Monday 1 September 2014

2014 Melt Season Review

I've decided not to await the actual minimum for 2014 to do a post on the season. Now that virtually all the data is in and I've had an afternoon off to do the bulk of the work, here is my 2014 season summary.

2014 may well turn out to be the highest extent since 2007. A hell of a turn around from 2012, but I've yet to see someone make a convincing case that there is a recovery of sea ice that will stop the decades long decline. All extent and area data is from the Wipneus dataset.

Fig 1. Arctic Ocean Extent.

Laptev has been one of the notable features, on which I have posted previously. Now the ice edge proper, not just a region of broken ice or a polnya, is within the 85th parallel, and Laptev has one of the lowest extents on record.

Fig 2. Laptev Sea extent.

Figure 3. Laptev Sea extent on 30 August for 1979 to 2014.

Beaufort is ahead of 2013 but it is virtually certain to be one of the highest minimum extents in the post 2007 period.

Fig 4. Beaufort Sea extent.

The East Siberian Sea (ESS) continues to exhibit a persistent mass of ice due to a large export of multi-year ice earlier this year. At the time I blogged that this could be a significant factor in the 2014 melt season, later I changed my mind on that, which was wrong. The ESS tongue of ice has kept extent up by around 0.2 to 0.3 million (M) cubic km, enough to have brought 2014 back into the middle of the post 2007 pack.

Fig 5. East Siberian Sea extent.

As of 30 August the Central Arctic remains high compared to 2012 and 2007, but aside from volume (which will be addressed when the PIOMAS data is out), 2014 in the Central Arctic is still clearly a post 2007 year.

Fig 6. Central Arctic extent on 30 August.

Compactness (Area/Extent) is a measure of the compaction of the pack. 2014 has consistently shown high compactness, however this is largely due to the Central Arctic (the largest single region).

Fig 7. Arctic Ocean compactness.

Note the similarity between the 2014 (red) plots in both Arctic Ocean and Central Arctic.

Fig 8. Central Arctic Compactness.

For the rest of the post (and this will be long), I have calculated anomalies using a 1981 to 2010 baseline, the same period as used by NCEP/NCAR to try to use anomalies for NCEP/NCAR to throw light on the events of the 2014 melt season. Arbitrary anomaly baselines are possible in the NCEP/NCAR monthly interface, but not in the daily, and I need daily resolution for this. I'm moderately satisfied with the results and am seriously considering moving all of my sea ice data to a 1981 to 2010 baseline.

Starting with the whole Arctic region (NSIDC domain), I think the graph needs explaining. Anomalies calculated using the same baseline for a set of regions can be used to decompose the roles individual regions play in the total anomaly because the sum of the regional anomalies is equal to the anomaly for all the regions combined. In the following I've used some regions that are aggregations of the normal Cryosphere Today regions I use. CAA is the Canadian Arctic Archipelago, ESS is the East Siberian Sea.

Arctic Ocean: Beaufort, Chukchi, ESS, Laptev, Kara, Barents, Greenland Sea, Central Arctic, CAA.
External Pacific: Okhotsk, Bering
External American continent: Baffin/Newfoundland, Hudson Bay, Gulf of St Lawrence.

And within the Arctic Ocean (Internal to the Arctic Ocean):

Int Pacific: Beaufort, Chukchi.
Int Siberian: ESS, Laptev.
Int Atlantic: Kara, Barents, Greenland Sea.
Int Central: Central Arctic, CAA.

Fig 9. Extent anomalies for the whole Arctic region.

Starting with extent anomalies: From January into March was a period of low overall anomaly (black line - extent lower than the 1981 to 2010 average), this was due to a combination of low extent in the Arctic Ocean and External Pacific sectors, with the north American extent being higher than normal. Looking at temperatures the low extent in the Arctic seems to be due to very warm winter temperatures affecting the Atlantic edge, which will be seen to be the case further down. The external Pacific seas show warm temperatures, notably over Okhotsk.

However higher extent in the north American continental sector (Ext American) is rather harder to explain, except for the possible influence of exceptionally cold conditions to the south of Hudson and Baffin. Closer examination reveals almost all of the positive extent anomaly to be due to the Baffin/Newfoundland region.

Fig 10. Late winter surface air temperature anomalies.

From March there was a period where extent in Ext American was high, while on the Atlantic extent was low, with Arctic Ocean extent being around average. Here in terms of temperature Hudson Bay and Baffin were both experiencing below average temperatures during this period. While the Arctic Ocean and External Pacific Regions were both largely above average.

Fig 11. Early spring surface temperature anomalies.

Referring again to figure 9 it can be seen that from mid June there was a substantial change in behaviour as anomalies of extent dropped sharply indicating far more extent loss than normal for the 1981 to 2010 average. This event is analysed in the following two graphics.

Figure 9 from mid June to early July shows a sharp drop in extent anomalies. This is composed of greater than average loss in the Arctic Ocean with a sharply greater than average extent loss in the American sector.

Figure 12 below demonstrates that the Arctic losses were predominantly due to a longer term process of above average loss in the Siberian sector, with Central Arctic and Pacific extents within the Arctic Ocean dropping less severely.

Fig 12. Arctic Ocean extent anomaly decomposed into four regions.

Turning then to figure 13 below, it is seen the the American sector extent crash was due to earlier than normal loss of ice, this was due to a warm band of surface temperatures some 3 to 7 degC above normal playing a role in early melt out of Hudson Bay. However Baffin and Newfoundland engaged to an equal degree with this loss of ice, this may have been due to compaction caused by southerly winds, NCEP/NCAR vector wind shows moderately strong winds heading north up Baffin Bay during this period. Hudson Bay shows northerly winds over this period, possibly leading to the southward bias of ice distribution in Hudson Bay at this time (e.g. Bremen) and compaction reducing extent.

Fig 13. American continental sea ice extent anomaly decomposed into three regions.

However as shown in figure 9 there was also a large component of above average loss in the Arctic Ocean, and as shown in figure 12 a major factor here was the Siberian sector. Further decomposition of this data (fig 14, below) shows that by far the greatest component in this was the Laptev Sea, where the polnya that had opened up earlier in the year began to grow substantially.

Fig 14. Siberian Sector extent anomalies decomposed into two regions.

The biggest feature of the summer, seen in figure 12 is the levelling of anomalies after around 12 July to a much shallower rate of descent. Whole Arctic anomalies of extent typically show an anomaly levelling in July, but this is due to the end of early melt in Hudson and Baffin resulting in open water, neither of which would be a factor in figure 12. Indeed, in recent years, it is more typical for anomaly rates losses to continue within the Arctic Ocean.

Looking at sea level pressure (SLP) from NCEP/NCAR the period 22 May to 10 July is taken to be roughly representative of the period of large anomaly falls during summer 2014, the period 10 July to 29 July (the most recent available data) is taken to represent the period of more modest extent anomaly drops within the Arctic Ocean.

Fig 15. Average SLP during the period of strong extent anomaly drops in the Arctic Ocean.

Fig 16. Average SLP during the period of modest extent anomaly drops in the Arctic Ocean.

The two periods have markedly different atmospheric set ups. During the period of strong drops in extent anomaly there is a strong high pressure tendency with clear skies (MODIS at the time) under the period of maximum insolation around the solstice. However by the second period there is a strong low pressure tendency over the Kara Sea, with a weakened high over mush of the pack, indeed cyclonic systems were quite common in the late summer.

As the increase of PIOMAS volume this year has shown, and as overall temperature for the summer will show when the NCEP/NCAR monthly data is up, 2014 has been a weird year with weather that has not been conducive to melting ice. For me it has been a boring year. If the weather during the period of rapid extent drops from mid June to early July persisted we could have seen a more interesting year. However the multi-year export into Beaufort, the Chukchi and the ESS had always meant that the best that could have happened with 2014 was a place around or above 2007, more in the middle of the post 2007 pack. Looking at extent anomalies shows the impact of weather, as seen in the figures that follow.

The picture in terms of area is more complex and very variable, but a steep drop in anomalies purely due to area losses in the Arctic Ocean is clear after around 16 August, the final two weeks of August (fig 17, below).

Fig 17. Whole Arctic area anomalies, regional decomposition.

Looking at the regional breakdown within the Arctic Ocean (fig 18 below) shows that a major factor was area loss from within the Central Arctic.

Fig 18. Arctic Ocean area anomalies, regional decomposition.

And zooming into the Central Arctic sector, comprising of the Central region, and the CAA, we see that almost all the loss from the Central Arctic sector was from the Central Arctic region itself.

Fig 19. Central Arctic sector area anomalies, regional decomposition.

So after 16 August there was a sharp drop in anomalies. Here's what the pressure pattern for 16 to 29 August looks like. Note the dipole between the Canadian and Eurasian halves of the Arctic.

And here is what it was doing to temperatures aloft (850mb level). Note the tongue of warm air from Alaska into the pack.
A strong dipole dragging warm air over the pack and dropping the area, but too little, too late.

I'd like to say that there is always next year. But I think the massive volume gain within the Central Arctic region will have to work through the system before we see another 2007 or 2012. That may take some years.


iceman said...

The regional decompositions are instructive, and thanks in particular for highlighting the bump in compactness in the Central Arctic. This is one of the main puzzles of the melt season. I suppose it is related to the moderate pressure gradients. What about the NAO, per John Christensen's comment on ASIF in response to your blog post? Is the temperature inversion and attendant fog, which others have noted, part of the same causal chain?

While the main tent is getting ready to pull up stakes, there remain some entertaining and maybe noteworthy side shows. The CAA seems to have stalled recently but is likely to have a few more substantial drops. And I'm wondering whether that chunk of thicker ice in the southern ESS will detach with shifting pressure systems over the next several days.

Chris Reynolds said...

Sorry Iceman, but the compactness issue remains a puzzle to me. Colder temperatures likely play a role, and as figs 7 & 8 show, almost all the action in Arctic ocean compactness was due to the Central Arctic region.

I'm glad you found the anomaly decompositions useful, I did too. I can't believe I didn't use this method earlier. I'll have another look at that metric over the weekend. My earlier suspicion (around June/ early July) was that it was the central placing of the high pressure that played a role through keeping cool air in place and blocking out warm intrusions. But I need to have another look at that idea.

Chris Reynolds said...

I'll address the NAO over at Neven's since the comment was posted there.

Giovanni Ciriani said...

Hello Chris, very thorough analysis. Thanks for posting! I'm looking for JAXA's regional extent numbers, to run some correlation. Do you know where they can be downloaded from?

Chris Reynolds said...

Hi Giovanni,

I don't really follow JAXA, but don't think they have regional numbers.

There are a few of us connected to the Sea Ice Forum and Neven's Sea Ice blog who work with gridded data. I've done regional breakdowns of PIOMAS volume, Wipneus has done the work on NSIDC concentration to produce regional area and extent. Both of us use the Cryosphere Today regions, not the MASIE regions.

You can find the Wipneus dataset linked to in the first paragraph of this post. My breakdown of PIOMAS volume is available from the top right-side panel under Sea Ice Data, it's regional PIOMAS Data.

The Wipneus area and extent data is in the files:

'final' is the finalised data, covering most of the series from 1979, it's big so it's zipped. 'nrt' is the provisional data, which currently runs from 1 Jan 2014 to the most recent date Wipneus has processed.

Neven's Blog.
Sea Ice Forum.