Does the AO index of 2020 predict record average lows for 2021?

[binntho]

This poll is basically a result of my musings on the "When will the Arctic Go Ice Free?" thread, starting with this short post on the recurring "phase change" and "hiatus" discussions.

A recurrent theme of the "hiatus/phase change" brigade is this graph from the wizard of statistics, Tamino himself, covering the years 1979-2016 (october to september each year, not calender years). Tamino found a statiscally significant change in the rate of decline as shown, but remember that "statistically significant does not equate truth" as Pontius Pilate might have said.



The two main elements of the "phase change/hiatus" appear in this graph, i.e. the fast decline 2003-2007 and then what is essentially a flat line, with no signifant rate of decline.

But what happens if we add another 4.5 years to the plot? Below is my own version, this time using calender years (which changes the shape of the top left from Tamino's graph. The year 1987 does have 200.000 km2 more in Tamino's calender year graph than in mine, for which I have no explanation). I can just about manage a LOWESS smooth, piecewise-linear is way beyond me.

The rapid decline 2003-2007 is now easily matched by 2013-2016, the stepwise nature of the graph is much less obvious and the rate of decline following 2007 is now faster than that before 2003. In other words, both conclusions from Tamino's graph, 1) a period of fast decline, 2) followed by a hiatus, seem to have vanished.

Which is not to say that there is not a stepwise change in the middle,  a noisy dataset can easily create artifacts that vanish as the timeseries expands.

[binntho]

There is another aspect of the graphs we so often see, and it is the strangely quiet period before the rapid decline Tamino found. Others have pointed out the inherent unlikelihood of 9 successive datapoints landing above the trendline, but even more interesting is the lack of variation between these 9 "quiet" years.

Is this the anomaly that has confused so many people and caused them to see the rapid decline/hiatus and start searching for phase changes and whatnot?

After having marked the "quiet anomaly" in the graph below, I noticed the fairly regular downward "punches" shown with red lines in the second graph, the years between in red. It looks as if the "quiet period" missed it's punch?

[binntho]

My thought at this point was that the punches were caused by interlocking climate cycles. The most obvious place to start looking was the AO index, and I must admit that I was quite surprised to find both the quiet period and the "punches" here.

The match is not perfect - but the "quiet period" is clearly visible and most of the "punches" match unusually high average AO indices from the year before.

My hypothesis is that the AO index is one of the climatic cycles that can cause a "punch", and thus that a high AO index of the year before is a strong predictor of a "punch" the following year. Hence the year 2021 marked in the lowest graph, showing the AO index of 2020.

So this poll is basically about this hypothesis and the prediction made on the basis of the unusually high AO index of 2020: I predict that 2021 will be the lowest on average during the satellite era, both in extent and volume.

[Sepp]

I do quite like the theory, unfortunatly, the number of data points won’t give any statistically significant result.

However my main objection is another one: From my perspective there are too many year before the minimum year, which are already quite below the trendline, but do no show an AO peek (in their previous year). So this could lead to the assumption, that a AO is needed for a rebound two years later. This conclusion sounds strange.

This is written from a data point of view only, without any deep understanding of AO.

P.S. I am actually more curious, what the reasons for the rebounds after the minima are, but this belongs to another thread.

[Glen Koehler]

    1) Extent has the "hemmed in by the edge of the basin" constraint in the High Arctic, and it is further muddled by variable thickness and concentration, especially around the edges that can vary from a thin veneer of slush to thick MYI, yet they all count as equal values in an Extent cell grid.  So I don't put much faith in Extent as a monitor of ASI status.  It would be interesting to see the hypothesis presented as an annual average VOLUME vs. AO graph.  Or at least AO vs. AREA instead of EXTENT.

     2) At risk of inventing a pattern out of noise, the quiet years ended in 2007 with a bigger punch than the other punch years (i.e. largest drop below trend line).  That makes me wonder if some sort of unresolved melt tension built up over the quiet years that was then released by AO or some other trigger.  Perhaps that is consistent with the lack of an AO uptick during the quiet years preventing release of "melt tension" (if there is such a thing).

