I had a look at the 7 day extent loss categories (0-100k, 100-200k, etc) during the melt season how how their numbers have changed during the satellite record.
I split the melt season into March 8th (just to avoid issues with leap years) to September 30th, and looked at how many 7 day extent losses there were for each category, for each year.
First off, the number of 7 day periods with any melt is shown below.
A bit of variability, with 1999 being the main anomaly, and a slight upward trend, but nothing really significant.
Below is the number of 7 day extent losses between 0-500k.
The trend here is downward, with the first 5 years averaging 144.6 and the final 5 years averaging 122.2. The years 2012 and 1999 stand out more than any other.
Below is the number of 7 day extent losses greater than 500k.
Here we see a clear upward trend, with 2012 standing out like a sore thumb. The average of the first 5 years in 39.4, while the last 5 years is 59 (though possibly skewed by 2012) which represents an almost 50% increase in the number of 7 day periods where extent losses exceeded half a million km2.
But it seems that it's the extreme melt periods that are really growing. Below is the number of 7 day extent losses greater than 700k.
The trend is extremely clear, from an average of 4 for the first 5 years to 23.4 for the last 5 years, an increase of over 500%.
So it seems one of the main contributors to the decline in Summer/Autumn sea ice is the increase in extreme melt periods, rather than general increases or a lengthened melt season. This would suggest that as the pack has weakened, the unfavourable weather conditions can now cause rapid ice losses, rather than the moderate losses seen in the past. This explains the downward trend in low to moderate 7 day losses (trend strongest between 200 and 500k) and the upward trend in losses above 0.5 million km2, especially above 0.7.
I might post up a bit more on this later tonight or in the next few days, or if anyone wants any details for particular years, just ask.