Sunday 18 September 2011

What's up with D'Aleo?

Well known denialist Joseph D'Aleo has recently written a load of specious claptrap at the internet's great repository of specious claptrap; Watt's up With That. Tamino exposes D'Aleo's cherry picking (denialists so love their cherries). That noted, my disagreement is with the overall gist of D'Aleo's post. Which is essentially a boring re-run of a tired denialist meme that Arctic sea-ice loss is due to natural processes in the atmosphere and ocean.

In a recently released pre-print Polyakov et al 2011 note that atmospheric factors play a role in the loss of Arctic sea-ice.
Data collected by the International Arctic Buoy Programme suggest that the thinning of the sea ice in the early to mid-1990s was attributable to an increase in ice area export through Fram Strait. Through that time period, this increased export has been linked to the positive phase of the Arctic Oscillation, which increases the cross-strait gradient in sea level pressure. More recent atmospheric circulation anomalies dominated by a dipole pattern (which is different from the Arctic Oscillation) also seem favorable to increased advection of sea ice towards Fram Strait.
They also note the impact of ocean changes on the Arctic.
Further, oceanographic observations carried out in the Eurasian Basin of the Arctic Ocean suggest that the thermal state of the Arctic Ocean interior has an impact on the Arctic ice cover. Observational and modeling results suggest that gradual warming of intermediate waters of Atlantic origin, the so-called Atlantic Water (AW) of the Arctic Ocean, helped precondition the polar ice cover for the extreme ice loss observed in recent years. Polyakov et al. (2010) argued that, on a time scale of several decades, ice thickness losses due to anomalous ocean heat flux could be comparable to losses due to local atmospheric thermodynamic forcing. Observations in the 2000s documented a new pulse of AW warming....

The patterns of temporal AW temperature changes in the Arctic Ocean’s Eurasian Basin and of MYI coverage are in good agreement (Figure 2), suggesting a plausible role for anomalous oceanic heat in recent changes of the Arctic ice cover.

So we know that the change in sea-ice is due to natural processes in the atmosphere and ocean, as the denialists like to claim - natural changes in atmosphere and ocean are driving the loss of ice.

Actually we know no such thing.

The first thing to note is that Gillett et al demonstrate that the Arctic warming since the 1970s is due to anthropogenic not natural factors.

Johannessen 2008 examines the correlation between annual sea-ice extent and CO2 levels and finds a relationship. He uses sea-ice extent and CO2 data from 1961 to 2007, as a test of the method he splits the period into two, the correlation from the first part is very close to that of the second demonstrating that the method can (subject to cautions outlined in the paper) be used predictively.

Figure 2 of Johannessen 2008

NB - regards the following two paragraphs: See first reply and my second reply - applying correlations to non-stationary series is not sound. For the purposes of my argument it's enough to leave the issue as figure 1, which shows the agreement between CO2 levels and annnual average extent.
When dealing with non-stationary series (series with trends in them),normally one would want to detrend both series before considering correlation. However normally one is concerned with whether the small-time variations are accounted for by the independent variable when assessing its correlation with the (suspected) dependent variable. In this instance we know that regional processes that cannot possibly all be directly related to CO2 are major players in the short timescale. So here Johannessen is quite correct in not detrending. After all what's the point in asking "Does the trend of A explain the trend in B?" Then starting the analysis by removing the trend in A and B - obviously that would be pointless.

PS 3/5/12 - due to recent research I've reconsidered my view of the Johanessen paper and approach and now support it. More here.

Johannessen finds that for the full period, 1961 to 2007 the relationship between CO2 and annual sea-ice extent is 90%, i.e. CO2 explains 90% of the reduction in sea-ice. It is worth noting here that the final three data points, 2005-2007 fall below the trend line, with 2007 substantially below the trend line, implying that either weather or positive feedbacks have a greater role in the losses of 2007. As I've argued previously, it's likely a feedback that's actually an atmospheric change, see here.

So surely one position must be wrong?

How can it be the case that the Arctic sea-ice loss is due both to natural processes like the Arctic Oscillation and ocean changes, and that 90% of the change is explained by human emitted CO2?

As discussed previously, Bitz & Roe in 2004 examined the relationship between initial thickness of sea-ice and the rate of thinning in model studies. They found that the thicker sea-ice thinned more than thinner. In one season sea-ice only grows thermodynamically up to around 2 metres thick. So to grow thicker it has to exist for more than one season, hence thicker ice is older and Bitz & Roe's finding can be translated as "older sea-ice thinned more than younger." The odd thing is that this relationship holds true even for models with a fixed ice layer, i.e. an immobile simulated ice pack.

Bitz & Roe propose that the solution to this apparent problem is in the thermodynamics of sea-ice. By definition it takes years to grow thick robust multi-year sea-ice, whereas the thinner less robust first year sea-ice can grow in only one season. This means that first year sea-ice is more rapidly able to rebound from reductions in its extent, whereas it takes years for multi-year ice to recover.

So, for example, the impact of the high export of sea-ice through the Fram Strait during the '90s could more readily be seen in multi year sea-ice because it was not able to rapidly grow back, what would replace it would have been first year sea-ice. Whereas first year sea-ice exported through the Fram Strait would be rapidly replaced by more first year ice in the following freeze season. Likewise for the other factors impacting sea-ice. This has lead to the loss of multi year sea-ice and the decline of the pack until we find ourselves now with what Maslanik calls "A Younger Thinner Arctic Ice Cover..."