     3) A change in AO that includes reversal of the polar jets, or otherwise affects surface wind patterns through polar cell displacement or spltting, could change Extent even without changing the amount of ice.  That could happen by cyclonic winds to compress the ice pack in one year vs. anti-cyclonic winds to disperse the pack the next.  So if there is an AO effect, it may be moving the ice around without affecting the amount of ice as much.  That would allow a consistent rebound effect after a temporary AO-induced punch to compress the ice pack in one year, but not eliminate it, then relax that compression the next year.

     4) Does AO have any consistent effect on Trans Polar Drift rate or direction?  Or the so-called Beaufort Gyre?  A change in Fram export or Gyre pattern would be a way for AO to affect the amount and distribution of ASI. 

     5) An AO effect on wind speed could also affect the ice through a thermal effect from Ekman pumping or shoaling.  The additional heat expenditure from stirring the pot could lead to cooler conditions in the following years, creating a post-AO punch rebound pattern.  Or AO induced wind change could move ice into or out of atypically hostile/supportive locations for ice melt.

     6) Even without a temperature or wind/drift effect, a pressure-based clear-sky pattern associated with AO could affect early-season melt pond momentum.

[nadir]

Binntho,

I like your observation, but to make it a theory we need a mechanistic explanation behind it. So I have these questions, for you and for anybody else (some wonderings are similar to Glen’s above)

- Are you using average AO yearly? Is it more preponderantly high at a particular time of the year?

- what does a higher AO for the previous year imply for the next to receive a punch? Does it set up a very mobile Beaufort ice and high Fram Export? Is it setting up an atmospheric pattern of strong Arctic dipole and/or strong Greenland blocking? Does it mean lower snow volume in the Winter in between at mid-high latitudes?

- Do you think this correlation has a cause behind both the high AO one year and ice loss the next one?

[bbr2315]

I think that we may see both a rebound and a disaster which may sound contradictory, but let me explain.

We are now in the midst of a major snowy period in the Northeast US and I do believe our weather in North America is quite connected to the Arctic. Our last truly cold and snowy February and March periods were 2014 and 2015, since then they have been fairly milquetoast although March-April 2018 was also VERY cold and snowy.

I think the connects to the low-latitude sea ice in Bering, Okhotsk, and Hudson Bay, and portends a near-record advance in these regions this year, and a very late melt. This is why this year's #s aren't abysmal as well, as Okhotsk is EXCEEDINGLY high right now and will continue to gain ice for a while.

While we have yet to see snow-water-content truly explode across the continents, North America is now showing signs that this may be underway. The Great Lakes are not yet frozen, but they may still freeze if short-long range forecasts of the TPV anchoring in the vicinity come to fruition.

If the TPV does anchor in the Great Lakes, we are likely to see +SWE anomalies in North America. Eurasia may see a similar pattern as we head into spring. The major -500MB anomalies across the relatively low-latitudes of the continents will also act to push major cold pulses into the mid-latitude oceans, which are WARM.

The end result of this should be advection of oceanic heat into the Arctic on a significant scale this springtime, I would think we will see major +500MB anomalies across the sectors of the Arctic fronting onto the oceans (Bering, Barents, maybe Baffin?) and we may also see the characteristic +500MB anomalies of late across northeastern Russia.

This could have two effects, both correlating to a severely negative Arctic Oscillation (IMO).

1) Cold and snow will remain entrenched across the continents, especially North America, into April, and possibly May and June. This is likely to PRESERVE the ice in Hudson Bay, and probably Okhotsk as well. Not sure what will happen with Baffin.

2) We currently have horrible extent / thickness in Barents and the CAB peripheral to Barents. This may improve a bit through March, however, the melt season is likely to start very early and with much gusto along the Atlantic front, in my opinion. Kara and Barents could go to 0 fairly quickly, and after that, it is the CAB that is now the ATL front....

While on the surface this may yield area and extent numbers that are "normal" the distribution of the sea ice this spring and early summer is likely to be quite lopsided, with anomalous extent remaining in areas it will melt anyways, and negative anomalies developing in regions that formerly retained sea ice through summertime. The sum of this equilibrium is most definitely a weather pattern that is NOT in equilibrium with recent human history, and could very possibly result in major heat accumulation in the Highest Arctic, i.e., Atlantic influx all the way to the heart of the CAB.

