RealClimate logo

Background on the role of natural climate variability in West Antarctic ice sheet change.

Filed under: — eric @ 14 August 2019

This is a summary of some of the key details that underpin the discussion of anthropogenic vs. natural forcing in driving glacier change in West Antarctica. This is useful background for the paper by Holland et al. (2019), discussed in another post (here).

We’ve known for some time that Pine Island Glacier (PIG) and Thwaites Glacier, the two largest of several fast-moving outlet glaciers that drain a large fraction of the West Antarctic ice sheet (WAIS) into the Amundsen Sea are critical to the stability of the ice sheet as a whole. Way back in 1979, Terry Hughes argued that these glaciers make the WAIS susceptible to large-scale collapse, which almost certainly occurred during some previous interglacial periods and contributed several meters to sea level rise. In the mid-1990s it was discovered that melt rates under the floating portion of the glaciers was orders of magnitude greater than previously assumed (Jacobs and others, 1996). Shepherd and others (2002, 2004) showed that this melting at the margin had resulted in thinning upstream, and retreat of the grounding line (the point at which the glacier goes afloat). It quickly became obvious that melt rates must have increased in the preceding few decades. Otherwise these glaciers would already have retreated even further. The culprit was suspected to be the increased inflow of Circumpolar Deep Water (CDW) on the Antarctic continental shelf, where it contacts the floating margins of the glaciers.

These ideas were validated in 2010 by direct observations made by an autonomous underwater vehicle under the PIG ice shelf (note: an ice shelf is the floating portion of a glacier; it should not be confused with the continental shelf). The submarine observations (Jenkins and others, 2010) showed that CDW was flooding the cavity below PIG, >30 km upstream of areas that were at least partially grounded as recently as the early 1970s. Although CDW is just a few degrees above freezing, it provides enough heat to melt the ice from below at rates in excess of 50 meters (vertical) per year. Independent estimates derived from satellite observations of ice speed and thinning rates (e.g. Rignot and others, 2008) agreed well with such numbers, sealing our basic undestanding of what was going on.

Now, the reason that glacier melt in West Antarctica has increased is not because Circumpolar Deep Water itself is getting warmer (although it probably is). Instead, it’s clear that more CDW is getting from the ocean surrounding Antarctica onto the Antarctic continental shelf and reaching the glacier margins. As shown in a seminal modeling study in Geophysical Research Letters (Thoma et al., 2008), how much CDW gets onto the shelf is strongly influenced by the strength and direction of the winds at the edge of continental shelf. It is useful to picture this as wind-driven upwelling (Ekman pumping). Westerly winds (blowing fromthe west) along the edge of the continental shelf divert cold surface waters northward because of the Coriolis effect. This surface water is replaced by the upwelling of warm water from below. The upwelled CDW then makes it’s way along the continental shelf and up to (and below) the floating ice shelves. While this picture is greatly oversimplified*, the essential insight is that stronger westerlies (or merely weaker easterlies) along the shelf edge should tend to cause more CDW to get onto the shelf. Numerous modeling studies since the original Thoma et al. work have supported this. Perhaps more important, it’s been verified by observations (more on that below).

Many scientists have assumed that there must be a link between the melting glaciers and the ozone hole. In fact, I got into this area of research partly in response to a press conference given by a well-known glaciologist who made such a claim in response to a reporter’s question, around 2010. We know that ice in West Antarctica is melting from below because it is bathed in warm Circumpolar Deep Water, and that more Circumpolar Deep Water gets onto the continental shelf when the local continental-shelf-edge winds are more westerly. We also know — as I noted above — that the strength of the westerly circumpolar winds around Antarctica has increased, in part because of the depletion of stratospheric ozone. It’s easy to link these separate ideas, but this links largely falls apart under scrutiny. The problem is that these are not the same winds! The circumpolar wind belt is centered around 52°S, very far north of the area of shelf-break winds that Thoma et al. (2008) wrote about, which are centered on about 70°S in the Amundsen Sea. Moreover, there is no correlation between the winds in the Amundsen Sea region and the Southern Annular Mode (SAM) index, a widely-used measure of the strength of the circumpolar westerlies. And the seasonal timing is wrong — the Amundsen Sea winds have increased largely in winter and fall, whereas the influence of the ozone hole is limited to spring and summer.

