An interesting Greenland-esque paper within the Skeptical Science weekly listings. J.L. Chen, C.R. Wilson, J.C. Ries, B.D. Tapley Rapid ice melting drives Earth’s pole to the east. The article linked by SkSc is here & the full paper here. I would warn folk not to be carried away by the size of the arrow in the paper's figure 2b. The units 'mas' are milliarc-seconds which are by my reckoning about 30mm long.

John Christensen, I would suggest you are wrong to state that Werner et al 2011 (full text here) "argues that there is a strong connection between changes in ocean currents and Arctic surface cryospheric conditions." They present evidence of a " strengthened Atlantic Water inflow after ca 1860 AD. However, ... (other data) suggest cool surface water conditions until the mid of the 20th century. Changes in all studied proxies indicate warmer temperatures for the past few decades (see Spielhagen et al., 2011) and coincide with positive Atlantic Water temperature anomalies and a retreating sea ice margin for the ca last 100 years (Divine and Dick, 2006; Polyakov et al., 2004, Polyakov et al 2005). The writing in Werner et al 2011 is not without ambiguity (I would say rather too much ambiguity) but to attempt to pile on the significance of AW warming in the Arctic prior to the latter half of the 20th century is surely a further misinterpretation of Werner et al 2011 (and his references to boot). And if you wish to extend the AGW narrative (now truly off topic) to incorporate pre-industrial/non-fossil-fuel-use as "having some impact", it would be good to do more than hint at what level of "some impact" you have in mind.

John Christensen. The 'G' in AGW stands for 'global'. The poles climatically act as massive heat sinks, receiving heat from lower latitudes and radiating it out into space. That transfer of heat from lower latitudes into the Arctic can only increase under AGW. Warmer air temperatures & warmer ocean currents are the primary cause of ice loss and rising Arctic temperatures which are amplified by lower albedo due to melted ice and wetter air over the now exposed warming oceans. The Arctic is certainly not the fastest warming region of the planet because of a 40% increase in atmospheric CO2 over the last century-and-a-half.

John Christensen. If you analyse global temperature data as you did the 1979 PIOMAS data, you will note than the present warming began before 1979. Indeed, it can be seen from 1976. It may be that 1976 within its time can be considered as 'just another wobble' in a non-warming world, but it was a warming wobble and thus has all that is required to have Arctic-melting abilities. If PIOMAS covered 1976, it would surely show melting on an annual timescale for that year. The Arctic Ice was not doomed because of 1976 or 1979. It is doomed because 1976 has proved to be more than 'just another wobble,' being the start of a world experiencing AGW.

Hi NeilT. Decadal-scale CO2 changes show in the smoothed red line on this graph.

The spikes in atmospheric CO2 concentrations mentioned above (off topic, mind) are ignorable in the grand scheme of things. However they have their climatical importance inflate by skeptical myth-makers. The biggest climatic wobble is ENSO which a few months later results in a positive wobble in global average temperatures and following later again is a wobble in the rise in CO2. Skeptics love saying this "proves" temperature is the cause of rising CO2 and often graphs are produced to demonstrate their point. There is even the occasional learned paper published saying the same thing, (this one, Humlum et al (2012) even getting peer review, mainly by not putting the effect it investigates into its true climatical context). And it is not put in context because these skeptical messages are all a load of nonsense. This graphic demonstrates it quite well.

I can claim with some confidence that Mahlstein & Knutti 2012 use Average Sea Ice Area for the 3 months August to October, not September SIA values. See graph here. (The implications of this difference on its own is perhaps less than other considerations that I reckon together turn their projection of 2°C global temperature rise required for seasonally ice-free Arctic into 1°C. But I feel a bit more nmmber-crunching is required before explaining furhter.)