This would explain why the North Pole went blue or nearly so in 2014, despite its apparent recovery re: extent.

[Tor Bejnar]

I voted "Yes" mostly because the trend is down and partly because of Binntho's intriguing hypotheses.

A couple questions arise for me:

• What effect does an AO have on retention or loss of Arctic fresh water?
• What effect does a low/high ice NH autumn have on Arctic fresh water retention?

[Comradez]

NH snow extent is significantly below normal heading into the spring snow melt season, with significant albedo feedback possible, so I am tempted to vote yes, if only for that reason alone...

[HapHazard]

I find the upward punches more interesting than the downward ones, honestly. But really it's just pulling patterns out of noise. Revisit this in a few centuries, maybe.

[interstitial]

It seems to me the noise level on extent based on wind and weather makes predictions similar to reading tea leaves or casting wands. I accept the general downward trend but much beyond that seems contrived. Average annual extent has more value IMO.


I think Volume is a more reliable measure with more a stronger signal in the signal to noise ratio. Unfortunately the only volume numbers I am aware of are Piomas. So far the most significant observation that I am aware from Polarstern was that Piomas does not accurately represent current arctic conditions. For all the argument defending Piomas Hycom seemed to be much more accurate.

[binntho]

I'll post a monthly update on where we are regarding my hypothesis. So far we have the January numbers, and 2021 is in 6th place. Does not bode well ...

[binntho]

I do quite like the theory, unfortunatly, the number of data points won’t give any statistically significant result.

However my main objection is another one: From my perspective there are too many year before the minimum year, which are already quite below the trendline, but do no show an AO peek (in their previous year). So this could lead to the assumption, that a AO is needed for a rebound two years later. This conclusion sounds strange.

This is written from a data point of view only, without any deep understanding of AO.

P.S. I am actually more curious, what the reasons for the rebounds after the minima are, but this belongs to another thread.

You raise an interesting point which is that there are so many years before the "quiet" period that are below the trendline. So what happens if we plot the deviations from the trend line?

The intriguing fact that emerges is that it is the "quiet period" that seems to pull the trendline upwards! The missing punch (or two? - the gap between the 1995 and 2017 punches is 12 years) seem to have kept the ice above the long-term trend than it would otherwise have followed, perhaps further explaining the seeming precipice once that period ended.

[binntho]

    1) Extent has the "hemmed in by the edge of the basin" constraint in the High Arctic, and it is further muddled by variable thickness and concentration, especially around the edges that can vary from a thin veneer of slush to thick MYI, yet they all count as equal values in an Extent cell grid.  So I don't put much faith in Extent as a monitor of ASI status.  It would be interesting to see the hypothesis presented as an annual average VOLUME vs. AO graph.  Or at least AO vs. AREA instead of EXTENT.

You are right about extent - but it was the measure used by Tamino and recently referred to by yourself. I did volume too and the conclusion seems to be the same. There is less year-to-year variation in volume than there is in extent, but what still stares out from the plot is the "quiet period" in the middle, admittedly slightly disrupted in 2000, but as you can see from the second graph, the "quiet period" has an amazing 10 points above the trend line.

However, it is the very deep dip in 1981 and 1982 that pulls the trendline down and skews the picture somewhat. This dip is not a mistake by me (as would be all too likely) but can be verified on the PIOMAS website.

[binntho]

I've added punchlines to the same years as for extent. Again the correlation between peaks in AO index and lows in average volume is quite good, although not perfect.

[binntho]

I find the upward punches more interesting than the downward ones, honestly. But really it's just pulling patterns out of noise. Revisit this in a few centuries, maybe.

Be my guest! I'll probably not be around by then. But until such a distant future, playing around with the numbers we have is the best we can do.

[oren]

I have somewhat belatedly approved posts by nadir and bbr, see above.

[binntho]

Binntho,

I like your observation, but to make it a theory we need a mechanistic explanation behind it. So I have these questions, for you and for anybody else (some wonderings are similar to Glen’s above)

- Are you using average AO yearly? Is it more preponderantly high at a particular time of the year?

- what does a higher AO for the previous year imply for the next to receive a punch? Does it set up a very mobile Beaufort ice and high Fram Export? Is it setting up an atmospheric pattern of strong Arctic dipole and/or strong Greenland blocking? Does it mean lower snow volume in the Winter in between at mid-high latitudes?

