Wednesday, 6 May 2015

April 2015 Status. Part 2.

Part 2 of my summary of April sea ice state focusses on ice thickness as seen in PIOMAS. So for those who have been awaiting the more detailed thickness distribution plots by region here is the detail that may help form opinions as to prospects for the coming melt season.

I have used PIOMAS gice data in this post, gice is explained here. However the PIOMAS gice data is available in two sets. Monthly gice data is available from January 1979 to December 2014, but not for January to April of 2015. Daily PIOMAS gice data is available from 1 January 2000 to 30 April 2015. The problem is that when trying to calculate monthly average gice from daily gice data I was getting significant errors. After I had satisfied myself that my code was fine I emailed Dr Zhang and discussed it with him, it turned out that daily PIOMAS data is from the last timeslot of the day, while monthly is the average of all 75 timeslots in each day for all days in the month. This was news to me, but such detail of the inner workings of the model are not the sort of material relevant to the papers describing the internal physics of the model.

As a result I have a long term time series, that doesn't cover this year, and a short term series that does cover this year. I will be using the long term monthly data in upcoming posts, but for the purposes of current ice state as this year unfolds I will use the daily data. For this post all gice data is for day 120, which is 30 April.

That background aside, here are what I see as key regions of the Arctic Ocean, Beaufort Sea, East Siberian Sea, and the Central Arctic, preceded by the whole PIOMAS domain. For reference here is the plot of April 2015 grid box effective thickness.

Note the structure in the upper left quarter of that plot, headed towards Wrangel Island, a ridge of thicker ice through the Chuckchi Sea. That is indicative of the intrusion of multi year ice exported over the winter from the Central Arctic. Compare its position with the following plot from ASCAT, which is a satellite radar system, not a model.

That export has happened, what role might it play in the coming melt season?

Whole PIOMAS Domain.

The thickness distribution of overall volume gives the first indication of what the rest of the data will reveal. A pack of two distinct states.

The 'monomodal peak' is less pronounced in April than earlier in the winter, being 'smeared' by weather and processes like sea ice ridging. However the peak around 2m thick is less pronounced then in 2013, at 19.2k km^3 April 2015 volume for ice below 3m thick is 1.1k km^3 below the average volume of ice under 3m thick for 2007 to 2014, being the second lowest volume for such thickness of ice in the post 2007 period. In contrast volume above 3m thick is 5.0k km^3, roughly 1.6k km^3 above the volume average for 2007 to 2014 for ice over 3m thick.

Looking at the gice distribution for the entire PIOMAS domain, the pattern is repeated, with volume higher for higher thickness, and lower for thinner thickness categories. The difference here is far less marked than for the grid box effective thickness. Grid box effective thickness is the thickness weighted integral of the gice thickness, variation in the gice distribution for a given grid box will have a greater effect on the grid box effective thickness for that grid box than it will for the sum of volume for the entire PIOMAS domain broken down into gice thickness.

The Arctic Ocean is the region in which the minimum is set, so for completeness here is the overall distribution for the Arctic Ocean in terms of both grid box effective thickness and gice.

(If you think the above is identical to the Whole PIOMAS Domain - look again, it did give me one of those 'have I screwed up?' moments).

Beaufort Sea.

The distribution of grid box effective thickness is far thinner than before the record year of 2012, which was actually quite a thick year. This suggests a degree of caution is needed when considering ice state and the development of the following melt season. However for ice volume below 3m thick, at 0.99k km^3 Beaufort April volume is substantially below the 2007 to 2014 average (1.1k km^3) and is the lowest overall volume since 1978 (although it is only slightly more than 1 standard deviation below the average). All of the previous years with lower volume in the <3m thick band have significant increases of volume above 3m thick. This is not the case in April 2015, volume for grid boxes reporting over 3m thick is just 0.07k km^3.

It is worth noting that 1982 had no volume from grid boxes over 3m thick, Beaufort is strongly influenced by export from the Central Arctic.

The volumes from the gice sub grid thickness distribution are significantly down, and are generally low for the post 2007 period.

I will have more to say about export, however I noted in my previous post that HYCOM-C-Ice, The Drift Age Model, and ASCAT showed a strong export of multi-year ice extent into Beaufort, Chukchi, and the East Siberian Sea. Those data sources don't give me hard numbers. The numbers from PIOMAS suggest that any such export (which there has been) is sparse and fragmented, leaving Beaufort looking quite vulnerable to strong melt and loss of extent given the right weather.

East Siberian Sea.

In 2010 a large export of multi-year ice from the Central Arctic led to large losses of volume in the following summer, and a remnant of low concentration ice that nonetheless kept extent up throughout the summer preventing very low extent in September 2015. This mass of multi-year ice can be seen in the plot for 2010, as a large volume between 3 and 4m thick, dark blue line.

Now in April 2015 the ice state (red) shows a monomodal spike indicative of first year ice dominance and negligible volume for thicker ice. At present 2015 does not look like a re-run of 2010. The state described above for Beaufort, which also holds for Chukchi does not suggest that there is a mass of multi-year ice poised to play a role in the 2015 melt season in the East Siberian Sea.

In the gice data below, the pulse of multi-year ice can be seen in 2010 (here it is purple - sorry), while April 2015 in the East Siberian Sea shows low volume in the 'monomodal peak' around 2m, and in the thicker mechanically deformed ice above 6m thick.

2015 is not one of the lowest overall volumes in the post 2007 period for 30 April.

I've used 3.30m as a demarcation in the following graph, and have just had to explain the use of that break to a commenter on another post. Here is the core of the explantion:
The 2m thickness falls into the 1.93 to 3.30m band, nominal thickness 2.61m. So I have used the upper limit of that band after examination of the data confirmed my suspicion that it would be a reasonable choice. That thickness (3.30m) should encompass mainly first year ice below it, with mainly deformed ice and multi year ice above it.

So using a 3.30m cut off, here are the gice volumes above and below that level for the East Siberian Sea on 30 April, post 2007.

It is clear that 2015 is a typical post 2007 year, so on the basis of ice state I see no reason to expect a re-run of 2010 with persistent extent late into the season. But before I leave the East Siberian Sea, I'll roll a belated easter egg to you.

Taking grid box effective thickness data, here is the relationship between thick and thin ice for the East Siberian Sea.

Now if only everything in sea ice data were that tidy....

Central Arctic.

And as is to be expected for those following the pack, volume in the Central Arctic remains high with a strong presence of thick multi-year ice.

2012 involved a substantial melt of ice within the Central Arctic, does this mean that with similar weather the Central Arctic will be much more resistant? Most of the thickest ice is crushed up against the Canadian Arctic Archipelago, away from the ice edge even in a year like 2012. I think prospects for this year are all in the weather.

A Tale of Two Ice Packs.

In closing I need to give some more detail into why I am seeing two radically different ice packs in this year's April data. I have used gice data for 30 April 2015 and have concentrated on the Arctic Basin, the region within which most of the 'work' in setting the coming September minimum will be done. The Arctic Basin is Beaufort Sea, Chukchi Sea, East Siberian Sea, Laptev Sea, and the Central Arctic. I have then separated out the Central Arctic, with the rest of the seas being the 'peripheral seas' of the Arctic Basin. Cryosphere Today regions are used.

Again the demarcation of 3.30m is used applied to volume from the PIOMAS gice sub grid thickness distribution. The results are tabled below for the post 2007 years. The colouring on this graph is from Excel (conditional formatting) and is only intended to be an aid to the eye.

2010 stands out, within the Central Arctic ice volume is low, due to the export of ice from that region, while in the Peripheral region volume for above and below 3.3m thick was the highest for the post 2007 period, because much of the export was into the Beaufort/Chukchi/East Siberian Sea regions.

The highest overall volume, and by far the highest volume for ice thicker than 3.3m thick is for 30 April in the Central region. Meanwhile, ice volume below 3.3m thick in the Peripheral region is average, that for above 3.3m is at the low end of the distribution of volume for post 2007 years. So the place where melting will play the greatest role in the coming melt season, the periphery, one cannot say based on ice state in PIOMAS that conditions are any worse than they have been most years since 2007. It really is only within the Central Arctic that one sees conditions that would argue against strong melt.