Thank you, Chris R & Artful D. I now have Mahlstein and Knutti, 2012 full text. While I have yet to entirely pin it down, I feel I am at the "Houston we have a problem" stage. We talk of SeptSIE reaching 1M sq km by Year X and balk at the IPCC when they say this Year X will be decades away. Mahlstein and Knutti, 2012 used to support this assertion have some very curious "observational" data showing in Fig 2. M&K2012 use 10 year means for observations so any recent change in Ice trend will not feature. But they use the models to establish the Ice trend is linear with Global Surface Temps so that isn't so much of a problem. Where I have a problem is this - Look at the last 17 dots on Fig 2. How can such 10-year averages for Arctic Ice sit in the range 5.8 - 5.0 M sq km for averages using 1980-2006 with 2007 just sub 5 M? My take on it is that this data in M&K2012 is AugSeptOctSIA. They are plotting an Ice-free summer not September while using Area not Extent. Even using Area, you would expect this to involve larger temperature rises. Using annual AugSeptOctSIA v UAH(Arctic Ocean), it hits 1M at 3.1 deg C. Zero takes 3.9 deg C, twice the temp rise when plotting SSIE or SSIV. (Note this is data to 2011 only. R-sq = 0.57 which suggests linearity (I'm told), although not as impressive as SSIV v UAH with an R-sq = 0.74.) Using SeptSIA v UAH, 1M sq km arrives at 2.2 deg C Arctic. With Arctic Amplification that is a rise of less than 1 deg C global. Where this seems to be going for me is that the IPCC use M&K2012 as the one alternative method for calculating that the Year X is likely to be decades away. If M&K2012 are predicting something significantly different to SSIE = 1M sq km, or using significantly different observational data, that alternative evidential support showing "best estimate of 2°C above present" is looking at the wrong Year X. It thus becomes the exact opposite of 'support'.

If a linear regression of the most recent half of PIOMAS SSIV (1996-2012) is projected to zero, it gets there in 2019. But this projection has no physical basis. It is but curve-fitting. But if PIOMAS SSIV is plotted against, say, UAH Arctic Ocean Lower Troposphere Temperatures, then there is a physical basis. The graph here plots SSIV against Annual (Oct-Sept) Ave Temps 1980-2012 & shows zero arriving when the temperature is 1.9 deg C above the 1986-2005 average. Without the means to seek that level of Arctic temperature increase within CMIP5 models, a simplistic alternative would be to take Arctic Amplification from IPCC AR5draft as x2.7. Thus zero ice equates to a 0.7 deg C rise in Global Temps above the 1986-2005 average which IPCC AR5draft table 12.2 sees as occurring 2016-35 (for all RCPs bar RPC8.5). An alternative that ventures into 'curve-fitting' would be to fit a linear regression to the UAH temperatures being used. This yields a best-guess of 2022. A similar exercise for NSIDC SSIE yields a UAH rise of 2.5 deg C, although the data doesn't look entirely linear. A regression 1996-2012 yields 1.9 deg C, the same as PIOMAS. In the literature there is Mahlstein & Knutti 2012 (abstract only) who link SSIE to CMIP5 global temperature. Their figures show presumably NSIDC SSIE data averaged over years that fits their CMIP3&5 model predictions. It would be interesting to read the full paper & see how they reach their 2 deg C global temp rise for ice-free Arctic. Myself, I cannot reproduce their graphs showing observed SSIE v global temp. My version for data 1979-2007 gives 1.2 deg C to SSIE=1M sq km.

The CMIP5 models show decadal variations. If today's higher trend is due to such variation, I see no discussion of it anywhere. Discussion of decadal variation seems not to exist while the likes of Goosse et al 2009 and Holland et al 2008 appear to be discussing annual variation (like 2007 & potentially 2012). Note Goosse et al 2009 say such variability should peak at SSIE=3M sq km, ie in the next few years. Without such discussions, Massonnet et al 2012 isn't robust enough for the IPCC's citation. And we haven't even mentioned PIOMAS SSIV. Okay it is a model output but so is CMIP5. We can extrapolate PIOMAS but what is missing is the physical basis for such projections. If we had them, then a good old rant at the IPCC would be arguable. Without it, we are just curve-fitting.