- Do you think this correlation has a cause behind both the high AO one year and ice loss the next one?

I'm not sure if I am even close to calling this a hypothesis, let alone a theory. But as for your questions, I average the AO index over the calendar year. I did look at using different timeframes (summer-to-summer, 18 months average etc.) but it didn't really make any difference.

Through the years I think we have learned that clear weather with lots of sunshine during the melting season, followed by a one or more hefty storms towards the end, are what has the biggest melting effect. Perhaps moving "downslope" from high AO to low AO makes these circumstances more likely?

[Glen Koehler]

     This is about long-term trends, not a decade long "quiet period" and change in interannual variability due to AO.  But it does point to AO as an important influence on ASI.

Accelerated decline of summer Arctic sea ice during 1850–2017 and the amplified Arctic warming during the recent decades
https://iopscience.iop.org/article/10.1088/1748-9326/abdb5f#erlabdb5fs5

Abstract
      "The 168 year trends of summer (July–September) sea ice area (SIA) variations in six Arctic regions during 1850–2017 are analyzed. SIA has been significantly decreasing in most Arctic regions since 1850.  The rate of retreat for the period of 1948–2017 accelerated multi-fold.  For the nearly four decades since 1979, most Arctic regions are experiencing the highest reduction rate. 
     Besides the increasing surface air temperature, the key drivers to the accelerated summer Arctic sea ice decline are found to be the combined global warming and the regional Arctic warming exerted simultaneously by the Arctic Oscillation, North Atlantic Oscillation, Atlantic Multidecadal Oscillation and Pacific Decadal Oscillation during the last several decades.  The dynamical and thermodynamical warming, driven by the internal variability of the teleconnection patterns, occurred in the last several decades, in particular on the multidecadal timescales.  This leads to Arctic amplification that accelerates the positive ice/ocean albedo feedback loop, resulting in accelerating summer sea ice decline."

[binntho]

Accelerated decline of summer Arctic sea ice during 1850–2017 and the amplified Arctic warming during the recent decades
https://iopscience.iop.org/article/10.1088/1748-9326/abdb5f#erlabdb5fs5

Abstract
 ...  the regional Arctic warming exerted simultaneously by the Arctic Oscillation, North Atlantic Oscillation, Atlantic Multidecadal Oscillation and Pacific Decadal Oscillation during the last several decades.  ...

So they are saying that theses oscillations are causing Arctic warming in and of themselves. How? Ill have a gander later, see if the answer is in the arcticle.

EDIT: The article is quite heavy going but looking at the bottom line of table 2, the (summer) AO seems to have the closest correlation with summer SIE of all the different influencers. My hypothesis is that the AO causes the strongest cyclical effect, but I looked at the average AO of the previous year.

Just below figure 4 they seem to be saying that the DA has a much stronger relationship with natural fluctuations in SIA than the AO.

Anyway, as for my question, in the conclusion:

This analysis suggests that the accelerated decline of SIA in recent decades was not uniquely driven by a single factor, but by a combination of global warming and internal variability of the climate system, particularly on the multidecadal timescales, which contributed to Arctic temperature amplification and anomalous atmospheric and oceanic circulation patterns over the pan-Arctic.

So they are claiming that the internal variability of the system drives the accelerated decline. But I don't really see how they can reach that conclusion. That there is correlation between the various cycles and SIE and SIA cannot be considered "drivers of accelerated decline", and even if the various indices have been increasing over time (i.e. the AO and the NOA), this is most probably due to a global warming, which is itself accelerating.

Which leaves us with global warming as the only driver of acceleration, but this warming expresses itself both in higher temperatures in the Arctic and in increasing indices of the various cycles, so that as the melt accelerates, the correlation between the cycles and SIE increases - but this does not imply causation!

In my (never really very humble) opinion the authors confuse correlation with causation!

[Killian]

You are all much more polite about this than you were about my El Nino/ASIE theory.

 

I  suggest overlaying the EN's over the AO and extent and see what you get. ICYMI, my theory was that an EN has a negative impact on ASI within two summers of the EN, which would usually equate to between a few months and something less than 18 months.