IJIS

[Bill Fothergill]

For a long time Vidaloo has been touting the use of averages and their extrapolation to predict a year-round ice free arctic a few years down the line, a very wrong approach when averaging numbers that have very different underlying physics, such as winter ice cover and summer ice cover. Looking at average July-September ice cover could give more insights.

Because he has a commercial interest on doing it

Really? What is it?

Yep, as with oren, I would like to know more background.

I eagerly await...

[icy voyeur]

Looks OK.  This is an daily extent change anomaly map of all seasons from 1989 to current.  The first thing I did is fill in any data gaps with simple linear interpolation.  Then I expressed each day's extent as the percent change from the previous day's extent.  I then created a "baseline" by averaging the daily change for all years between 1989 and 2005 and smoothed that with a 5-day running average. 

Wouldn't it make more sense to apply the 5 day smooth first and then calculate the % change from prior day?

[Espen]

IJIS:

10,573,160 km²（December 10, 2016）up 207,700 km² from previous and lowest measured for the date.

[Pmt111500]

IJIS:

10,573,160 km²（December 10, 2016）up 207,700 km² from previous and lowest measured for the date.

Well that was fast.

[dnem]

Looks OK.  This is an daily extent change anomaly map of all seasons from 1989 to current.  The first thing I did is fill in any data gaps with simple linear interpolation.  Then I expressed each day's extent as the percent change from the previous day's extent.  I then created a "baseline" by averaging the daily change for all years between 1989 and 2005 and smoothed that with a 5-day running average. 

Wouldn't it make more sense to apply the 5 day smooth first and then calculate the % change from prior day?

Sure, it might well.  But I doubt it would change the baseline against which the % change was calculated very much.  I also played around with using just plain old deviations from the average km2 extent change but thought that expressing it as percent change to the current extent better highlights the changes that have gone on in recent years.  For example today's double century gain on top of the all time current low extent makes it a much bigger % increase than other late fall/winter double century gains in the data set.  I started playing around with the data to get at the question of whether or not extent gains and loses have gotten noisier in recent years.  Is the top beginning to wobble?  Yes, I'd say.

[Jim Pettit]

IJIS:

10,573,160 km²（December 10, 2016）up 207,700 km² from previous and lowest measured for the date.

Even with that very large increase, 2016 is blowing away all other years in terms of annual average. Nothing else is even close:

[jdallen]

IJIS:

10,573,160 km²（December 10, 2016）up 207,700 km² from previous and lowest measured for the date.

Well that was fast.

I'm thinking a bunch of that increase was Hudson.  It should be getting the effects of the breakout that's about to drop into the Mississippi valley & Great lakes in a day or so.

[bbr2314]

IJIS:

10,573,160 km²（December 10, 2016）up 207,700 km² from previous and lowest measured for the date.

Well that was fast.

I'm thinking a bunch of that increase was Hudson.  It should be getting the effects of the breakout that's about to drop into the Mississippi valley & Great lakes in a day or so.

Yes, you can see on NASA, Hudson is now gaining significant ice every day.Should be mostly covered shortly. But after that the gains may come to a halt as continued cyclonic activity limits ice formation (or results in outright loss) along both Atlantic and Pacific fronts (with exception of additional growth in Okhotsk + Baffin Bay).

[icy voyeur]

Looks OK.  This is an daily extent change anomaly map of all seasons from 1989 to current.  The first thing I did is fill in any data gaps with simple linear interpolation.  Then I expressed each day's extent as the percent change from the previous day's extent.  I then created a "baseline" by averaging the daily change for all years between 1989 and 2005 and smoothed that with a 5-day running average. 

Wouldn't it make more sense to apply the 5 day smooth first and then calculate the % change from prior day?

Sure, it might well.  But I doubt it would change the baseline against which the % change was calculated very much.  I also played around with using just plain old deviations from the average km2 extent change but thought that expressing it as percent change to the current extent better highlights the changes that have gone on in recent years.  For example today's double century gain on top of the all time current low extent makes it a much bigger % increase than other late fall/winter double century gains in the data set.  I started playing around with the data to get at the question of whether or not extent gains and loses have gotten noisier in recent years.  Is the top beginning to wobble?  Yes, I'd say.

I have some background in data analysis. From my perspective, what you did was frankly wrong. It's foundational mathematics. Smoothing a derived metric rather than smoothing the original measurements is frankly wrong. Ask yourself "what is the point of smoothing?" and the underlying question, "what is the nature of noise in the data?".

The measures of ice extent include noise, both as measurement artifacts and, depending on ones perspective, weather. Smoothing by averaging minimizes that noise. If you smooth a derived metric, like daily percent change, you are effectively smoothing the derivative. It doesn't commute.

It's trivially simple to manipulate data within the equivalent of a spreadsheet but you do need to ask yourself what is it you are actually doing. And smoothing, in particular, should not be applied capriciously. Maybe I'm missing something but if so I'd expect you to be able to elucidate a reason why smoothing the derivative rather than the original data made sense for this instance when it is generally known to be a flawed procedure.

