2018 sea ice area and extent data

[RoxTheGeologist]

How much of the recent cliff is driven by Kara vanishing suddenly? Because if it is, the cliff should end in a day or two and we go back to the slow loss that has otherwise characterized this season.

I think that will depend on what happens in the Beaufort Sea (and perhaps the ESS as well).

We might finally be seeing the result of all the extra insolation that we had early in the season on the Pacific side. A lot of extra energy was soaked up by the open water in the Bering/Chukchi and the Laptev, even if the days weren't entirely sunny. Water absorbs roughly twice the energy from sunlight as ice, and ten times more than snow.

I agree with Neven, the Beaufort and ESS are going to be very important for the final extent numbers.

[magnamentis]

I agree with Neven, the Beaufort and ESS are going to be very important for the final extent numbers.

i think the heat soaked, and stored in unknown places and now suddenly popping up again out of nowhere is not really logical but what do i know, just does not sound reasonable to me while:

talking about that area, i think that's quite a temp for up there and interestingly temps went up by about 7-10C within 2 days after the coastal ice finally gave way, this may hint at what could happen the year when there will be no ice at all, probably a big jump of 10C or more arctic wide.

[CalamityCountdown]

In both 2016 and 2017, NSIDC SIE extent declined by just under 1,000,000 km2 during the upcoming 10 day period. So anything less than daily century extent losses during the rest of this month (July) would be below the trend set during the last couple of years

[Juan C. García]

[ADS-NIPR-JAXA] ASI Extent.

July 24^th, 2018:   

[Neven]

It's a bit too much this year. 

[Neven]

Double century drop. Nice!

[gerontocrat]

NSIDC Total Area as at 24 July (5 day trailing average =  5,143,824 km2
This is at 183 k above the 2010-2017 average

Total Area loss 84 K, Central Seas 55k, Periphery loss 17 k, Other Seas loss 12 k  
Analysis of individual seas.

Pacific Side
- The Bering Sea area is now 1 k,
- Chukchi Sea loss 10 k,
Atlantic Side
- Baffin Sea loss 9 k,
- Greenland Sea loss 7 k,
- Barents Sea loss 0 k -  area is 2 k, i.e. melting season over,
- The Kara Sea area loss 8 k, Area now just 74 k.
- The Laptev Sea area loss 7 k.
CAB
- Beaufort Sea loss just 1 k,
- The Central Arctic Sea GAIN 9 k,
- The Canadian Archipelago loss 6 k,
- East Siberian Sea loss 31 k .
Other seas
- St Lawrence area at 2 k,
- Hudson Bay area loss 12 k,
- The Okhotsk Sea area is 5 k.

Area loss slightly above average. Extent losses are catching up with area losses again, both NSIDC and JAXA daily measures especially. The question is - will this be reflected in higher Area losses over the following days, or is it simply Extent catch-up. See graph

Note:
- Okhotsk Sea moved to other seas - it is so detached from the Arctic Ocean (sea, estuary, lagoon, whatever).
- Beaufort Sea moved to CAB - as it is one of the four seas that are always the most reluctant to melt. None as yet have lost 50% of their area maximum. Some will disagree, "but this post belongs to me"
[/quote]

[gerontocrat]

JAXA ARCTIC EXTENT 7,132,042 km2（July 24, 2018）

Just to add to Juan's post.... Wait a minute. Juan's gone to bed and missed it. He will not be pleased. And I will not be pleased if JAXA tomorrow say "whoops - glitch while we were testing."

One cannot deny that an extent loss of 1 million km2k 8 days and 4 100k+ days in a row followed by a daily loss of 201k is impressive. As a result :-
- Extent is down to 122 k km2  (1.7%) above the 2010's average extent on this date,
- and only 127k (1.8%) above 2017,
- Extent loss to date is now 500k km2 (6.9 %) below the 2008-2017 average, with 72.9 % of the average melting season done.

Resulting minimum from average remaining melt is down to 4.43 million km2, (excluding 2012 from the average 4.52 million km2). Range of results from last ten years remaining melt is 3.60 to 4.91 million km2.

8 days losses above average in a row amounting to 1 millionkm2, that 2017 feeling fades yet more but lingers on- extent losses are not yet enough to catch up on the slow melt to date. There is, on average, just 27% (50 days) of further extent loss to go. Could the melting season last a bit longer - Yes. Daily extent loss, on average, is already in significant decline. Nevertheless, A September minimum down one bin in the range of 4.50 to 5.00 million km2, and perhaps yet another bin lower, now looks increasingly the most likely outcome. But that is still around 0.5 million km2 above 2nd place.