If it’s not the ozone hole, then what has caused the local winds to change, and to bring more CDW onto the continental shelf (if indeed this is what has happened)? Well, that’s where much of my own work, and that of my coauthors on the new paper, has focussed in the last few years. In 2012, we published a paper articulating the problems with the ozone-hole argument, and pointing out that a much better explanation for the recent glacier changes in West Antarctica was forcing from the tropics. The greatest control on wind variability in the Amundsen Sea is the state of the tropics, which can be characterized roughly by the state of the El Niño-Southern Oscillation (i.e., whether it is a neutral, El Niño, or La Niña year). Just as El Niño event causes widespread climate anomalies in the Northern Hemisphere — such as increased rainfall in southern California — it also causes changes in the West Antarctic. Indeed, the Amundsen Sea is one of the areas on the planet that is most strongly dependent on ENSO (e.g. Lachlan-Cope and Connolley, 2006). Our work showed that the changes in Amundsen Sea winds that had occurred over the last few decades were very well explained by changes in ENSO. We also noted that because big ENSO events had occurred in the past, it was quite plausible that wind conditions not that different than those of today had also occurred in the past.

A number of other papers have supported these findings. Dutrieux et al. (2014) showed that CDW flow onto the shelf, and ice-melt rates under the PIG, decreased during a major La Niña event. Smith et al. (2017) showed evidence that the PIG ice shelf retreated right around the time of really big El Niño event of 1941 (as we speculated in our 2012 paper), and Hillenbrand et al. (2018) showed that CDW may have first begun to flood the Amundsen Sea at about the same time. Finally, Paolo et al. (2018) showed that the influence of El Niño events on West Antarctic glaciers could be measured by satellite observations: El Niño events tend to be correlated with both increased melting from below, and increased snowfall above, and the variations in the altitude of the ice sheet surface (varying by a few tens of cm) can be detected by satellite altimetry.

In short, a lot of research has demonstrated the importance of ENSO in determining conditions in West Antarctica. This has meant that we cannot rule out the idea that natural variability in Amundsen Sea winds, driven by natural variability in ENSO, as the primary driver of observed glacier retreat in West Antarctica.

Our new paper makes the case that while ENSO dominates there is a significant anthropogenic component as well. See the main post on our new paper in Nature Geoscience, here.

*Ekman pumping is actually too weak to account for the observed flow and the reality is quite a lot more complex. For more details on this, see e.g. Arneborg et al., 2012, and Nakayama et al., 2018.

9 Responses to “Background on the role of natural climate variability in West Antarctic ice sheet change.”

  1. 1
    Dan DaSilva says:

    The West Antarctic Volcanic Rift System just so happens to correspond to the areas of notable glacier melt and alleged warming. In contrast, East Antarctica, which holds > 80% of Antarctic ice mass and does not have any known underlying volcanoes, is significantly increasing in ice mass. (Copied with slight edit from another famous climate site who shall remain nameless so as to not antagonize the readers.)

    [Response: Let me guess, WattsUpWithThat? That’s the most likely place for such ignorance. –eric]

  2. 2
    jgnfld says:


    Please show us some data on how the volcanic rift system has suddenly started to put more heat into the base of the WAIS. Other evidence should also be around as well, oh, like a sudden increase in meltwater specifically at the glacier base/rock boundary over the past while but not on the surface.

    Surely you’ve researched this and have the references at hand.

  3. 3
    Dan DaSilva says:

    Reply #1, Eric
    Yes, you are correct. Where is the ignorance?
    1) West Antartic has a system of volcanos
    2) East Antarctica is 80% of the ice mass.
    3) East Antarctica has no known underlying volcanos.
    4) East Antarctica is significantly increasing ice mass.
    5) The implication that volcanic action is causing the melt.

    Thanks for the Response.

    [Response: #4 is not really true (it *may* still be gaining mass, but not signifiantly). #5 is where the ignorance lies. We know where the significant melting is and what is causing it. It’s not volcanos. If you don’t understand how we know this, you need to read the literature. There is a lot of it, most of it very accessible. You might start with this review paper: Cheers — Eric]

    [I should add: “But don’t let my expertise get in the way of your confidence. h/t J. Katzenstein:

  4. 4
    Dan DaSilva says:

    Eric, again thanks for the response and the cartoon. Let me assure you that I have no overblown regard for my own expertise in climate science. I am a climate dummy. That is why I ask questions and read both sides of the issue. Let me say that RealClimate is the only place (that I know of) where you can get a response from a real climate scientist.

    jgnfld, from what I have seen in the comments section here research and references are not required or even the less bit typical.

    [Response: Since your source was Wattsupwiththat, and since the idea you raised has been raised (and shown to be wrong) many times over the years, I have assumed your questions were in bad faith. If I was wrong, I apologize. There are a now quite a lot of good sources on information on the internet, though I always advise reading actual published papers by people who can provide at least some evidence they have expertise in the topic at hand. This doesn’t guarantee good information but it’s a very good filter. Wattsupwiththat has no such filters, and is a source of disinformation, full stop. — best Eric]

  5. 5
    Eric Steig says:

    By the way, Dan DaSilva’s “Wattsup-derived” #4 point is also wrong.

    From the latest work on this (Rignot et al., 2019): The contribution to sea-level rise from Antarctica averaged 3.6 ± 0.5 mm per decade with a cumulative 14.0 ± 2.0 mm since 1979, including 6.9 ± 0.6 mm from West Antarctica, 4.4 ± 0.9 mm from East Antarctica, and 2.5 ± 0.4 mm from the Peninsula (i.e., East Antarctica is a major participant in the mass loss).

  6. 6

    Thanks for a nice review.

    The question it raised for me was the variability of ENSO–what has it been, historically, and how might it change with alterations in climate? If “It quickly became obvious that melt rates must have increased in the preceding few decades,” the implication of the idea that the changes in CDW infiltration are driven ultimately by ENSO trends is that ENSO trends must then exhibit significant variability on multidecadal scales. Mann, Bradley & Hughes (2000) apparently found historical evidence of that, and correlations to more general climate changes:

    The mean state of ENSO, its global patterns of influence, amplitude of interannual variability, and frequency of extreme events show considerable multidecadal and century-scale variability over the past several centuries. Many of these changes appear to be related to changes in global climate, and the histories of external forcing agents, including recent anthropogenic forcing.

    Somewhat to the contrary was Collins, 2004:

    The most likely scenario (p=0.59) in a model-skill-weighted histogram of CMIP models is for no trend towards either mean El Niño-like or La Niña-like conditions.

    (Indeed, I note your companion post to this one says, similarly, that “there is little evidence for a long-term anthropogenic change in ENSO…”)

    But I wonder what a more comprehensive view of the literature reveals. If natural variability in ENSO state is driving the increased melt rates, then there must have been a shift in ENSO ‘balance’ toward a more EN-dominated mean state, sometime in the mid-to-late 20th century. It does appear in MBH (2000), if I’m reading their Fig. 17 correctly. But what does more recent work have to say about this?

    [Response: Well, our analysis in this paper uses the observed tropical SST as a forcing (essentially), and in these model experiments the trend in the Amundsen Sea winds is increased a bit. But the main train is driven by radiative forcing.

    To try answer your main question (while noting that I am *not* very current on this literature), I’ve long thought that part of the trend is owing to the tendency towards more “central tropical Pacific” ENSO. See e.g. this paper: Both models and data seem to be in general agreement here.

    On the other hand, the main trend in the last 50-60 years is towards eastern Pacific cooling, which is not what models do. The very latest/best dicussion of all this is likely this paper by Richard Seager: I haven’t read it carefully yet. I suspect 50-60 years is too short to really learn very much about the forced response.

    I hope that gives you a start. I would not take this my statements as a comprehensive review! — best, Eric]

  7. 7

    “Moreover, there is no correlation between the winds in the Amundsen Sea region and the Southern Annular Mode (SAM) index,”

    The non-correlation between the SAM and ENSO is also known. I am skipping the AGU meeting this year, but what I would have submitted was a way to show how the indices can be correlated. Last week I posted this analysis to my blog. This is part of an extended cross-validation of a fluid-dynamics-based model of cyclic climate indices that we published earlier this year.

    [Response: Tropical forcing accounts at least 25% of the SAM index. See Ding et al., 2012, “Influence of the tropics on the Southern Annular Mode”. –Eric]

  8. 8

    Eric, thanks for your inline and the literature suggestions! Your ‘review’ may not be comprehensive, but your ‘view’ will inevitably be more comprehensive than mine.

  9. 9
    Eli Rabett says:

    Allow a small nit. The ozone hole appearing in the spring is a RESULT of the circumpolar wind isolating the atmosphere above Antarctica and the fact that a lot of reactive oxygen being tied up in ClONO2. Thus stronger circumpolar winds isolating the region for a longer time so the heterogeneous chemistry would have more time to denitrify the Antarctic stratosphere would have a strong effect on the ozone hole, and it no reason to assume that this would not affect melting.

    There may be other reasons, but stronger winter time circumpolar winds are not one of them, it’s effect would only show up in the spring.

    [Response: Eli, I’m confused because you have some double negatives in there. “it no reason to assume that this would not affect melting.” What did you mean?

    In any case, while of course you are right about why the ozone hole appears in spring, the change in the winds that I was referring to is, indeed, a result of the ozone hole. The change in stratospheric heating because of the reduced ozone changes the vertical temperature profile which causes the circumpolar westerlies to spin up. The effect lasts into summer (in models at least). See e.g. Thompson et al., 2011. (Apologies if you know all this already — it will be useful for other readers in any event.)–Eric]