[Espen]

IJIS:

10,751,338 km²（December 11, 2016）up 178,178 km² from previous and lowest measured for the date.

[dnem]

 
[/quote]

Wouldn't it make more sense to apply the 5 day smooth first and then calculate the % change from prior day?
[/quote]

I have some background in data analysis. From my perspective, what you did was frankly wrong. [/quote]

Totally open to suggestions.  I tried a bunch of ways, smoothed and not.  I was simply looking for a different way to visualize the data synoptically as I was getting sick of looking at the piled up line graphs.  I find them hard to get anything from when there are so many lines.  I wanted a way to quickly and visually see when during the the melt seasons the ice was melting faster/slower than some sense of "normal."  I used the five day smooth from 1989 to 2005 as the normal.  Honestly it doesn't change that much whether you look at deviations from the raw gains/losses for the day or from the % change against the day before, smoothed or unsmoothed.  It's just a visualization, not any sort of test.

[Cato]

Quick increase of extension, as per the forecast. In the next few days the recurring synoptic pattern will be reinstated, with consequent massive advection of warm air from both Atlantic and Pacific into the CAB.

In spite of this, extension should keep growing significantly, for the contribution of Hudson, Ohotsk and the smaller contributions from Davis Streit and Kara (east of Novaya Zemlja).

I confirm my view that by the 20th of December, 2016 might no longer be at the top...

[magnamentis]

I confirm my view that by the 20th of December, 2016 might no longer be at the top...

i hold agains it again as i did for 15th of december similar forecast. one would have to look at the numbers and not only at a graph where things appear to be closer than they really are. there will be no recovery of the needed kind this year IMO while a 40+ days forecast is a bit difficult to make, not more than a gut feeling that might be wrong. it's just that the remaining open waters that were already frozen in earlier years are still to warm to just re-freeze in bulk so to say and then the air temps as well are not in favour of a "passing lane" refreeze IMO. let's see, always interesting to compare reality with early impressions, at least for those who are interested to learn :-)

game on

cheers

[Tigertown]

Sometimes you just need another perspective

[Espen]

IJIS:

10,853,038 km²（December 12, 2016）up 101,700 km² from previous and lowest measured for the date.

[budmantis]

Tigertown's perspective certainly adds contrast to the difference between this year's results versus the previous lowest extent, but the difference remains substantial regardless of the view.

[Tigertown]

I kind of expect the extent numbers to catch up with previous years. It is thickness and volume that won't be able to do so.

[budmantis]

I kind of expect the extent numbers to catch up with previous years. It is thickness and volume that won't be able to do so.

You might be right TT, but as you say, the volume will not catch up. I normally don't follow the arctic happenings this time of year, but I have to say with the record low ice buildup, I've been keeping track.

[magnamentis]

I kind of expect the extent numbers to catch up with previous years. It is thickness and volume that won't be able to do so.

as i mentioned this is an open race without certain outcome and for the fun of it ( meaning it ) the perspectives can change on a daily base. a few more of those flattening bends and it's one and a few more of those recent up-turns it's the other. looking at the weather forecast on both sides i still tend towards the flatting while it really only takes around 0.7k per day to reach the previous lowest till the end of the month.

again for those interpreting more into this, it's an open race and for the fun of it to reason a bit back and forth. no serious forecast can be made IMO

cheers

[Espen]

IJIS:

10,895,637 km²（December 13, 2016）up 42,599 km² from previous and lowest measured for the date.

[Espen]

IJIS:

11,010,469 km²（December 14, 2016）up 114,832 km² from previous and lowest measured for the date.

[oren]

Half a million in five days to finally lose the title, doable though not certain.

[Tigertown]

Half a million in five days to finally lose the title, doable though not certain.

Should intersect 2015 and possibly 2012 around Dec. 22nd, give or take a day.

[DavidR]

IJIS extent has now been lowest on record for 184 days this year meaning that more than half the days this year will be lowest on record. 

The variation from the 2002-2015 figures reached 5.04 std deviations on 20th Nov, and averaged 1.88 std devs for the year  compared with the previous record variation of of 2.40 std devs and average fo the record of 1.69 std devs.

[Tony Mcleod]

Half a million in five days to finally lose the title, doable though not certain.

Should intersect 2015 and possibly 2012 around Dec. 22nd, give or take a day.

Seems to me it'll be lucky to follow the same shape curve. I don't expect any intersection.

[magnamentis]

Half a million in five days to finally lose the title, doable though not certain.

Should intersect 2015 and possibly 2012 around Dec. 22nd, give or take a day.

i'm watching LOL U know my opinion considering the weather and winds but oren put it as good as one can at the moment. "spannung pur"

[Cato]

As per the forecasts, Hudson is freezing quickly and there's 5-6 days left of very low temperatures to further increase extension up there. Very cold conditions forecast for the Okhotsk and, in a few days, for Chukchi as well.

There's more than 1 million kmq left to be frozen between Hudson, Chuckchi and coastal areas of Okhotsk and Newfoundland where conditions will be generally cold for the next one week. Part of this increase will be offset by very warm conditions in Barents, where I expect also some compaction after the strong winds from the south.

Overall I keep expecting that 2016 will probably close the gap in about one week time (I'm always overoptimistic, I know...)

[Tigertown]

In 2015, the extent hit 11.83 million km² on 12-22. It is very possible to make up the difference in where we are now and that number in 7 or 8 days.

[Espen]

IJIS:

11,093,023 km²（December 15, 2016）up 82,554 km² from previous and lowest measured for the date.

[Jim Pettit]

As noted above, IJIS extent has been the lowest on record 185 of this year's 349 days, including the last 57 consecutive days. That's about 53% of the total. It was also 2nd lowest on an additional 56 days, and 3rd lowest on another 67. That means 2016 has spent a total of 308 days in one of the three lowest positions, for 88% of the total.

Despite the December refreeze, 2016 has by a wide margin seen the least Northern sea ice extent on record, as exhibited by this graph:



Interesting stat: even were extent to grow by 750k every day between now and the end of the year, 2016 would still end up with the lowest annual average extent.

November saw the lowest average for that month on record, and December is a solid lock. So even if IJIS's daily numbers catch up to previous years for a shorter or longer time span, the damage has already been done. For remember: not all ice is created equal.

Also: the sun starts s-l-o-w-l-y moving back north in just one week.

[Shared Humanity]

And that sun is going to find a lot of thin ice and open waters on both the Pacific and Atlantic sides.

[Tigertown]

You won't get any argument from me on that. Next year will be phenomenal, no doubt.

[Lord M Vader]

IJIS is back and we are up to 11,25 Mn km². Espen will give you the exact numbers later today

[Espen]

IJIS:

11,251,450 km²（December 17, 2016）and lowest measured for the date.

[oren]

Average gain slightly less than 80k for the last two days. The "target" is still within reach but far from guaranteed.

[magnamentis]

Average loss slightly less than 80k for the last two days. The "target" is still within reach but far from guaranteed.

even though we all understand the meaning that would make a "gain" if you want to edit that :-)

yeah this time it's a flattening and now we need an upturn to get back on "collision course" the lines meant to intersect or touch of course LOL

[oren]

Thanks magnamentis, corrected...

[pauldry600]

11.29m is where we are at noe.

Only 40k increase.

2010 slowed about now for a few days.

We are 200 to 300k off that

My math suck but we may stay just below and just breach 12m by Dec 31.

[Espen]

IJIS:

11,295,762 km²（December 18, 2016）up 44,312 km² from previous and lowest measured for the date.

[oren]

Looking at the regional AMSR2 extent charts, it seems the recent slowdown was caused by the Chukchi and Bering sporting a decrease, quite possibly into record territory for the date, and this despite Kara and especially Hudson exhibiting very fast growth which brought them almost to a "full" level.

[jdallen]

Looking at the regional AMSR2 extent charts, it seems the recent slowdown was caused by the Chukchi and Bering sporting a decrease, quite possibly into record territory for the date, and this despite Kara and especially Hudson exhibiting very fast growth which brought them almost to a "full" level.

I believe the Barents is retreating some with the weather as well.

[Espen]

IJIS:

11,360,060 km²（December 19, 2016）up 64,298 km² from previous and lowest measured for the date.

[Cato]

More or less in accordance with the forecast, freezing over hudson has been very quick, though partially offset by compaction in Barents. The gap has not been closed due to the slower than expected (by me) freezing rate over Okhotsk.

Between now and the 31th of December, about 800,000 kmq are needed to close the gap. It's not a big deal as it makes slightly more than 70k per day. And there is potential for further freezing over Okhotsk, Davis streit and Chuckchi, where conditions will be very cold in the next one week. IMHO there is more than 1 million kmq to freeze in those areas, as conditions over barents will keep being extremely warm and favourable to further compaction.

Based on synoptic considerations only, I still tend to believe that 2016 will not be leading by the end of December.

[Espen]

IJIS:

11,447,549 km²（December 20, 2016）up 87,489 km² from previous and lowest measured for the date.

[Paddy]

Based on synoptic considerations only, I still tend to believe that 2016 will not be leading by the end of December.

I'm reckoning it'll be about the lowest, plus or minus a little. Temperatures are still pretty high for the time of year over the Arctic seas.

[Shared Humanity]

We may edge out of the lowest spot in the next day or two.

[Tigertown]

Funny thing is that it looks like the year will end close to where it started. On Jan. 1st 2016, the extent was 12.23 million km².

[gregcharles]

I also find it funny that while 2010 has (for now) the lowest record until December 31, it was beaten on January 1 (albeit 2011) by 2016. The two lines crossed on the only day that the graph can't show that they crossed. In a weird way, 2016 is now in a race for last with 2016.

[Espen]

IJIS:

11,491,398 km²（December 21, 2016）up 43,849 km² from previous and lowest measured for the date.

[oren]

Missed by an inch. 2016 just can't let go.