The season surprises again.

[uniquorn]

Very cliffy

[RoxTheGeologist]

i think the heat soaked, and stored in unknown places and now suddenly popping up again out of nowhere is not really logical but what do i know, just does not sound reasonable to me while:

Obviously the heat is stored in the water column (where else could it be?). More bouys are required to see differences. A logical idea for it popping up out of nowhere would be mixing, both horizontal and vertical. The heat for the melt caused by GAC in 2012 was already entrained into the ocean.

[binntho]

... entrained ...   

[FishOutofWater]

Some of the Pacific water has flowed into the Arctic Ocean and then rotated eastwards in the Alaskan coastal currrent along the Alaskan coast into the Beaufort sea. Some of the Chukchi sea water has sunk below the ice. The water that entered from the Pacific is doing some damage to the ice now. Likewise salty Atlantic water is melting ice from below on the Atlantic side of the Arctic. Some salty water has flowed at depth into the shores of the CAA. That water is helping to open the passages of the CAA by melting ice from below.

Moreover, there's been an astounding amount of heat moving downwards in the upper layers of the north Atlantic and the Labrador sea has warmed all the way down to 3000m as warm salty water mixed downwards this winter and spring.

[sinocentric]

Apologies if this is the wrong place to ask this, but how common have 200k daily extent losses been in previous years?

[magnamentis]

i think the heat soaked, and stored in unknown places and now suddenly popping up again out of nowhere is not really logical but what do i know, just does not sound reasonable to me while:

Obviously the heat is stored in the water column (where else could it be?). More bouys are required to see differences. A logical idea for it popping up out of nowhere would be mixing, both horizontal and vertical. The heat for the melt caused by GAC in 2012 was already entrained into the ocean.

of course it's there but what i mean is that it has always been there, dispersion is responsible for the extent number not hidden heat IMO and again i say that perhaps we or those in charge should consider a change of the 15% threshold. i wrote about it but it went mostly unnoticed even though i hinted at the fact that previous periods were not impacted but current conditions would be illustrated much better. IMO we only have around 60-70% of the ice cover compared to 8 years ago but extent shows 90-110% of ice cover due to the 15% threshold.

if that were changed to 35% previous periods woulds change little but the last 3 years would show the difference.

but however, apparently the ice got so thin and fragmented over the last 3 years that even the 15% threshold can't hide the real state of the ice anymore.

[Neven]

Apologies if this is the wrong place to ask this, but how common have 200k daily extent losses been in previous years?

As far as I can see, since 2005, there was a 228,644 km2 drop on the exact same date in 2012.

[DavidR]

Apologies if this is the wrong place to ask this, but how common have 200k daily extent losses been in previous years?

There have been two drops of over 200 K both on 24th July. 2012 had a 228 K drop.  So 24 July 2018 this year wasn't the biggest on this day, but  it was bigger than every other day in the Jaxa record since 2003.

There have been 23 days where the decline was greater that 150K. Six in 2012, three  in 2015 and three so far this year. The earliest was Mar 23, 2014 and the latest was Aug 9th, 2012.

[gerontocrat]

Quote from magnamentis

again i say that perhaps we or those in charge should consider a change of the 15% threshold.

15% is OK by me. NSIDC Satellite #F18 is OK by me.

Yes 15% is arbitrary. Yes, #F18 uses old technology. But consistency in the data record for 39 years is a prize not lightly discarded.

But that does not stop me looking at area as a better idea of what ice cover there is.

And we also have PIOMAS volume to give an indication of the volume of ice remaining - which in the end is the metric that will (and does) matter most.

It would be great to have a load of new satellites up there armed with the latest Gizmos, loads of buoys wandering around the Arctic looking at the ice and water temperatures etc etc.   Much more accurate data - yes, comparison with previous years - no.

So looking at the current situation of sea ice requires more than one simple metric. But for the general public - it is all about NSIDC Sea Ice Extent as it is has been defined since 1979. Change the definition of the measure and a forest of trolls will muddy the water (ice) for years.

[Juan C. García]

JAXA ARCTIC EXTENT 7,132,042 km2（July 24, 2018）

Just to add to Juan's post.... Wait a minute. Juan's gone to bed and missed it. He will not be pleased. And I will not be pleased if JAXA tomorrow say "whoops - glitch while we were testing."

One cannot deny that an extent loss of 1 million km2k 8 days and 4 100k+ days in a row followed by a daily loss of 201k is impressive. As a result :-

Thanks, Neven for the graph and Gerontocrat for all the analysis! 
I feel like Peter on the great movie "While you were sleeping"... 
ADS changed a little their data, but they decreased it by 2K km2.

[ADS-NIPR-JAXA] ASI Extent.

July 24^th, 2018: 7,132,042 km², a double century drop of -203,392 km².

I am still waiting for the July 25th. data.
Hope it will appear before I go to sleep again!!! 
2018 is the eighth lowest on record, close to become the seventh.

[Juan C. García]

[ADS-NIPR-JAXA] ASI Extent.

July 25^th, 2018: 7,038,804 km², a drop of -93,238 km².
2018 is the eighth lowest on record.

[Wherestheice]

is it just me or does it look like we could be at 3rd lowest if it keeps melting at the current rate?

[DavidR]

is it just me or does it look like we could be at 3rd lowest if it keeps melting at the current rate?

Third is one of the most unlikely final ranks of all as it is only 48K above 2nd.  However if we continue to average 100K a day for the next  8 days we will be in third position. That would be extremely unusual at this time of the year. 

Fifth is even less likely  as it is only  12K above 4th. Based on the range of losses from this date we could end anywhere between 3.50 M km^2 (2nd) and 5.1 M km^2 (12th).   Based on losses in the last 10 years the highest position would be 10th with a final value of 4.82 M km^2.

But  we could move outside the range with a bigger or smaller decline.

[oren]

Looking at regional AMSR2 extent (the most accurate thanks to its high grid resolution) I have a few thoughts:
* Kara, Hudson and Baffin have about 300k remaining between them. This "easy ice" repository will soon run out.
* The big laggards Beaufort and ESS have joined the ranks at last, but will they keep up?
* The huge question mark for August is the CAB.

[seaice.de]

Looking at regional AMSR2 extent (the most accurate thanks to its high grid resolution) I have a few thoughts:
* Kara, Hudson and Baffin have about 300k remaining between them. This "easy ice" repository will soon run out.
* The big laggards Beaufort and ESS have joined the ranks at last, but will they keep up?
* The huge question mark for August is the CAB.

Just for the case if you want to create your own graphs:
https://twitter.com/seaice_de/status/1022411701864591360

[stjuuv]

Yes 15% is arbitrary. Yes, #F18 uses old technology. But consistency in the data record for 39 years is a prize not lightly discarded.

Maybe a stupid question, but is there any raw data available that could be used to recalculate previous years with a, say, 25% or 30% concentration cutoff?

It would be great to have a load of new satellites up there armed with the latest Gizmos, loads of buoys wandering around the Arctic looking at the ice and water temperatures etc etc.   Much more accurate data - yes, comparison with previous years - no.

With a lot of people saying that the arctic has gone through a permanent change, how long until having comparisons with 1980s will become irrelevant except for the uninformed general public? Of course, now that the supposed change has already arrived, it's a bit too late, but the early 2000s might have been a great time to make changes and still capture the relevant years with the new methods and only losing comparisons with the years that will become irrelevant anyway. With more changes to follow, and the 15% concentration likely becoming even less useful in 10 or 20 years, wouldn't making reasonable changes as soon as possible be the best option?

[oren]

I don't think the 15% threshold makes that much of a difference. You can always check out the area graphs, where the 15% effect is quite negligible. These graphs mostly have good correlation with extent graphs (esp. for the hi-res algorithms like JAXA and UH AMSR2), showing that the extent threshold is a robust one.

[magnamentis]

Quote from magnamentis

again i say that perhaps we or those in charge should consider a change of the 15% threshold.

15% is OK by me. NSIDC Satellite #F18 is OK by me.

Yes 15% is arbitrary. Yes, #F18 uses old technology. But consistency in the data record for 39 years is a prize not lightly discarded.

But that does not stop me looking at area as a better idea of what ice cover there is.

And we also have PIOMAS volume to give an indication of the volume of ice remaining - which in the end is the metric that will (and does) matter most.

It would be great to have a load of new satellites up there armed with the latest Gizmos, loads of buoys wandering around the Arctic looking at the ice and water temperatures etc etc.   Much more accurate data - yes, comparison with previous years - no.

So looking at the current situation of sea ice requires more than one simple metric. But for the general public - it is all about NSIDC Sea Ice Extent as it is has been defined since 1979. Change the definition of the measure and a forest of trolls will muddy the water (ice) for years.

good reply and i'm aware that my idea about altering that threshold to something higher than 15% is not shared by many here which is no big thingy.

while i believe it would provide some improvement and reduce discussions due to discrepancies and uncertainties,  there are definitely guys here who know a lot more about sea ice and related stuff so at the end of the day can live with it too.

thanks for sharing your views, a well worth to consider and/or valuable contribution as always

have a good day

[magnamentis]

Maybe a stupid question, but is there any raw data available that could be used to recalculate previous years with a, say, 25% or 30% concentration cutoff?

apparently you're got my key-point which perhaps i was not able to convey sufficiently well.

from my limited sources and know how i came to the conclusion that a 30% thershold would NOT alter the history much but would represent the present much better.

i'm aware of the consistency argument and i agree in FULL but i tried to guide the eyes of the pros to the assumption that due to the fact that in earlier years only a fraction of the total extent was prone to the threshold and hence the numbers would change only little for the past, now most of the total extent is prone to that threshold, hence the current staten is poorly represented.

this idea of mine was to improve the present data usability without altering the past data in a way that consistency would get lost.

[Ned W]

is it just me or does it look like we could be at 3rd lowest if it keeps melting at the current rate?

Third is one of the most unlikely final ranks of all as it is only 48K above 2nd.  However if we continue to average 100K a day for the next  8 days we will be in third position. That would be extremely unusual at this time of the year. 

Fifth is even less likely  as it is only  12K above 4th. Based on the range of losses from this date we could end anywhere between 3.50 M km^2 (2nd) and 5.1 M km^2 (12th).   Based on losses in the last 10 years the highest position would be 10th with a final value of 4.82 M km^2.

But  we could move outside the range with a bigger or smaller decline.

Yes, because some of the "ranks" are very narrow, the pattern of probabilities is rather counter-intuitive.  Courtesy of the predict-o-matic, here are the probabilities of the various ranks.  Note that 2nd, 4th, and 6th are all over 15% probability, while 1st, 3rd, 5th, and 7th are all below 5%.  Apparently this year the Arctic doesn't like odd numbers.   



Meanwhile, here are the probabilities of the (non-overlapping) bins.  It's much more of a normal (in both senses) distribution:



Best estimate as of today is 4.41 million km2

[stjuuv]

this idea of mine was to improve the present data usability without altering the past data in a way that consistency would get lost.

Wouldn't consistency in this case mean that the extent numbers should use the same threshold over all available years? If you start measuring extent with 30% concentration from some point and keep referencing past years with 15% concentration extent, this would be inconsistent in any case, even if the past years "would change only little".

In my view, having consistency would mean going back, using the raw data and calculating all the previous extent data using a 30% concentration cutoff as well, keeping the data consistent over all years. Which is why I was wondering about the possibility of doing that.

[Ned W]

In my view, having consistency would mean going back, using the raw data and calculating all the previous extent data using a 30% concentration cutoff as well, keeping the data consistent over all years. Which is why I was wondering about the possibility of doing that.

Sure, it's possible.  I'm doing this right now, just out of curiosity.  As input I'm using the NSIDC monthly ice concentration maps (1978-2017).  I'm throwing out the land, the pole hole, etc., reprojecting them to equal-area grids, and counting pixels with at least 15% concentration, then counting pixels with at least 30% concentration.

The processor is inefficient and there are > 470 months.  It will take a while, and I need to go to a meeting.  When I get back I'll see if it's finished and post any results.  If someone else beats me to this, well, that's life...

[magnamentis]

In my view, having consistency would mean going back, using the raw data and calculating all the previous extent data using a 30% concentration cutoff as well, keeping the data consistent over all years. Which is why I was wondering about the possibility of doing that.

Sure, it's possible.  I'm doing this right now, just out of curiosity.  As input I'm using the NSIDC monthly ice concentration maps (1978-2017).  I'm throwing out the land, the pole hole, etc., reprojecting them to equal-area grids, and counting pixels with at least 15% concentration, then counting pixels with at least 30% concentration.

The processor is inefficient and there are > 470 months.  It will take a while, and I need to go to a meeting.  When I get back I'll see if it's finished and post any results.  If someone else beats me to this, well, that's life...

you do what i was hoping for, unfortunately i lack the skills and the resources so my suggestion was that someone would sooner or later come up with a calculation to proof the idea either useless or useful or neutral

 

[Neven]

My guess is it won't change much in the standings. Or not as we expect. DMI used to have a 30% threshold, but they have 15% now too.

Remember, the objection against extent is that everything over 15% gets counted as 100%, everything below it is counted as 0% (same with area). If you raise the bar to 30%, there's this segment between 15 and 30% that is now counted as 0% too, whereas area counts it as 15-30%. So, basically you're letting sea ice disappear to artificially lower extent.

The strength of both measures (with their weaknesses) is their combination.

[gerontocrat]

JAXA ARCTIC EXTENT 7,038,804 km2（July 25, 2018）

Better late than never?

Just to add to Juan's post....
One cannot deny that an extent loss of 1.1 million km2  in 9 days remains impressive.
As a result :-
- Extent is down to just 98k km2  (1.4%) above the 2010's average extent on this date,
- and only 95k (1.4%) above 2017,
- Extent loss to date is now 480k km2 (6.6 %) below the 2008-2017 average, with 73.6 % of the average melting season done.

Resulting minimum from average remaining melt is down to 4.41 million km2, (excluding 2012 from the average 4.50 million km2). Range of results from last ten years remaining melt is 3.60 to 4.85 million km2.

9 days losses above average in a row amounting to 1.1 million km2, that 2017 feeling fades yet more but lingers on- extent losses are not yet enough to catch up on the slow melt to date. There is, on average, just 26% (49 days) of further extent loss to go. Could the melting season last a bit longer - Yes.  On the other hand, could extent loss sharply reduce? Yes.

A September minimum down one bin in the range of 4.50 to 5.00 million km2, and perhaps yet another bin lower, now looks increasingly the most likely outcome. But that is still approaching 0.5 million km2 above 2nd place.

The season surprises again.

[magnamentis]

My guess is it won't change much in the standings. Or not as we expect. DMI used to have a 30% threshold, but they have 15% now too.

Remember, the objection against extent is that everything over 15% gets counted as 100%, everything below it is counted as 0% (same with area). If you raise the bar to 30%, there's this segment between 15 and 30% that is now counted as 0% too, whereas area counts it as 15-30%. So, basically you're letting sea ice disappear to artificially lower extent.

The strength of both measures (with their weaknesses) is their combination.

of course you have a valid point/opinion only that i think 25% i.e. is closer to zero than to 100% hence if already not accurate at least it's less flawed but then i'm curious as to what the results will be of the calculation in progress.

after all the subject itself would not even be that important if it were not for the extreme room the current data leave for speculations and if that would be more of a middle way there would be a few extremes less to guess/assume/discuss.

thanks for your take on this which you made clear several times and it's always a good thing to compare different assessments to perhaps widen the range of acceptance among people.

[Tor Bejnar]

Warning: armchair 'scientist' at play:
I've been musing on how to make forecasts using 'only' concentration data by looking at the percentage of 95%+ concentration ice on (pick a date - July 19) that melted by mid-September (each year) and the percentage of 90-95% concentration ice on July 19 that melted by mid-September, etc.  Maybe even qualify further, like percentage of 95%+ concentration ice on July 19 that melted by mid-September north (and then south) of 80N [or finer geographic locations, e.g. Neven's Pacific Sector, etc. ...]

For instance:  Most spots on the Arctic with 95% concentration on July 19 have more than 15% concentration on September 19, but in 2012 (at least) nearly all spots with 95% conc. (or any lesser conc.!) on July 19 south of 80N in the Pacific Sector had less than 15% concentration on September 19.  (My method won't care about actual spots, just pixel counts within an area at each concentration percentage.)

Do any of these 'pigeon holes' show trends or consistencies?

[Neven]

of course you have a valid point/opinion only that i think 25% i.e. is closer to zero than to 100% hence if already not accurate at least it's less flawed but then i'm curious as to what the results will be of the calculation in progress.

What if it's actually 100% covered with ice, but 75% of this ice is itself covered with melt ponds (which the satellite sensors mistake for open water)? Remember, this is the main reason that sea ice extent was introduced as a measure.