On face value, the paper's conclusions appear convincing enough to fit the IPCC's citation requirement. Consider Fig 1. The NSIDC SSIE data doesn't look too out of place & adding 2012 would make little difference as these are 5 year means (to 4.6M sq km to 2012). But while the average NSIDC trend also fits snug in Fig 2, Massonnet et al make zero discussion of the evident change in the NSIDC trend. The second half of the SSIE data (trend 190K sq km/decade) no longer fits snug! What then lies behind the recent higher trend? Fig 3 displays SSIE having periods of higher trends in the CMIP5 models, occurring (to 1sd) for SSIE=2.2-4.4M sq km which is where NSIDC data is today.

I did tap in what became a rather long comment on this thread but it seems not to have survived. Maybe it fell victim to captcha or moderation or I was on the wrong thread. So I re-write it (& perhaps more understandably for the re-writing) & post in installments. What actually is the alternative to these 'conservative' IPCC projections for an ice-free Arctic? Note that the IPCC want a 'physical basis' and natural variation accounted for. Is the literature providing such an alternative? I don't know of such. If not, are there appropriate reasons for the IPCC's inclusion of those 'conservative' CMIP5-based projections? Does the quoted literature provide a solid-enough basis? I am inclined to say No. Although each paper needs assessing individually, I will address here only Massonnet et al 2012, which should be a strong candidate being titled "Constraining projections of summer Arctic sea ice"

As the IPCC is but a review of the literature, I thought it would be useful to have links to the references from Chapter 12, section 4.6.1 Boe et al 2009a http://www.atmos.ucla.edu/csrl/publications/Hall/boe_et_al_2009a.pdf Massonnet et al., 2012; http://www.the-cryosphere-discuss.net/6/2931/2012/tcd-6-2931-2012.pdf Stroeve et al., 2012 http://www.agu.org/pubs/crossref/2012/2012GL052676.shtml Wang and Overland, 2009 http://www.pmel.noaa.gov/pubs/outstand/wang3261/wang3261.shtml Wang and Overland, 2012 http://www.agu.org/pubs/crossref/2012/2012GL052868.shtml Zhang, 2010b, http://www.agu.org/pubs/crossref/2010/2010GL044988.shtml Massonnet et al., 2012 http://www.the-cryosphere.net/6/1383/2012/tc-6-1383-2012.pdf (Anonymous Referee #2. http://www.the-cryosphere-discuss.net/6/C1297/2012/tcd-6-C1297-2012.pdf )

P-maker, Re snow cover thickness. I had a memory of a map of snow thicknesses for a couple of months in a paper of some age. I don't remember seeing actual observation data of snow thickness elsewhere so it has stuck in my mind. And hey presto, it is figure 2 in The effect of Eurasian snow cover on regional and global variations. Barnett et al 1989. http://oceanrep.geomar.de/13248/1/Barnett.pdf This may be useful to you in scoping snow cover water content (but I'm yet still unclear as to your objectives).

P-maker, A couple of points. But before that, I am not Lodger. (Perhaps in China or Japan it may sound like I am!) On this site especially, Lodger is someone else entirely. Firstly, is the volume/weight convertion that much of an issue? Average Sea Ice density is something like 0.91 sg. I would have thought Greenland snow quickly tends to ice with an sg of >0.917 as it cools below freezing (0.94 at -200°C so not by much). But then given the accuracy of the data to hand, an sg of 1.0 is probably accurate enough (or 1Gt=1cukm.) Secondly, I am still not entirely sure of your requirements. The one value that perhaps could be your aim may be something along the lines of the Arctic's heat content or annual rise thereof. The talk of ice that has melted but re-frozen on its way "out of the building" so not appearing on the GRACE data could be put in context by considering heat energy. The energy to melt or re-freeze ice is 334 kJ/kg. The GIS gets down to temperatures of no more than -35°C. Even warming of this coldest ice all the way up to freezing/thawing temperature requires less than 73 kJ/kg. So it would take a lot of even the coldest ice to re-freeze significant volumes of melt water. Thus, assuming there are no vast under-ice lakes of melted water accumulating, I would assume that re-frozen melt water probably isn't a major factor in any energy balance calculations. Beyond Sea Ice & Greenland, while there is a steep decline in NH snow cover in early summer, the winter snow cover has not declined so (assuming average depth is unchanged) a year-on-year increase in NH energy through snow melt doesn't mirror the year-on-year loss of Arctic Sea Ice volume which is down throughout the year. Melting permafrost: well that is another matter.

