[UPDATE Aug 24-28: The previous record low for daily Arctic sea ice extent was 4.16 million sq km, set on September 14, 2007. The new record was first set on August 24, 2012 and now stands at 3.85 million sq km. Sea ice extent was reduced by more than 430,000 sq km in five days (Aug 23-27), the most rapid late August loss on record. Click on thumbnail at right for latest Arctic sea ice extent as of today (based on latest NSIDC daily data).
- Aug 22: 4.29 million sq km
- Aug 23: 4.19 million sq km
- Aug 24: 4.09 million sq km ***
- Aug 25: 3.97 million sq km
- Aug 26: 3.94 million sq km
- Aug 27: 3.85 million sq km
*** New record low for daily Arctic sea ice extent set on Aug 24.]
This year’s arctic ice melt season is generating extraordinary interest. 2012’s apparent descent toward a new record low in extent and area is dramatic enough, but it also comes as new analysis shows that summer sea ice loss is 50% more than previously thought in terms of volume, according to preliminary satellite data from CyroSat 2. Virtually ice-free summers in the arctic sea could well arrive by 2030, with troubling implications for accelerated albedo feedback and possibly disruptive changes in the jet stream.
A new record low, eclipsing 2007, does seem increasingly inevitable with each passing week. National Snow and Ice data Center data showed Arctic sea ice extent at 4.29 million square km yesterday, just under 2007’s September average, and a level only reached on September 7 back then. To be sure, 2012 is starting to bottom out, but most years have seen a similar pattern around now.
Here’s a snapshot of the 2012 melt season (with the small crosses denoting recent daily values), compared to the previous five years.
Just how low could 2012 go?
To try to answer that question, I’ve tried my hand at short-term prediction. I used a rolling five-day average to generate bi-weekly projections, eight in all. The model is very simple – the most recent five-day August average is used as a predictor for the September average, based on a regression on the last 11 years. To illustrate, the following chart shows projections generated for five-day averages centred on August 3, 10, and 17. It also shows the regression for the August 22-26 five-day average, which will become available in a few days.
The fit is not that good early in August, but improves considerably by mid-August; over that same time, 2012 has gone well below 2007. Now here are the other rolling projections, including the latest one generated today and one to come in a week.
Since daily data becomes available only the following day, the prediction for each five-day average is actually only available three days after the nominal “centre” date. So today’s projection (August 23) is based on a five-day average from August 18 through 22 (i.e. centred on August 20).
The current projection is 3.67 +/- 0.17 million sq km. As shocking as such a breach of the 4 million mark would be, it could be argued that this might even be somewhat conservative. The most recent years all tend to hew quite close to the regression line (e.g. 2007) or else fall below it (e.g. 2008). On the other hand, the 2012 melt season might yet surprise us once again.
Here are the projections as they have evolved since the beginning of August, including today’s.
Over the course of the month, the confidence interval has narrowed and the central value has stabilized considerably. It will be interesting to see how this develops over the coming days and weeks.
[UPDATE Aug 24: Yesterday, the NSIDC extent fell 100,000 sq km and stands at 4.19 million sq km, just 30,000 sq km above the previous record of 4.16 million sq km set on September 14, 2007. The new record should come any day now, as can be seen from the updated chart.]
UPDATE Aug 25: Yesterday, the extent fell another 100,000 sq km and stands at 4.09 million sq km, setting a new record.
UPDATE Aug 26: Yesterday, the extent fell another 100,000 sq km and stands at 3.97 million sq km, breaching the 4 million mark for the first time.]