Both measures have their flaws, but of course, it's true that at the end of the melting season the melt ponds have drained/frozen over, and so area would be the more accurate measure. When media report only the September extent number, the public isn't getting all the information, but at the same time, if the ice pack is so riddled with holes that area is actually lowest on record, this will be reported.

But that's September. Right now, it's important to look at both extent and area (and compactness, by dividing area with extent). Speaking of which...

Two slow area days, while extent does two centuries, and so compactness goes up again, putting 2018 in the upper half of the pack:

[gerontocrat]

NSIDC Total Area as at 25 July (5 day trailing average =  5,077,154  km2
This is at 187 k above the 2010-2017 average

Total Area loss 67 K, Central Seas 38k, Periphery loss 17 k, Other Seas loss 12 k  
Analysis of individual seas.

Pacific Side
- The Bering Sea area is now 1 k,
- Chukchi Sea loss 11 k,
Atlantic Side
- Baffin Sea loss 10 k,
- Greenland Sea loss 7 k,
- Barents Sea loss 0 k -  area is 2 k, i.e. melting season over,
- The Kara Sea area loss 7 k, Area now just 67 k.
- The Laptev Sea area loss 3 k.
CAB
- Beaufort Sea loss 15 k,
- The Central Arctic Sea GAIN 13 k,
- The Canadian Archipelago loss 5 k,
- East Siberian Sea loss 10 k .
Other seas
- St Lawrence area at 1 k,
- Hudson Bay area loss 10 k, Area now 100k
- The Okhotsk Sea area is 4 k,down 1k.

Area loss slightly below average. Extent losses are catching up with area losses again, both NSIDC and JAXA daily measures especially.

Qu1. The question is - will this be reflected in higher Area losses over the following days, or is it simply Extent catch-up as it was on this day.
Qu2. Will the Kara Sea completely melt out? (Even in 2012 - still the minimum year, it did not).

[Richard Rathbone]

Warning: armchair 'scientist' at play:
I've been musing on how to make forecasts using 'only' concentration data by looking at the percentage of 95%+ concentration ice on (pick a date - July 19) that melted by mid-September (each year) and the percentage of 90-95% concentration ice on July 19 that melted by mid-September, etc.  Maybe even qualify further, like percentage of 95%+ concentration ice on July 19 that melted by mid-September north (and then south) of 80N [or finer geographic locations, e.g. Neven's Pacific Sector, etc. ...]

For instance:  Most spots on the Arctic with 95% concentration on July 19 have more than 15% concentration on September 19, but in 2012 (at least) nearly all spots with 95% conc. (or any lesser conc.!) on July 19 south of 80N in the Pacific Sector had less than 15% concentration on September 19.  (My method won't care about actual spots, just pixel counts within an area at each concentration percentage.)

Do any of these 'pigeon holes' show trends or consistencies?

This is the way the Slater projection works. Its quite decent for September predictions from around now.

[Juan C. García]

[ADS-NIPR-JAXA] ASI Extent.

July 26^th, 2018: 6,978,040 km², a drop of -60,764 km².
2018 is the eighth lowest on record.

[gerontocrat]

JAXA ARCTIC EXTENT 6,978,040 km2（July 26, 2018）

Just to add to Juan's post....
- An extent loss of 61k is about 20k below the average for this date,
- Extent is at  111 km2  (1.6%) above the 2010's average extent on this date,
- and 119k (1.7%) above 2017,
- Extent loss to date is now 500k km2 (6.7 %) below the 2008-2017 average, with 74.4 % of the average melting season done.

Resulting minimum from average remaining melt is down to 4.43 million km2, (excluding 2012 from the average 4.5 million km2). Range of results from last ten years remaining melt is 3.64 to 4.88 million km2.

There a discussion some time ago about how extent losses come in waves - a few days above and then a few days below average. Are we in that below average spell now? So  that 2017 feeling fades yet but lingers on- extent losses are not yet enough to catch up on the slow melt to date. There is, on average, just 25.6% (48 days) of further extent loss to go. Could the melting season last a bit longer than that - Yes.  On the other hand, could extent loss sharply reduce? Yes.

A September minimum in the range of 4.50 to 5.00 million km2, and perhaps yet another bin lower, now looks increasingly the most likely outcome. But that is still approaching 0.5 million km2 above 2nd place.

[Ned W]

OK, I've finished reprocessing all the monthly NSIDC ice concentration data, to see what the effect would be of changing from a 15% threshold for extent to a 30% threshold. 

My guess is it won't change much in the standings. Or not as we expect. DMI used to have a 30% threshold, but they have 15% now too.

Yup, it basically changes ... nothing.  Except it makes the extent numbers artificially lower.  But there's no trend, no difference in the effect in recent years vs earlier years:



Needless to say it's not statistically significant. 

Remember, the objection against extent is that everything over 15% gets counted as 100%, everything below it is counted as 0% (same with area). If you raise the bar to 30%, there's this segment between 15 and 30% that is now counted as 0% too, whereas area counts it as 15-30%. So, basically you're letting sea ice disappear to artificially lower extent.

The strength of both measures (with their weaknesses) is their combination.

Yes, exactly.  Switching to a 30% threshold would delete 0.72 million km2 of extent on average, so we'd hit the "ice-free Arctic" threshold very slightly earlier.  But it would be an artificial distinction.

[Stephan]

OK, I've finished reprocessing all the monthly NSIDC ice concentration data, to see what the effect would be of changing from a 15% threshold for extent to a 30% threshold. 
[...]
Yes, exactly.  Switching to a 30% threshold would delete 0.72 million km2 of extent on average, so we'd hit the "ice-free Arctic" threshold very slightly earlier.  But it would be an artificial distinction.

Thanks for this work. It seems like the difference between 15% and 30% extent is season-depending as mostly five peaks appear in almost regular manner in every five-year period.

[Ned W]

OK, I've finished reprocessing all the monthly NSIDC ice concentration data, to see what the effect would be of changing from a 15% threshold for extent to a 30% threshold. 
[...]
Yes, exactly.  Switching to a 30% threshold would delete 0.72 million km2 of extent on average, so we'd hit the "ice-free Arctic" threshold very slightly earlier.  But it would be an artificial distinction.

Thanks for this work. It seems like the difference between 15% and 30% extent is season-depending as mostly five peaks appear in almost regular manner in every five-year period.

Indeed, you're right.  The difference is mostly stationary across months, except for July (when it's slightly larger, on average) and September (when it's slightly smaller on average).  Probably due to the phenology of melt ponds? 

[gerontocrat]

NSIDC Total Area as at 25 July (5 day trailing average =  4,997,527  km2
This is at 159 k above the 2010-2017 average

Total Area loss 80 K, Central Seas loss 56k, Periphery loss 14 k, Other Seas loss 10 k  
Analysis of individual seas.

Pacific Side
- The Bering Sea area is now 1 k,
- Chukchi Sea loss 11 k, area is now below the 2010's average,
Atlantic Side
- Baffin Sea loss 9 k,
- Greenland Sea loss 6 k,
- Barents Sea loss 0 k -  area is 2 k, i.e. melting season over,
- The Kara Sea area loss 2 k, Area now just 67 k.
- The Laptev Sea area loss 4 k.
CAB
- Beaufort Sea loss 13 k,
- The Central Arctic Sea loss 9 k,
- The Canadian Archipelago loss 7 k,
- East Siberian Sea loss 10 k .
Other seas
- St Lawrence area at 1 k,
- Hudson Bay area loss 10 k, Area now 90k
- The Okhotsk Sea area is 4 k.

On average, this is when daily area loss declines sharply - on this day to 50k. Thus an area loss of 80k is 30k above average. Extent losses are still catching up with area losses again, both NSIDC and JAXA daily measures especially.

Qu1. The question is - will above average area losses continue over the following days, or is it simply Extent catching up as it was on this day (extent loss was 120k).
Qu2. How low will the Greenland Sea go ? Fram export likely to be minimal if 2017 repeated (per Wipneus graph on PIOMAS).

[LDorey]

Actually I think that's very interesting, when I look at that it looks like the gap between the two is increasing, esp on "Bad" years, aka 2007, 2012, those big spikes are showing there's a lot more Ice in between 15% and 30% than there was, esp on iffy years, how far along in 2018 did you process? with the current state of the ice I would have expected a much larger difference...Personally I don't think 30% tells us much more or less than 15%, they are just arbitrary lines in the sand, but the year to year difference between the two is interesting, Can you for instance spit out the yearly difference by month in a graph.

Ie a plotted line for difference between 15% and 30% in each month from 1980 to now? see if the difference in the trend for winter months vs summer months? if it's not a total pain I'd also be interested in seeing the trend for the difference between 15% and 90%, ie how much more "slush" there is kicking around year to year...and if a spike shows up on bad years before the actually cliffs show up

Liam

OK, I've finished reprocessing all the monthly NSIDC ice concentration data, to see what the effect would be of changing from a 15% threshold for extent to a 30% threshold. 

My guess is it won't change much in the standings. Or not as we expect. DMI used to have a 30% threshold, but they have 15% now too.

Yup, it basically changes ... nothing.  Except it makes the extent numbers artificially lower.  But there's no trend, no difference in the effect in recent years vs earlier years:



Needless to say it's not statistically significant. 

Remember, the objection against extent is that everything over 15% gets counted as 100%, everything below it is counted as 0% (same with area). If you raise the bar to 30%, there's this segment between 15 and 30% that is now counted as 0% too, whereas area counts it as 15-30%. So, basically you're letting sea ice disappear to artificially lower extent.

The strength of both measures (with their weaknesses) is their combination.

Yes, exactly.  Switching to a 30% threshold would delete 0.72 million km2 of extent on average, so we'd hit the "ice-free Arctic" threshold very slightly earlier.  But it would be an artificial distinction.

[stjuuv]

Yup, it basically changes ... nothing.  Except it makes the extent numbers artificially lower.  But there's no trend, no difference in the effect in recent years vs earlier years:

Would you mind sharing the numerical data for this as well? It indeed looks like there is no long term trend, but I'm sure others can find interesting ways to present this data as well. Thanks in any case, this looks like it's settled the question quite well.

[Juan C. García]

[ADS-NIPR-JAXA] ASI Extent.

July 27^th, 2018: 6,878,603 km², almost a century drop of -99,437 km².
2018 is the seventh lowest on record.

[gerontocrat]

JAXA ARCTIC EXTENT 6,878,603 km2（July 27, 2018）

Just to add to Juan's post....
- An extent loss of 99k is about 20k above the average for this date,
- Extent is at  100 km2  (1.4%) above the 2010's average extent on this date,
- and 108k (1.6%) above 2017,
- Extent loss to date is now 490 k km2 (6.4 %) below the 2008-2017 average, with 75.2 % of the average melting season done.

Resulting minimum from average remaining melt is down to 4.41 million km2, (excluding 2012 from the average 4.49 million km2). Range of results from last ten years remaining melt is 3.69 to 4.86 million km2.

That 2017 feeling fades yet but lingers on- extent losses are not yet enough to catch up on the slow melt to date. There is, on average, just 25 % (47 days) of further extent loss to go. Could the melting season last a bit longer than that - Yes.  On the other hand, could extent loss sharply reduce? Yes.

A September minimum in the range of 4.50 to 5.00 million km2, and perhaps yet another bin lower, now looks increasingly the most likely outcome. But that is still approaching 0.5 million km2 above 2nd place.

[Ned W]

[snip]

[gerontocrat]

NSIDC Total Area as at 27 July (5 day trailing average =  4,918,862  km2
This is at 139 k above the 2010-2017 average

Total Area loss 79 K, Central Seas loss 60k, Periphery loss 12 k, Other Seas loss 7 k  
Analysis of individual seas.

Pacific Side
- The Bering Sea area is now 1 k,
- Chukchi Sea loss 9 k, area is now below the 2010's average,
Atlantic Side
- Baffin Sea loss 7 k,
- Greenland Sea loss 6 k,
- Barents Sea gain 1 k -  area is 3 k, i.e. melting season over,
- The Kara Sea area loss 4 k, Area now just 61 k.
- The Laptev Sea area loss 6 k.
CAB
- Beaufort Sea loss 7 k,
- The Central Arctic Sea loss 26 k,
- The Canadian Archipelago loss 5 k,
- East Siberian Sea loss 4 k .
Other seas
- St Lawrence area at 2 k, gained 1k!
- Hudson Bay area loss 8 k, Area now 82k
- The Okhotsk Sea area is 4 k.

On average, this is when daily area loss declines sharply - on this day to 50k. Thus an area loss of 80k is 30k above average. Extent losses are still catching up with area losses again, both NSIDC and JAXA daily measures.

Qu1. The question is - will above average area losses continue over the following days?
Qu2. How long before the St. Lawrence finally  gets to zero ice? The sea where the ice apparently persists beyond reason.