year of missing data between them.

    Other problems are 1) the divergences between the two Isfjord records before 1935 and 2) the strong warming between 1910 and about 1920, which isn’t seen in any other North Atlantic record, nor for that matter in any other record anywhere else on earth.

    And thanks to Jostermikk for the metadata on Jan Mayen. However, while this station has obviously had a checkered history it does give reasonably robust results. The four records used to produce the combined record track each other closely during periods of overlap:

    And the combined record matches the records in surrounding areas (Iceland, Greenland, Norway) quite closely, which Svalbard certainly doesn’t.

  11. Roger Andrews says:


    “But it will be very difficult to make the MWP go away in Greenland.” I would say impossible. Not even the Team can stretch the data that far.

    On your curate’s egg hypothesis, however, I’m still looking for the good things in this paper. We’ve rubbished the Svalbard analysis, and now that I’ve had time to look at the GISP2 analysis my reaction is “so what?” I can clearly see the Medieval, Roman and Minoan warm periods in the data. I don’t need Fourier transforms and wavelet analysis to pick them out for me. And I could come pretty close to Humlum et al.’s green-line projection using a pencil and a human eyeball Mark 1.

    But maybe I’m missing something.

  12. tallbloke says:

    Hi Roger,
    well, here at the talkshop, we’re always interested in the study of combinations of cycles when they potentially offer explanatory power. For me, the interesting past of the Humlum et al paper is this:

    the character (persistence, period and amplitude) of such cyclic patterns
    is difficult, as the variations quite often come and go, lasting
    only for a limited period at each appearance. For this reason, they may
    prove difficult to characterise fully from a normal Fourier power spectrum.
    Especially the dynamics over time of the individual cycles can
    be difficult to analyse. However, as will be shown below, Fourier analysis
    remains an extremely valuable tool for the identification of such
    recurrent natural climate variations.
    To overcome the problem encountered when cyclic variations
    change their period and amplitude, we here also employ wavelet analysis
    to identify and describe oscillating variations in climate series as a
    supplement to the Fourier analysis. Wavelet analyses are able to pick
    up even oscillations that last for a relatively short time and change
    their phase between one appearance and the next.”

    Although we’ve found that we can only stand each other’s company for limited periods at a time, I doff my cap to Paul Vaughan for his awareness of this, and the pioneering work he has done in taking a morlet approach to solar-planetary and climate cycles.

    As an example of our own work on cycle analysis in relation to climatically important variables, I would offer the post Tim Channon and I did a while back on deconstructing lean’s TSI reconstruction into cyclic periods which are close to periods found in the planetary periods and the Fourier analysis of climate data.

  13. Roger Andrews says:


    Point taken. Combining wavelets and Fouriers may improve cycles analysis and lead to better predictive capabilities, and if so your curate’s egg hypothesis is confirmed. (And yes, I’m familiar with the Lean TSI analysis – see )

    Now here’s another data problem. There are four Greenland ice core records, but the GISP2 record used by Hummum et al. is the only one that shows the Medieval, Roman and Minoan Warming Periods. The other three – NGRIP, GRIP and DYE3 – don’t show any of them. (Data at )

    Based on these results I’m now retracting my earlier statement that it’s impossible to make the MWP go away in Greenland. It’s actually very easy. All we need do is replace GISP2 with NGRIP, GRIP or DYE3 and hey presto, no MWP. (And not much in the way of cyclicity either – the NGRIP, GRIP and DYE3 records are essentially monotonic over the last 4,000 years.)

  14. tallbloke says:

    GISP2 +NGRIP+GRIP+DYE3 turns up over 2000 results on google. The fourth from the top is a wikipedia article about DYE3:

    “The follow-up U.S. GISP2 project drilled at a glaciologically better location on the summit (72°36’N, 38°30’W, 3200 masl). This hit bedrock (and drilled another 1.55 m into bedrock) on July 1, 1993 after five years of drilling. European scientists produced a parallel core in the GRIP project. GISP2 produced an ice core 3053.44 meters in depth, the deepest ice core recovered in the world at the time.[28] The GRIP site was 30 km to the east of GISP2. “Down to a depth of 2790 m in GISP2 (corresponding to an age of about 110 kyr B.P.), the GISP2 and GRIP records are nearly identical in shape and in many of the details”.

    It’s not my expertise area, so feel free to disagree with the Wiki author whilst I munch some popcorn. 🙂

  15. Roger Andrews says:


    Sending you some graphs to look at while you munch your popcorn

  16. tchannon says:

    popcorn futures +21
    One of the snappiest performing commodities in recent years

    Mexicancun jumping beans are dead.
    Perdurbanations are a good bye

    Weather is showing another severe low (945mb) Friday the coming week. Lets hope that does not happen.

    Around 16th Dec

  17. Roger Andrews says:


    Didn’t get a response from you on my email, so I’m posting GISP2 (orange) vs. GRIP (blue) over the last 4,000 years as a tinypic. Hope it works.

  18. Jostemikk says:

    I’m happy you appreciated my tiny contribution, Tallbloke! I’ve been a long time lurker here on The Talkshop, and just love it!

    Here are the three locations used in the Svalbard series:

    We are not talking about “two climate zones”.

    Here are the years with common temperature data for Isfjord Radio and Longyearbyen:

    And a comparison between Jan Mayen and Longyearbyen:

    As can be seen, there is one big anomaly starting ca 2004. Willis E. mentioned the 2006 april temperatures, and Anthony wrote about Svalbard in May 2005 in “Where the!@#$% is Svalbard’s Weather Station?”

    The April temperatures on Svalbard Airport 1990-2011:

    Don’t forget Ole Humlums big contributions on his website (Yes, Humlum as in Humlum et. al). heat island effects in Longyearbyen heat island effects in Oslo

  19. tallbloke says:

    Hi Jostermikk,

    It did make me smile when Willis complained about the stations being in two different climate zones. I had a picture in my mind of a wasteland of ice and snow on one side of the island, and people lying in hammocks under palm trees on the other side… 🙂

  20. Roger Andrews says:

    I showed in the graph above why it was questionable to use the GISP2 ice core record. Now here’s why you don’t want to use the Svalbard surface temperature record.

    First a plot of surface air temperature anomalies from six long-term records in and around the North Atlantic – doesn’t matter which is which. They track each other quite closely, so we can reasonably conclude that that there’s nothing seriously wrong with any of them. (This, incidentally, is the way you verify surface temperature records).

    Now let’s superimpose Svalbard

    “Nuff said.

  21. Contrari says:

    Just a small comment to Eschenbach concerning Svalbard; it was not taken over by the Germans in WW2. They made a one-day naval raid there in 1943, apart from that the islands were on Allied hands.