Of course, this also applies to 2014, and that was a year of poor melt. Last year an anonymous commenter here said my prediction for 2014 was too low because of a multi-year ice intrusion, I suspect that this overplayed the role of multi-year ice, and that as I have since concluded, weather was a stronger factor in 2014.

I have said before that I thought we would see a poor summer melt due to the intrusion of thicker multi-year ice into Beaufort and Chukchi, and further movement into the East Siberian Sea. Now I have numeric data I doubt that this will amount to much and doubt that it will play a major role in the coming season. I presented my prediction for this September's extent in this post last month. But that is based on whole Arctic ocean volume, so is likely to overshoot this year being biassed upward by conditions in the Central Arctic. I need to consider what to do about that.


Anonymous said...

I hate to ask you for more but in the last two tables what would the pre-2007 years look like. For example what would the numbers be for a few1980s or 1990s look like?

David Rennie said...

Great analysis as always, I look forward to reading it in depth over the weekend.

Is there any chance you could send me the consolidated data used to create the PIOMAS Volume distribution with grid box effective thickness All Piomas Domain.

I am just looking for the high level stuff you used to create the graph so I guess an excel table with 40 columns and 6 rows would be about the right amount of data. I can guess with a ruler and magnifying glass but the actual figures would be good.

Chris Reynolds said...


Those last two tables are calculated using gice, as I outline at the start of the post I only have equivalent data back to 2000.

I could do the same using monthly gice for April 1979 to 2014. That would lack 2015, but would give a long term perspective. Otherwise to include April 2015 I can only go back to 2000.

Let me know and I can do it. Also do you want numbers or graphs?


I haven't got round to updating my online data yet, I intend to tomorrow. But if you look at the top right side panel of the page "Sea Ice Data", the top link is to my regional PIOMAS data.

The volume by thickness breakdown is available as a csv file. See under the heading 'The Data', item 4. These are from a Google Drive account I use, let me know if you can download and make use of that data.

I warn you, it is big. But for the Arctic Ocean (regions 3 to 11) there is data as region 17 at the end of the file.

If that data is going to be too detailed for you to use I can easily extract just April and post in a separate file.

1) Have a go at downloading that data so I know you can get it.
2) If you need me to extract April it will be very easy for me to do and I can post as a separate file for you tomorrow if needed.

Let me know how you get on with 1 and whether you need 2.

Chris Reynolds said...


Sorry, not reading what you said closely enough.

"40 columns and 6 rows would be about the right amount of data."

Try this:

DateIndex,0 to 0.9m,1m to 1.9m,2m to 2.9m,3m to 3.9m,4m and above

iceman said...

Your observation of "two radically different ice packs" sounds like a good narrative theme for this melt season. The Beaufort region of what you have marked as "Peripheral Seas" has a considerable amount of thicker and/or MYI, owing to the combination of gyre circulation and anomalous winter winds that pushed the ice deeper south than usual (compared with recent years). Nearby that thick ice is early open water, which looks set to be joined soon by melting off Barrow. It appears that the southern Beaufort will pick up some heat in the coming week - from both insolation and southerly winds - and then the thicker ice will get pushed into the warmer or more fragmented zone. This should give an early indication of whether the "protective arm" of thick ice will remain effective later in the season.

Chris Reynolds said...


But as explained at the end of the above post, having thought that MYI would play a big role I am now no longer convinced there is enough volume of thick ice to have much of an impact. It looks to me like we are not in for anything near a 2010 style survival of ice due to this process.

iceman said...

You were right: the "protective arm" in the Beaufort will continue to wither away over the next few weeks. There appears be a rear-guard arm near the 80th parallel that could afford some protection to the central basin. We'll see how it holds up under the next week of warmth and melt ponding.

Chris Reynolds said...

I'm just about to post about Chukchi, Beaufort compactness dropped early in the season, but has declined at roughly average levels since. So yes, it looks like Beaufort will be protected to some degree by winter's export of MYI.