NLPatents, Me and my abacus make it roughly 289 years, 9 months & 21 days. In other words, not many centuries. Whether or not the increased melt is something like a smooth 30Gt per year or not, half of Greenland is a contribution of some 3.5m to SLR. This is far from outlandish.

P-maker, Surely the graph above does exactly what you ask - presenting the "key" acceleration of ice loss, at least for Greenland. The data of the Tedesco et al Figure 8 (also presented in fig 5.19 of the NOAA Arctic Report Card and linked here for easier access - http://www.arctic.noaa.gov/reportcard/images-terrcryo/g-fig5.19.jpg ) when converted from Mass Anomaly to Rate of Change in Mass Anomaly (as the graph in the post above does) provides a measure of the Acceleration of the Mass Anomaly through its slope. Thus the regression line labelled 30 Gt.yr^-2 is a measure of the acceleration in Greenlan'ds ice loss over the decade 2003-2012. There is still a question of whether this acceleration is constant or not. A regression on the data 2003-2010 yields an acceleration of 18.5 Gt.Yr^-2. This isn't a conclusive measure of an increasing acceleration. But I'd suggest it does indicate that an increase in acceleration of some value is occurring. So hold on to your hats.

Bosbas, The usual definitive thesis for the accuracy of PIOMAS is Uncertainty in modeled Arctic sea ice volume. Schweiger et al. (2011) here linked as a 5 Mb PDF. (Normally it is linked to behind a paywall.) It is still not the easiest of reads. There are two difficulties with making PIOMAS accurate. The first is deriving a figure via the model from point thickness measurements. The second is deriving the change in thickness/volume with time. This second difficulty is (as I remembrer it) the big one. It is also the one that will become a lot less difficult when the likes of CryoSat2 start to provide thickness data. So far prelimenary results appear encouraging for an accurate PIOMAS. Other accuracy figures that sometimes appear are the internal accuracy of the PIOMAS model, but these are (as I understand it) independent of the other (external) accuracy assessments.

The PIOMAS anomaly graphed year-on-year 2007-2012 through the end of the melt with timing of minimum shown. 2012 was the latest minimum since 2008, changing the trend toward earlier minimums. Graph here (usually 2 clicks to 'download your attachment').

Just to add a pedantic flavour to the Silly Season, John Mason sets out on thin ice with his "This truth is incontrovertable..." Churchill quote at the head of this post. Churchill was actually in denialist mode when he said it. The context can be read in Hansard where he bemoans the panic caused by the 1916 Zeppelin attacks on London (or anywhere else they found worthy of lobbing the odd bomb at). Indeed, beyond the panic, the main problem would become "this infernal barrage ... a remedy worse than the desease" (Churchill quote from elsewhere). Churchill would in later years, of course, hold an entirely opposite view and argue for a strong Royal Air Force which here in 1916 he was arguing (at some great length) against forming. Yet Churchill even in 1916 was indeed denialist. There was real danger from German air power. By 1918, Gotha bombers were planning a murderous fire blitz on London and Paris, plans that should have given every reason to panic in light of the learning that had been gained by Der Englandgeschwader, plans only cancelled at the last minute when German High Command (thankfully) lost its nerve.

(Part 3 of 3.) A more detailed approach is to consider where the melt will come from. The case that it will be from ice flow and not ice melt is quite convincing. Pfeffer et al 2008 constrain SLR to below 2m by 2100 because the sources of the ice are constrained by geology. (Some missing refs add further constraint, if I didn't just dream them.) Greenland shouldn't be a source of unexpected SLR. Glaciers have recently been reported speeding up 30% (No ref to hand) which fits if you consider the actual ice loss = net ice loss (measured by GRACE) plus snowfall and so hasn't doubled since GRACE measurements started. Also satellite data shows faster glaciers but no doubling. There is also studies showing glaciers slow dwon when the melt really kicks in. So we are left with the possibility of surprise SLR from Antarctica which cannot be dismissed. However, the studies that address the issue (eg Bamber et al 2009 or Hartmut et al 2012) consider it very unlikely. So it's not over, but the fat lady is clearing her throat (and if all these links work & point at the right papers, I will sing also).

(Part 2 now of 3.) But that is not to dismiss H&S2011 out of hand. And definitely not to ignore its main message - If global temperatures are not brought down, SLR will keep on going. There have been very large SLR in the past. MeltWaterPulse-1A 14k years ago averaged 4 metres per century over 300+ years. Half of that SLR may be due to a NH Ice Sheet collapse (abstract only) and happened a second time 8,200 bp but ice sheets of this configuration are thankfully gone now. During the Eemian sea level was higher than today and study impliesthe cause was mainly Greenland melting more, not Antarctica, although SLR remained high (possibly 2m per century or more) right up to the 'highstand'. High Northern insolation was suggested as a cause of the high rate. The linear approach (ie melt rate is linear with temperature) has been used to model future SLR and gives a max SLR of 1+m per century either modelled or correlated with past SLR. These pretty much conform to IPCC with added ice cap melt.

Hi M. Owens. (A reply to you way up thread. I tried uploading it yesterday but it wouldn't go. So I went to the pub. Or was it the other way round. It's still sticking so here is part 1 of 2.) I could just cut & paste references here but there are rather a lot & in all truth, a couple of the ones I had in mind are proving elusive this evening. So I thought a very quick run through the approaches within the references. The IPCC has been reducing its max SLR for 2100 ever since FAR in 1990 but has not got much credibility in doing so. At the other end of the scale is Hansen & Sato 2011. Now I don't have a problem with most of this paper. It is saying SLR is gonna hurt. It is only when it starts saying how quickly that I get all contrary, especially at fig 7. This waving of 5m SLR by 2100 goes back to Hansen 2007. Hansen affirms he cannot prove his doubling model is an improvement on the linear one. He is only "confident that it provides a far better estimate than a linear response for the ice sheet component of sea level rise under BAU forcing." In H&S 2011 he provides GRACE data from Greenland & Antarctica with tentative doubling projection lines. I am unconvinced. Here is the Greenland data graphed up to 2011 & the doublings are not apparent. Also the energy requirements are far too high by the last couple of decades of melt before 2100.

Lennartvdl, I find Levermann et al 2011 that you link to is a frustrating read. My interpretaion is that they are happy with Pfeffer et al 2008 as far as Greenland is concerned and agree a full collapse of the WAIS is not on the cards but argue that a partial collapse affecting the Amundsen Sea sector could happen which could cause 1.5m SLR if the whole sector all went. PIG & Thwaites that Pfeffer constrain by geology also contains ice for 1.5m SLR (It would require a speed-up of discharge from 2km/yr to a 90 year average of 53km/yr that Pfeffer et al dismiss.) Whether Levermann et al question Pfeffer et al on Amundsen or PIG/Thwaits or both, I cannot make out. Levermann et al do, however, consider geology & at least question its influence on ice discharge rate & SLR, unlike Hansen & Sato 2011 whose Section 6 has more the feel of a justification than an investigation.

Some light reading for those interested in the poor performance of Arctic Ice models. "Limitations of a coupled regional climate model in the reproduction of the observed Arctic sea-ice retreat" Dorn et al 2012 (full text) sounds like it should throw at least some light on the matter. (I have yet to read it.) There are also a couple more relevant articles in this weeks new research listing at SkepticalScience but only available as abstracts.