Post 2007 Summer Melts: Ice or Atmosphere?

The issue of what is causing the increased melt season sea ice losses post 2007 is intriguing, there seems to be two candidate players, the ice and the atmosphere. In my opinion 2013's muted melt gives a hint as to what the main player is.




Since 2007 the annual range of area and extent has increased, this is due to increased losses in the melt season with autumn/winter area and extent remaining broadly unchanged. In terms of area and extent that winter maxima should not be affected is not surprising. The maximum area and extent are largely set in seas outside of the Arctic Ocean, so conditions post 2007 within the Arctic Ocean will not easily impact the maximum in area or extent. However the increase in losses are notable.

From the above table it can be seen that for Cryosphere Today Area (CT Area) and NSIDC Extent the annual melt has increased when the average is used for pre and post 2007. For CT Area the average melt has increased by 10%, for NSIDC Extent the increase is 15%. That this is due to a real increase in all of the years from 2007 to 2013 is seen by ordering the years in terms of their melt season losses, all of the years 2007 to 2013 are found to be in the top ten of losses for both CT Area and NSIDC extent.

I've previously blogged about the summer pattern and research that shows it generates an Arctic Dipole pattern in summers from 2007 to 2012. This pattern draws in warm air from the north Pacific and drives it over the pack, which is reasonably assumed to increase melt. However there is another factor, sea ice has thinned after 2007, and thinner ice will melt more readily exposing water, driving further loss of ice by ice albedo feedback. Indeed for the peripheral seas, Beaufort through to Bering, the average September thickness is ~0.5m for 2007 to 2012, but is ~1.0m for 2001 to 2006 (PIOMAS data, my calculations), and it is losses in these seas that set the minimum. 2013 thickness cannnot be calculated due to full data not being available as yet.

Those scientists I've discussed this matter with have thought it's probably a mix of the two factors, both ice state and the dipole anomaly. Now 2013 may help to shed a little more light on their relative roles.

I've calculated the losses between each monthly average using CT Area and NSIDC Extent, then the average loss has been calculated for each intermonth loss from 1980 to 1999 (my usual baseline average). This average has then been used to calculate the anomaly of intermonth losses for the whole satellite period throughout each summer, these can be consider monthly loss rates.

It can be seen that all of the post 2007 years show a highly anomalous pattern of loss rates, with strongly positive loss rate anomalies. June in particular is seen to be an aggressive loss, this is probably due to larger amounts of flat undeformed first year ice facilitating ice pond formation, which initially masquerades as open water, but preconditions the flooded ice for melt later in the season due to lower albedo, more detail here.

The ice state is a given that has only got worse, until the 2013 melt season when a slight recover in thickness seems likely. However the 2013 muted melt was due to atmospheric conditions, notably a slow start caused by cold conditions in May. Most years since 2007 have seen a positive Arctic Oscillation (AO), until 2013. However this is a state that has persisted since 2003.

Therefore the AO isn't looking a good candidate for playing a role in the increased melt season losses.

However since 2007 a new pattern of atmospheric circulation centred on anomalous ridging over Greenland has manifested.

The above pattern has been correlated with NCEP/NCAR sea level pressure (SLP) for each summer month from 1979 to 2013. The correlation index series produced is intended to give an indication of the strength of the above pattern in the SLP data. I've called this the Summer Pattern Correlation (SPC).

Due to the AO and the SPC sharing a common element of pressure over the Arctic Ocean the post 2003 period shows increasingly positive SPC values, however in contrast the AO the SPC shows very high correlations after 2007, except July 2010. This is reflected in visual examination of SLP maps over that period.

Notably the SPC shows a strong negative pattern in 2013, and indeed pressure maps for 2013 show that the average summer pattern was almost the complete reverse of the map plot shown above. So one could argue that the positive dipole of 2013 has drawn in warm air from the Atlantic, not the Pacific. To address this concern I've used 850mb temperatures to detect flow above the ice rather than surface temperatures which tend to be pegged to zero above the ice.

For the June July August mean temperature from 2007 to 2012 the cold pool over the Arctic is seen to be shifted towards the Atlantic, and the coldest region is indented from the Chukchi Sea, this is due to what I interpret as the intrusion of warmer air from the Pacific, causing the deformation in the cold pool as indicated. In 2013 there was no similar deformation from the Atlantic side. So I don't think the 2013 melt was due to a warm dipole driven intrusion from the Atlantic.

From my reading the evidence suggests that the prime mover in the increased melt season losses after 2007 is ice state, not the Arctic Dipole. However one year out of seven is not strong, so I stress that I consider the evidence as suggesting this conclusion.