ORH_wxman

It is definitely weird that we saw an abrupt flip. I wish we knew what caused it. It seems to coincide with the formation of the deep cold pool anomaly in the North Atlantic sometime in spring of 2013 when the deep blocking from early that spring broke down. But I'm not sure what is cause and what is effect and also what is just reinforcing feedback. Probably some combo but interesting nonetheless. The hadgem model from Hadley center seemed to predict this on some level after 2012...which is a nice score by that model.

Yeah if you read the Arctic sea ice forums, most of them are convinced every season is going to be a record low. There's a select few posters on there that do a good job and are pretty objective about the data but you can ignore the other 90% of them. As skier said, most of them just post anecdotes...pictures of buoy webcams or zoomed-in visible satellite pics that show ice "rubble" and then claim that means it has to melt out this year. There's no doubt that this year started off really bad coming out of winter...likely worst on record...but it's clear now that we didn't carry that momentum into spring and summer. It will still be a pretty low ice year because of how bad we started, but I don't think we will get near any records. Maybe volume can still be close if we get some hostile weather in August.

Extent has been keeping pace with the bottom years for the time being even though area has not. Seems like we've gotten a pretty compact ice pack now which has aided in the extent decline. We also lost a good amount of ice in the Greenland sea over the past few weeks as export has come to halt. I'd expect at some point we're going to see a decent stall (or slowdown to be more accurate) in extent loss once the easy pickings are gone...still some fast ice that broke off lingering int hje Laptev and Baffin Sea has some stubborn ice too that will probably contribute to losses in the next few days. The late Andrew Slater's page has the extent minimum coming in around 4.6 million sq km (assuming the value on 9/9 is close to the min), so I think we'll get below 5 million sq km for the daily min. His method has been remarkably accurate for a 50 day forecast...rarely missing by more than a couple hundred thousand sq km. His method also does have a pretty big flattening out in the extent loss in late July and early August...so that supports the idea above, but path to the min isn't always as accurate. The 2012 August losses are pretty easy to see the reason...it was such low concentration already by this point...the difference compared to 2017 is pretty stark:

PIOMAS had a mid-month update....2017 is still lower than 2012 by a very slim margin. It is 150 km^3 lower than 2012...it was 180 km^3 lower on July 1st update so basically no change. However, the difference in the CAB has shrunk to almost zero...compared to July 1st which had quite a bit more CAB volume in 2012 versus 2017. So I think it is only a matter of time before 2012 runs away from this year...but we will see. Until the total number diverges, it is still a race. This seems to be supported by the area numbers too (NSIDC area as measured by SSMI/S)...after starting off the month with some big drops as noted earlier in this thread....we've come to a screeching halt in area loss. We are sitting at 5.62 million sq km for area right now which is more than 800,000 sq km above 2012 at this point (with the big August cyclone still looming in 2012). We are actually ahead of 2013 and 2014 right now as well, but I do not believe we will stay there because both of those years went into big pauses later this month.

Just get us a Kp index of 9 like October 2003. Only time I ever saw legit curtains with the naked eye.

Yeah I'm not sure what he is seeing that indicates we will go lower than 2012...there's still an outside chance we go lower on volume only because the very thick ice near Greenland and the CAA is quite a bit thinner than 2012...but the rest of the pack is in superior shape than 2012 was at this point, so it is hard to see how the area and extent get anywhere near it. Volume will be a task...it will require probably very hostile weather from here on out. We entered July nearly tied with 2012 for volume and so far this July has been significantly colder than 2012. We had a decent chance to beat 2012 on extent/area coming into May, but the weather since then has all but eliminated that chance now...never say never, but I'd put the odds probably below 10%. Maybe if we get a 2007-esque pattern later this month and right through all of August, we will make up ground.

Greenland is actually having a historically cold summer. Kind of weird obviously in the recent warm context but it's a reminder that natural variability still plays a large role.

Finally melting all that snowpack on the CAB ice and getting melt ponding. The bigger question though becomes is this happening too late? 2013 saw similar losses in the same period (670k) before stalling later in the month. OTOH, years like 2008 (the closest match to 2017 for area on June 30th) decided to go nuclear a few days later (about July 5-9 losing almost 800k) and finished pretty low...albeit still around 3.00 million sq km.

Also, probably one of the more underrated events during the whole decline of sea ice in the past decade-plus is the winter of 2007-2008. 2007 might not have been a complete game-changer had the 2007-2008 winter not exported a ton of that compacted multiyear ice. While the summer of '07 melted out a lot of volume, even more was exported the following winter...the '07 summer pushed/compacted all the MYI leftover from the 2006 season (and previous seasons) toward Greenland/CAA and then a really hostile winter/spring pattern proceeded to export a lot of that ice...expediting the transition from a lot of MYI to mostly FYI.

2012 and 2011 were fairly different in volume by September though, 2012 was clearly lower...they were close in July. A much better example IMHO would be 2007...it had a much higher September volume than years like 2013 and barely lower than 2009...yet it had one of the lowest extents/areas at the minimum due to the extreme compaction (caused by the weather)....way lower than 2009/2013.

Here's the distributions of the final area minimum should 2017 follow the same path of each previous year's melt after June 30th. (I.E. you can see that if we followed 2015's melt after June 30th, our minimum would be at 3.00 million sq km) Using percentiles, the 10th percentile is 2.87 million sq km and the 90th is 3.87 million sq km with a median of 3.31 million sq km...however, that is using the entire distribution since 1979 and as I noted in a discussion with skierinvermont, the recent years may be starting to trend a little higher for ice loss after June 30th...this isn't conclusive yet, but I think it should be mentioned. If we only use 2007-2016, then our percentiles would change...you'd get a 10th percentile of 2.80, a 90th perecentile of 3.31 and a median of 3.12. Note how the 90th percentile for post-2007 is the same as the median for the entire distribution...this is because in the past 10 years, we haven't had any of those very slow post-June 30th melt years like we saw especially in the 1990s....but even 2005 and 2006 were quite slow in area loss after June 30th. Thus, given all the information...my prediction for 2017 min will be 3.10 plus or minus 200,000 sq km...so a range of 2.90 to 3.30. This covers a lot of rankings of course. We could theoretically finish anywhere from #3 to #8.

Final June area numbers are in and my prediction will probably be somewhere around 3.1 million sqkm for area. I'll have the full expected distribution of results in the next day or two when I have time to run them on excel. We finished the month 250k higher than last year...so that is probably going to put the chances at beating 2012 near zero considering even last year's epic melt out post-June didn't break it.

2012 likely already has lower volume now than 2017 (they were basically tied in the mid-month PIOMAS update) and the area at this point as measured by NSIDC SSMI/S (CT SIA by proxy) was significantly lower in 2012 which suggested a lot more melt ponding than this year. We will need a pretty incredible weather pattern to finish lower than 2012 IMHO. It's not impossible, but I'd put the chances as very unlikely.

It means 2016 was 330k lower than 2017...you can kind of tell just looking at the years. All the negative years are the bad ones like 2007, 2010, 2012, 2016....and the ones positive or higher than this year are the better ice retention years like 2014, 2013, 2009, etc. This year has kind of an in-between vibe to me...like a 2008 or maybe 2015 (talking area...because 2015 was low extent, but very compacted so the area actually wasn't that low in the post-2007 context).

With a couple days left before the June 30th prediction point, here's where we stand on area: 2016: -330k 2015: +30k 2014: +110k 2013: +280k 2012: -700k 2011: -320k 2010: -650k 2009: +640k 2008: +130k 2007: -160k

FWIW, the outliers for post June 30th losses since 2007 are 3.70 million sq km in 2010 and 4.51 million sq km in 2016. That's still a pretty wide range that has to be taken into account in the uncertainty.

Agreed...the last two years especially make the more recent trend more noticeable. The 2010-2014 (sans 2012) stretch was kind of middle of the pack but we've seen pretty good losses the past couple years and when you start adding in 2012, 2008, and 2009, it might be a newer normal. We'll have to see, but it's becoming harder to envision losses that are closer to 3.5 million sq km after June 30th versus 4 million sq km in the post 2007 years (average since 2007 has been 4.1 million sq km versus 3.56 from 1996-2006 or 3.73 if we include 1979-2006) I'll probably adjust my forecast this year to assume losses closer to the 2007-2016 average post-June 30th.

Wipneus has finally started giving some NSIDC updates again, so we can compare to past years on CT SIA....the following shows where other year's are in relation to this year (i.e., 2016 was 240k lower than this year on this date) 2016: -240k 2015: +200k 2014: +240k 2013: +520k 2012: -590k 2011: +10k 2010: -620k 2009: +780k 2008: +150k 2007: -130k You can see the closest match right now is 2011, though 2011 saw nuclear drops in the final 4 days of June, so it will be tough to keep up with that year for June 30th readings (which aren't available until July 1st). Traditionally, I have used the end of June's value to predict the final outcome of CT SIA with moderate success...though last year was a huge miss outside of the 5% confidence intervals. It was the largest loss of area after June 30th in the data base including 2012. So we'll see how well it does this year. As a reminder, here are the minimums on CT SIA (or the NSDIC equivalent) each year since 2007: 2007: 2.919 2008: 3.003 2009: 3.424 2010: 3.072 2011: 2.904 2012: 2.234 2013: 3.554 2014: 3.483 2015: 3.094 2016: 2.427 You can actually see how using past years to predict what 2016 would be all fell outside of what actually happened....the graph below shows what 2016 "would have" finished with if it followed that year...I plotted the actual result of 2016 on the end to show how it was lower than previous variations of area loss after June 30th. The closest result would have been following the losses after June 30th in 1989...but that still would have produced a value of 2.53 million sq km. Previously, there had been no trend in area loss after June 30th, which made the prediction somewhat reliable. But we'll have to see if 2016 is the start of a downward trend or if it was an outlier.

The last two winters have definitely been poorer for first year ice with much warmer conditions than the previous 3 winters...though the thermodynamic thickening curve would still have the FYI achieving most of its possible thickness even in those warmer winters...we probably need another 2C or so to really start seriously denting the FYI thickness. I do wish we had better subsurface data on the SSTs up there, because I'm wondering if last year's El Nino played a role in the rapid melt during the 2nd half of the summer by increasing the flow of warm subsurface into the pacific side of the arctic...since last year didn't really have favorable melt weather outside of a very intense dipole for about 2 weeks in August.

Agreed with that. The very low volume at the start of the season will make it easier to get into the top 3-5 lowest seasons even if the weather remains kind of mundane, but 2012 is so far ahead of the others that we'll need some pretty big melt weather to catch it.

There's actually a pretty good dipole pattern that gets going later this week, but it doesn't look like it will last more than a few days....but I also recall that July 2015 didn't initially look like it was going to last, but it was able to for most of the month. We'll have to see.

The rapid regression of PIOMAS seems to suggest that it might have been a little low in May versus reality. Hard to say for sure. We'll see what it does in the next couple weeks.

The area losses look like they are mostly coming from Hudson Bay and perhaps the Kara. The Kara is more meaningful, but Hudson bay can deceive us sometimes in both directions...the main basin looks pretty high concentration to me right now. Of course, we don't have SSMI/S area data right now...that's really the most important one for predictive purposes because unlike AMSR2 data, it is sensitive to melt ponding.

That's kind of weird. I know the winters there have gotten quite a bit colder in the past decade or two vs the 1990s (we've seen the whole autumn snow cover feedback on the Siberian high)...but it was the opposite in the warm season. I wonder why it would start cooling.

Yeah it looks like the only time a drop that high occurred in the record was between 1957 and 1958.