2019's Arctic sea ice extent was on a roll -- downward -- for longer than usual. But now it's pulling up short and looks very unlikely to fall below 2012's minimum (JAXA data):
2019 is still only a slight deviation, but 2012 was still melting rapidly at this time of the year, and I don't just can't see 2019 making headway against that.
(Added 10 pm Pacific Time: the JAXA number for 8/19/2019 SIE has decreased from -258 km2 below 2012 yesterday, to, now, -309 km2. That's a huge jump in one day. Puts a record even further out of reach. It's not going to happen.)
So when might 2012's record low be broken?
Below I've taken the annual lows since 1979, using NOAA's daily data, and, leaving 2012 out of the equation, calculated the trend in the annual minimum. (I'm switching between JAXA and NOAA data only because of where I have the relevant graphs; they don't differ by much, but the graphs not so easy to replicate):
This trend for the min is -78,000 km2/yr. Given 2018's min of 4.55 Mkm2, the trendline won't fall below 2012's min (3.34 Mkm2) for 16 years after 2018 -- that is, in 2034.
2012's summer was clearly extraordinary. There was a summer cyclone that chewed the ice up, meaning it had more surface area exposed to the air and sea and so melted faster. (Ice chips in your restaurant water melt faster than ice cubes.)
Of course, my calculation assumes no years between now and then will be extraordinary lows due to, say, the same kind of natural variability that was seen in 2012.
2012 was extreme natural variability in action. Similar to the average USA48 temperature that saw 1934 the warmest year for many decades after. (Probably aided some by [anthropogenic] Dust Bowl conditions.) It was only with the 1997-98 El Nino that its annual average was surpassed. But now 1934 ranks only 7th highest of USA48 average temperatures. There same will eventually happen with 2012's record SIE low.
Added 8/22: Here's the situation as of 8/21: