PIOMAS September 2014: Massive Volume Gain

The story of the 2014 melt season has been one of unexpected ice survival due to weather being largely unfavourable for largescale ice melt. Now, after a delay to NCEP/NCAR reanalysis data, the PIOMAS volume data is out and it is possible to see how much volume gain will be carried over into the 2014/15 winter, and very likely beyond.



First the PIOMAS Main Series, I use that term to differentiate from the gridded data I predominantly use. The main series being the numbers available from the PIOMAS public website, here. The timeseries of the daily minimum volume is shown below.

Fig 1. Main series PIOMAS daily minimum volume, 1979 to 2014.

The interruption to the long term exponential decline period is strengthened by summer 2014. However by 2014's minimum the increase from the minimum of 2012 is 3.137k cubic km, of this increase, 1.719k cubic km is due to 2013, a lesser 1.418k cubic km is due to 2014. So 2014 is not the whole story, 2013 played a greater role.

Fig 2. PIOMAS Volume anomalies from 1980 to 1999 baseline.

As shown in figure 2 above, 2014 has seen the anomalously strong spring melt typical of post 2007 years, but strongly exhibited after the 2010 volume loss event. In 2013, a year of poor melt, before the summer season events started with a robust spring melt despite poor weather. I consider that the muted spring melt of 2014 is probably due to increased volume in the Central Arctic, which I have shown to be a major regional player in the post 2010 spring melts (link), in 2014 this has been combined with poor melt weather.

Referring now to results from the PIOMAS gridded data, figure 3 (below) shows a stacked plot of volume contributions from five thickness bands for the whole of the PIOMAS domain (can be considered 'all Arctic').

Fig 3. Volume contributions to September PIOMAS volume from grid cell thicknesses in five thickness bands.

Looking from left to right. The long term decline of the thickest ice can be seen (3m and above, purple and light blue). Note that this is for September, prior to 2007 ice over 3m thick was a major factor even at the end of summer. Also one can see a small increase in ice 1 to 2m thick (red), with a commensurate decline in volume from ice 2 to 2.9m, this was due to ice thinning at the end of summer as the whole pack moved to thinner ice.

Then in 2007 there is a step drop in volume whose effects remain with us even through to this September, that was a result of unusual weather in 2007 acting upon an ice pack weakened by decades of attrition from anthropogenic global warming. In 2010 another volume loss event cleared out the thickest ice setting the stage for further declines that culminated in the 2012 area/extent crash, also ushering in a period of unusually aggressive spring melts as discussed above.

My point in revisiting all of this is to stress the impacts of 2013 and 2014's poor melt summers. Ice state is now at a point similar to the years 2009 and 2008. With the extra volume in the Central Arctic, the transport of older thicker ice into Beaufort and through to Chukchi  (Pacific sector) will buffer those regions against anything but a 2007 style weather driven crash. For some years, until this volume pulse works through the system, spring melts will be less aggressive, and melt in the Pacific sector of the Arctic Ocean will not be as aggressive as in some recent years, so repeats of 2012 are now rather unlikely, probably for much of the rest of this decade (in my opinion).

The following graph is one I've kept using for the last two years. It shows ten day increments in volume and is effectively the rate of change of sea ice volume, the post 2007 years are shown. Hope the large size works in people's browsers.

Fig 4. Ten day increments of volume loss, the volume change is from the stated date minus the volume tend days earlier.

In figure 4 2014 is shown in red, it can be appreciated that the differences are actually small, volume losses have been a similar order of magnitude to other post 2010 years throughout the melt season. This curve being largely driven by an invariant factor for each year; solar radiation. Weather merely modulates this factor.

I have recalculated this plot as anomalies from the average ten day rate of change for 1980 to 1999 to show changes from that past baseline and highlight recent changes.

Figure 5. Ten day rate of change of PIOMAS volume expressed as an anomaly from the 1980 to 1999 baseline.

Figure 5 clearly shows the spring melts as being larger than in the past (more negative anomalies), however 2014 is seen to be a weak spring melt, especially when compared to 2012 (orange). After 1 July the losses become less than the average (more positive anomalies), and 2014 is seen to have the highest anomalies (weakest late summer melt rate) of any post 2007 year.

****

As usual, in introducing the breakdown from PIOMAS gridded data here is the map-plot of grid box effective thickness.

Fig 6. Map plot of PIOMAS grid box effective thickness for September 2014.

The large mass of blue is ice over 2m thick, however also bear in mind the mass of darker red surrounding it. Note that ice in the 0m category is not likely to represent real ice, as can especially be seen around the coasts (which is why PIOMAS thickness neglects grid boxes below 0.15m thick).

The contrast with 2012 is striking, as can be seen in figure 7 below, sorry I've not got the time to do a blink comparison right now.

Fig 7. Map plot of PIOMAS grid box effective thickness for September 2012.

In recent years, by September most of the volume is in the Central Arctic, this is the case this year. So the general picture painted in my August PIOMAS post where I noted that "78% of the total PIOMAS volume increase from 2012 is from grid box thicknesses between 2.0 and 3.9m thick." The thickness distribution has thinned by September, but the take home message remains that the Central Arctic holds almost all of the extra mass of ice. This can be appreciated by looking at the timeseries of regional volumes in figure 8 below.

Fig 8. Regional volumes from PIOMAS gridded data. NB the scale for Central Arctic is on the right vertical axis due to the dominance of Central Arctic volume.

I've also taken the time to update my calculations from PIOMAS of percentage open water formation as a function of April thickness. The following graphs show the result of calculating the percentage of area in April for a given thickness that becomes open water by September.

 Fig 9. Percentage open water in September as a function of April grid cell effective thickness, for all years 1978 to 2014.

In figure 9 (above), all available years are shown, with earlier years darker blue and later years lighter, 2014 stands out in red. The plot shows the increasing open water formed as ice thickness in April goes below 2m thick. In recent years there has been erratic formation of large percentage open water for thicker ice, this is caused by there being fewer grid boxes with thicker ice, and the large inroads of open water into the Central Arctic region, notably in 2012 as will be seen in the following graphic.

2014 is still in with 'the pack' and at thinner grid box effective thickness shows behaviour similar to recent years, for example, taking 1.3m thick and moving up the graph the mass of plots transits from darker to lighter blue. This plot is really only shown to give an idea of spread, the useful detail is in figure 10 below.

Fig 10. Percentage open water in September as a function of April grid cell effective thickness, averages for selected periods and 2014.

Figure 10 (above) has the decades out of order, sorry, but once you've got round that: In the 1980s thicker ice showed the lowest open water formation, with the 2000s and 1990s producing marginally more open water formation than the 1980s (this may not be statistically significant). The real change comes in the post 2007 period (here 2007 to 2012), where the entire curve starts to lift upwards. This lifting is the start of the transition to a seasonally sea ice free state.

I have previously calculated the long period averages for such graphs for various seas, here is the Beaufort Sea.

Fig 11. Percentage open water in September as a function of April grid cell effective thickness, averages for selected periods up to 2012.

In the Beaufort Sea the plots are less tidy than for the whole Arctic, but the principle remains. In the 1980s and 1990s only half of the ice over 2m thick melted out, keeping the orange and red curves from 100%, down around 40%. Then in the 2000s thicker ice started to melt out for the same reason as in figure 9 this created large spikes as there was less thicker ice to melt out. After 2007 the character of the curve changes, it cleaves close to the 100% level because virtually all the ice of all April thicknesses had melted out by September.

Now, referring back to figure 10, it can be seen that in 2014 the behaviour of thicker ice was more like a pre 2007 year. There was so much thick ice in the Central Arctic, and the weather favoured ice survival, so hardly any thick ice melted out to reveal open water. However whilst this was the case, between 2 and 2.6m and between 1 and 1.5m 2014 was more like a post 2007 year.

For completeness, as these have been in all of my summer status posts, here is the thickness distribution plot for the whole PIOMAS domain. I don't think that there is much to be said about it that wouldn't be labouring the point.

Fig 12. Thickness distribution for all PIOMAS domain.

In summary: The extra volume in the sea ice system is massive, in my opinion it removes the chance of a 2012 type crash, leaving only the chance of abnormal weather to remove this mass of ice within a few years. The linear trend of volume loss in the Central Arctic from 1995 to 2012 was -344 cubic km/year (+/-36 cubic km/year). If years with weather not conducive to melt do not become a regular occurrence and the preceding typical summer weather reasserts itself, then the current volume gain of 3137 cubic km would take 9 years to remove, based on the assumption of linear trend losses. However a strong winter export into Beaufort/Chukchi/East Siberian Seas, or through the Fram Strait can be expected to reduce this significantly. That noted I think that the volume gains of 2013 and 2014 have severely reduced expectations of a virtually sea ice free state by 2020.

All of this is separate to my recent post and activity on the Sea Ice Forum about a slow transition. In fact this complicates the situation for me. Another crash this year would have been more helpful to my case as we could have seen whether Arctic Ocean volume in April was again around 19k cubic km. As things stand I will have to wait some years to see if, as I expect, once again the Arctic Ocean Winter volume seems to stall at what one would expect for ice around 2m thick. However I do expect that in years to come April volume will once again start to level at the volume expected for around 2m thickness. In a similar manner, whilst the increase of volume suggests that spring volume loss will be more muted in recent years than in 2010 to 2013, once this pulse of volume works through the system I expect that this phenomenon will also re-emerge.

On the subject of the 'Slow Transition' argument, I will be returning to that in a couple of upcoming blog posts. But over this winter any 'status' posts will be as and when needed, not every month.


The derived sea ice volume data I calculate from PIOMAS gridded data has been updated, apart from daily volume (gridded product not updated), it is available here. This data will continue to be updated over the winter, regardless of whether it is addressed in a blog post.