However that doesn't answer the question; why hasn't the sea-ice recovered after a period like 1990? We're now nearly two decades on, ample time for the survival of first year and subsequent generations of sea ice to replenish the 'stocks' of multi year ice. Or are we to accept that the observed loss of sea-ice is due to an unlucky and increasingly improbable succession of natural causes each coincidentally acting against the survival of sea-ice to cause the observed decline?

A more simple explanation is that the ice is not getting the chance to recover because in the background is a factor that is biassing the Arctic system against the survival of sea-ice. This is the increasing level of CO2, and the resultant warming and downwelling infra-red. Random factors such as weather variability (e.g. the 1990s positive mode of the Arctic Oscillation) act against the survival of ice. In a normal regime these would create variations in the ice pack, such as in the 1940s, variations that would be followed by recovery. But in the current regime of human caused CO2 increases the recoveries are not happening as they would without increasing CO2.

The ice is constantly chasing a changing equilibrium, one that is driving the sea-ice extent downwards, and humanity is responsible for this changing equilibrium due to the burning of fossil fuels.

Bitz & Roe, 2004, "A Mechanism for the High Rate of Sea Ice Thinning in the Arctic Ocean."

Gillett et al, 2008, "Attribution of polar warming to human influence."

Johannessen 2008, "Decreasing Arctic Sea Ice Mirrors Increasing CO2 on Decadal Time Scale."

Polyakov et al, 2011, "Recent changes of Arctic ice coverage." Preprint.


Peter said...

Sorry, but Johansson is wrong. Doing a correlation without detrending the series is meaningless. You could use the exact same logic to find a tree in Kew Gardens that's been growing since 1961 and claim that the height of the tree is causing Arctic ice to melt.

"After all what's the point in asking "Does the trend of A explain the trend in B?" Then starting the analysis by removing the trend in A and B - obviously that would be pointless."
This is correct as far as it goes, but misses the main point. The question "Does the trend of A explain the trend in B" is not answerable by correlation analysis.

Chris Reynolds said...

Thanks Peter,

That paper should never have got past peer review. I've just checked using Excel's CORREL function against a set of various functions e.g. log(x) vs x^2 gives 0.84 despite the two functions giving wildly different results. I should have checked before using that paper.

I'll look into whether there is any way of investigating the suggestion of a link suggested in panel a of the figure.

Joe D'Aleo said...

Please go to the original post not the Tamino Clift notes extraction.

I was presenting the admission from NOAA NSIDC in 2007 that the arctic was affected by multi decadal cycles in both oceans and that "Many questions still remain to be answered , but these changes in ocean circulation may be important keys for understanding the loss of ice in the arctic. I showed how they linked to the changes in the Pacific after 1979 and then in a step change after the Atlantic popped warm in 1995. I have a follow up answer to Tamino with more arctic data. Would be glad to email a preliminary version.

Finally I am not a denialist. The arctic like the globe warmed from 1977 to 1998. Global temperatures then flatlined. The arctic continued the decline as the Atlantic warm water plays a key role. See the link to the many media stories from the last warm AMO era about ice loss.


Joe D'Aleo said...

Original post was here.

Joe D'Aleo said...

Verity Jones of Digging in the Clay who authored the chart that sent Tamino into orbit has responded in this post.

Kevin O'Neill said...

Dr. D'Aleo: I was specifically puzzled by your graphic for "PDO+AMO & Arctic Mean Temp & TSI" - where does the TSI data come from? It appears to be the Hoyt 1880 to 1979 data with recent years tacked on. Isn't this the outlier of all TSI reconstructions? What are the correlations when other TSI datasets are used?

The consensus theory of global warming predicts stratospheric cooling. How does your position that warming is due to solar forcing account for stratospheric cooling?

Chris Reynolds said...

Joe D'Aleo,

You will find that the first link here is to your guest posting at Watt's Up With That.

Whether you accept what the evidence shows with regards Arctic and global warming is irrelevant to the question of whether you are in denial about AGW.

In your post here you betray your agenda when you say "I was presenting the admission from NOAA NSIDC..." Were you interested in a reasoned debate you would merely say that NSIDC/NOAA 'stated', you use the word 'admission' to try to bolster the fantasy that people like yourself are having to 'drag' science into accepting your views.

As for oceanic impacts. Given that the Arctic ice-pack is floating, only a gross ignoramus (144 times worse than an ordinary ignoramus) would be surprised that changes in ocean heat fluxes have an impact on the sea-ice. Indeed I've previously touched on ocean warming here in the context of atmospheric changes, which are likely due to sea-ice loss. Nothing in the Polyakov paper I started my main post with was a surprise to me.

As this page shows, once the global warming trend is removed from the AMO its impact is revealed to be small.

Tamino's prompt response is crushing, having read both the Digging In The Clay page, and the one to which it linked, I agree with Tamino.
"Let me sum up: Joe D’Aleo ends his post about Arctic sea ice with a graph labeled “Arctic Region Temperatures” which, it turns out, is based on a very limited set of stations, which his defender admits were cherry-picked and do not represent the “Arctic.”"