Tuesday 6 January 2015

PIOMAS December 2014

The PIOMAS data for December 2014 is now out. So here is the usual summary. I start with the usual map plots of thickness, this time I've done an animated gif for September to December thickness.





As is to be expected from my previous summaries, the bulk of the volume increase on past recent years is still from the Central Arctic region, for region definitions see Cryosphere Today's map.


Note that as usual the Central Arctic volume, being so much larger, is shown separately on the right side axis, and is four times the scale of the left axis. Barents and Kara also show increased volume from 2012, but this is small compared to the Central Arctic.

The following graph shows the difference between December 2014 and December 2012, again the volume gain in the Central Arctic is massive, from Beaufort to Laptev the picture is more mixed and arguably they haven't improved much on 2012 (following the record crash of that year).


This Autumn (Oct - Dec) has seen a strong transpolar drift (ring of arrows and blue colouring from lower left quadrant to the pole), which should have enhanced open water formation and freezing from the Laptev Sea, adding to volume gain. While the mean wind also exhibits wind along the track of the transpolar drift I show the plot of anomaly to demonstrate that this is more than usual for the 1981 to 2010 climatology.


More recently, the mean vector wind for December from NCEP/NCAR shows the above noted transpolar drift, however it also shows the emergence of a brisk wind causing Fram Strait export (yellow mass in upper right quadrant). This is the cause of the emergence of Fram Strait export noted by Wipneus (Hat tip to Neven).



Aside from PIOMAS most days I check out HYCOM Global ice motion from the CICE thickness 30 day gif. It's a very useful real time resource for assessing factors such as those noted above, which can be seen in their most recent animated gif (if I link to their site I will be out of date in a few months!).


From the above it may be suspected that there is a significant export into Beaufort and Chukchi, I'll be blogging on that shortly.

Over last winter I maintained a policy of using Beaufort, the East Siberian Sea (ESS), and the Central Arctic as bellwethers of Arctic Ocean state, this does not seem so applicable this year as we have such a difference between the Central Arctic region and the peripheral seas (Beaufort, Chukchi, ESS, Laptev, Kara, Barents, Greenland and the Canadian Arctic Archipelago).

So looking at the whole PIOMAS domain: In terms of the timeseries of volume due to grid boxes reporting thickness in five thickness bands, between 0 and 3m there is little difference from other post 2010 years. These lower thickness are dominated towards the thin end by volume from the regions outside of the Arctic Ocean, with expanse of December ice of under 1m thick, indeed that volume has changed little since 1978. There may be a change in contribution from within and outside the Arctic Ocean holding that volume up, as may be suspected when I come to autumn ice growth later in this post.


Ice volume from grid boxes reporting 1 to 2m appears to be levelling, over the recent past this is also suggested by 2m to 2.9m thick ice. The impact of 2007 on ice 2m to 2.9m thick was very severe, this is because ice over 2m thick in December is typically multi-year ice, however after 2010 the 2m to 2.9m category suggests a bounce back from that year's losses. Turning to thicker ice there is seen to be a marked increase in ice over 3m thick in December 2014, this accounts for most of the increase over 2012's minimum.

However longer term the increase is seen to be modest, despite that it has taken us back to volumes typical of around 2006, using the gridded data, this December is 15.07k km^3, 2005 was 15.04k km^3, 2005 was 15.98k km^3. However as can be seen from the 2012 to 2014 difference plotted above, almost all of the increased volume is limited to the Central Arctic region.

The pentad (5 year) breakdown of volume as function of grid box thickness shows the same monomodal spike as in 2013 and to a lesser degree in the average of 2008 to 2012 (averaging will round it). This spike at around 1m in December is symptomatic of open water freezing over to form first year ice which then thickens thermodynamically. With September 2014 seeing the fifth lowest extent (NSIDC) dominance of thin first year ice is still a major factor this winter.


However, what December 2014 exhibits and 2013 (also 2010 to 2012) lacks is the presence of a spike at 4m, that being due to the volume of ice that is 4m thick or greater. Current export patterns through Fram and into Beaufort/Chukchi suggest to me that this may not last long (for years), I'll deal with that in a separate post in due course.

This monomodal spike behaviour at around 1m thick in December is clearly seen in the post 2020 volume distributions with thickness. 2012, following a record smashing minimum shows this most clearly. What is significant here is the closeness of the peaks, at around 1 to 1.2m for all years 2011 to 2014, although 2013 is rather less pronounced. This suggests that, at least in PIOMAS, thickening has proceeded for these years rather independently of the variable weather over those years. The reason for this is that once the sun has set and temperatures drop, it is ice thickness that is the dominant factor in determining heat flux through the ice, and the growth rate of sea ice.


Turning now to the Central Arctic, again the impact of 2007 is seen in ice 2m to 2.9m (light green). While ice has thinned into thinner categories, notably the 1 to 1.9m thick band. Again it can be seen that the majority of the volume gain over 2012 (1500km^3) is in the Central Arctic in ice between 3 and 4m thick.

In my August post I calculated that 78% of the volume increase from 2012 was in the band 2m to 3.9m thick, within the Central Arctic. Revisiting that calculation:

Arctic sea ice volume from PIOMAS is now 2940km^3 above that in December 2012. 32% of this increase is from the Arctic Ocean excluding the Central Arctic, of that 19% is from the Kara and Barents Seas alone. The Central Arctic accounts for 61% of the volume gained between December 2012 and 2014, 93% of the increase in volume is due to ice thicker than 3m thick, with the implausibly small figure of 7% for the ice thinner than 3m being caused by the thinner ice in 2012 causing a negative difference. i.e. in 2012 there was thinner ice that isn't there this year.

So, relative to 2012, over half of the increase in volume this December is due to ice over 3m thick within the Central Arctic.


This mass of ice has now thickened, the strong appearance in the 2 to 2.9m thick band has now thickened and spread into volume from grid cells 3m and thicker.


So what of the process of thickening from September to December?

I have produced the following table of autumn volume gain within the Arctic Ocean, and autumn volume gain, outside of the Arctic Ocean. As a percent of total volume gain there is a clear decline outside the Arctic Ocean, while within the Arctic Ocean there is a clear increase. This is due to actual volume gain decreases and increases in line with the percentage changes.


How does one explain this clear difference? In the external seas the sea ice is already seasonal and cannot benefit from the growth thickness feedback, instead it is subjected to warming temperatures with anthropogenic global warming. Whereas within the Arctic Ocean once multi year ice dominated the increased growth with thinner ice and open water at the end of the summer is able to manifest.  So does what is going on in the Arctic support this interpretation?

I will start off by looking at thickening by regional seas. From this it is clear that the peripheral seas have seen little change in Autumn (Dec minus Sept) thickening, the main player in increased autumn volume gain being the Central Arctic. This is particularly evident after 2010, a year in whose winter there was a massive loss of thick ice in the PIOMAS model (and, I argue, in the real world). The cause of this being the increase in growth/thickness feedback caused by the loss of volume.


Breaking the data down into the peripheral seas within the Arctic Ocean and the Central Arctic, it can be seen that September to December (autumn) volume gains have been strongest within the Central Arctic, with relatively little change in the peripheral seas.


The Central Arctic, with its thicker ice had less potential for thermodynamic thickening, but as it has thinned larger volumes have transitioned to a thinner ice state with more potential for thermodynamic thickening. The peripheral seas seem level, there is a small upward trend, but the R2 is only 0.07 due to the noise in the series.

Using the following seas:

1 Barents
2 Greenland
3 Kara
4 Chukchi
5 Laptev
6 Beaufort
7 ESS
8 CAA

Numbered as in the following graph, Excel has a bug where it fails to show x axis labels with two y axes. A scatter plot shows the same general relationship, but it is rather messy. So I have used the 1980s average and 2010 to 2014 average of both thickness and autumn volume gain to calculate the difference between the 1980s and 2010 to 2014. This is then plotted on the following graph.


Despite the small number of data points the slopes of these lines suggest the same relationship between thinning and increase in autumn volume gain. As the ice thins in the peripheral seas the autumn volume gain goes up. The negative volume gains are for the Barents, Greenland, and Kara seas, which have been subject to warming Atlantic Ocean waters over the period between the 1980s and recent years.

The Arctic Ocean seems to be behaving as expected in PIOMAS, the ice is thickening and the volume gain of the last two years is being retained. Whether Fram export will seriously eat into this remains to be seen. By March/April we will see what ice state will be passed on to the 2015 melt season. My tentative opinion at this stage is that we will see another low ice state next year, but that beating a year like 2007 or 2011 looks unlikely and a new record below 2012 is only likely with exceptionally good melt weather.

Over my Christmas/New Year holidays I've decided to switch from using a 1980 to 1999 baseline for anomalies to matching the NCEP/NCAR baseline of 1981 to 2010. This proved a useful technique last year, as in this post, and I anticipate it to be valuable in the 2015 melt season. I will also be making use of the greater regions outlined in that post (about half way down).

No comments: