Support the Arctic Sea Ice Forum and Blog

Author Topic: IJIS  (Read 2014125 times)

Jim Hunt

  • Young ice
  • Posts: 4289
    • View Profile
    • The Arctic sea ice Great White Con
  • Liked: 274
  • Likes Given: 27
Re: IJIS
« Reply #4850 on: August 23, 2017, 03:28:08 PM »
Hmmm.  What would be better?

How about a PIOMAS predict-o-matic?

I'll eat my proverbial hat if that bottoms out in August!

P.S. I see Ned is ahead of me on the "Freezing Season" thread.
Reality is merely an illusion, albeit a very persistent one - Albert Einstein

Espen

  • Young ice
  • Posts: 3251
    • View Profile
  • Liked: 177
  • Likes Given: 0
Re: IJIS
« Reply #4851 on: August 23, 2017, 04:12:03 PM »
IJIS:

5,003,285 km2(August 22, 2017)down 44,622 km2 and 5th lowest measured for the date.
Have a ice day!

Ned W

  • Guest
Re: IJIS
« Reply #4852 on: August 24, 2017, 01:53:26 PM »
Extent for 8/23:  4943252 (drop of 60033; 5th lowest for this date).

2017 is now below 5 million km2 for the first time.

Live version of the predict-o-matic now says 4.27 (range 4.04-4.49), the same mean as yesterday but a slightly narrower confidence interval.  There is a 43% chance of 4th place and a 48% chance of 6th place.  Complete list of odds:

1st place:  0%
2nd place:  1%
3rd place:  2%
4th place: 43%
5th place:  4%
6th place:  48%
7th place:  2%
8th place or higher: 0%

Extent losses similar to those from 2007-2016 for the remaining days of the season would result in a 2017 minimum of 4.15 to 4.43, as shown below:


Ned W

  • Guest
Re: IJIS
« Reply #4853 on: August 24, 2017, 01:58:44 PM »
That chart shows why the probabilities of various ranks are so strange (40+ % chance of 4th place or 6th place, but only a 4% chance of 5th place). 

In order to achieve 5th place, 2017 would need to squeeze in between the two blue dots under the "re" in "previous years".   Those dots are the minimums from 2011 and 2015, and there's not much space between them!  Much easier to come in 4th (lower extent than 2015) or 6th (higher extent than 2011).

Bill Fothergill

  • Frazil ice
  • Posts: 278
    • View Profile
  • Liked: 0
  • Likes Given: 0
Re: IJIS
« Reply #4854 on: August 24, 2017, 02:19:59 PM »
That chart shows why the probabilities of various ranks are so strange (40+ % chance of 4th place or 6th place, but only a 4% chance of 5th place). 

In order to achieve 5th place, 2017 would need to squeeze in between the two blue dots under the "re" in "previous years".   Those dots are the minimums from 2011 and 2015, and there's not much space between them!  Much easier to come in 4th (lower extent than 2015) or 6th (higher extent than 2011).

Yep.
Under such circumstances, it can often be far better to avoid the use of equalities, and instead plump for "greater than, or equal to".

Hence one would talk about there being...
an X% chance of being 4th or higher,  (NB Higher, in this context, means a larger ordinal number.)
a Y% chance of being 5th or higher,
a Z% chance of being 6th or higher,

In the case you are describing, the probability X would be appreciably larger than Y, but Y would be only marginally greater than Z. (The trivial case is, of course, that there is always a 100% chance of the year finishing 1st or higher - using "higher" to mean the same as mentioned above.)

At first glance it can seem as if one is trying to dodge a difficult question, but, when one has that kind of spread in the respective probabilities, it can make the situation easier to understand. (At least for those prepared to invest in the effort of a modicum of thought.)

Espen

  • Young ice
  • Posts: 3251
    • View Profile
  • Liked: 177
  • Likes Given: 0
Re: IJIS
« Reply #4855 on: August 24, 2017, 04:13:59 PM »
IJIS:

4,943,252 km2(August 23, 2017)down 60,033 km2 and 5th lowest measured for the date.
Have a ice day!

Ned W

  • Guest
Re: IJIS
« Reply #4856 on: August 25, 2017, 12:55:29 PM »
Extent for 8/24:  4893557 (drop of 49695; 5th lowest for this date).

2017 is now only 9437 km2 above the 2014 minimum.

Live version of the predict-o-matic now says 4.28 (range 4.06-4.49).  There is a 41% chance of 4th place and a 51% chance of 6th place.  Complete list of odds:

1st place:  0%
2nd place:  1%
3rd place:  2%
4th place: 41%
5th place:  5%
6th place:  51%
7th place:  2%
8th place or higher: 0%

Extent losses similar to those from 2007-2016 for the remaining days of the season would result in a 2017 minimum of 4.17 to 4.44


Espen

  • Young ice
  • Posts: 3251
    • View Profile
  • Liked: 177
  • Likes Given: 0
Re: IJIS
« Reply #4857 on: August 25, 2017, 04:09:08 PM »
IJIS:

4,893,557 km2(August 24, 2017)down 49,695 km2 and 5th lowest measured for the date.
Have a ice day!

Shared Humanity

  • Young ice
  • Posts: 3980
    • View Profile
  • Liked: 430
  • Likes Given: 52
Re: IJIS
« Reply #4858 on: August 25, 2017, 05:48:29 PM »
Looking at this graph, unless we have a late melt season, SIE extent will do no worse than 4th. If the freeze season is delayed, SIE may be able to mimic the pale blue year on the graph. I am having trouble distinguishing the blues. Is this year 2007?

Lord M Vader

  • Nilas ice
  • Posts: 1263
    • View Profile
  • Liked: 26
  • Likes Given: 19
Re: IJIS
« Reply #4859 on: August 25, 2017, 06:26:18 PM »
SH: yup, should be 2007..

Ned W

  • Guest
Re: IJIS
« Reply #4860 on: August 25, 2017, 06:27:56 PM »
If you mean the blue line that reaches its minimum just above 4.0 in late September, yes, that's 2007.

It is hard to know how to talk about these ranks/places.  "Higher/lower" ranks, and "better/worse" places are ambiguous.  6 is a "higher" number than 5, but because we typically consider ranks close to 1 to be "good" (e.g., in races) we often think of 1 as the "highest" rank.  I struggle with this in every post here.

So I'm guessing that by "no worse than fourth" you mean that SIE will end up in 4th, 5th, or 6th place... meaning that 3rd, 2nd, or 1st would be "worse".  Which makes sense, from the ice's perspective....

Daniel B.

  • Grease ice
  • Posts: 659
    • View Profile
  • Liked: 8
  • Likes Given: 29
Re: IJIS
« Reply #4861 on: August 25, 2017, 06:55:22 PM »
Extent for 8/24:  4893557 (drop of 49695; 5th lowest for this date).

2017 is now only 9437 km2 above the 2014 minimum.

Live version of the predict-o-matic now says 4.28 (range 4.06-4.49).  There is a 41% chance of 4th place and a 51% chance of 6th place.  Complete list of odds:

1st place:  0%
2nd place:  1%
3rd place:  2%
4th place: 41%
5th place:  5%
6th place:  51%
7th place:  2%
8th place or higher: 0%

Extent losses similar to those from 2007-2016 for the remaining days of the season would result in a 2017 minimum of 4.17 to 4.44



That seems about right.  If melt picks up then 4th place is likely, otherwise 6th.

Shared Humanity

  • Young ice
  • Posts: 3980
    • View Profile
  • Liked: 430
  • Likes Given: 52
Re: IJIS
« Reply #4862 on: August 25, 2017, 08:21:46 PM »
If you mean the blue line that reaches its minimum just above 4.0 in late September, yes, that's 2007.

It is hard to know how to talk about these ranks/places.  "Higher/lower" ranks, and "better/worse" places are ambiguous.  6 is a "higher" number than 5, but because we typically consider ranks close to 1 to be "good" (e.g., in races) we often think of 1 as the "highest" rank.  I struggle with this in every post here.

So I'm guessing that by "no worse than fourth" you mean that SIE will end up in 4th, 5th, or 6th place... meaning that 3rd, 2nd, or 1st would be "worse".  Which makes sense, from the ice's perspective....

Yes, from the ice's perspective....more is better.

Dharma Rupa

  • Frazil ice
  • Posts: 493
    • View Profile
  • Liked: 55
  • Likes Given: 26
Re: IJIS
« Reply #4863 on: August 25, 2017, 10:19:39 PM »
If you mean the blue line that reaches its minimum just above 4.0 in late September, yes, that's 2007.

It is hard to know how to talk about these ranks/places.  "Higher/lower" ranks, and "better/worse" places are ambiguous.  6 is a "higher" number than 5, but because we typically consider ranks close to 1 to be "good" (e.g., in races) we often think of 1 as the "highest" rank.  I struggle with this in every post here.

So I'm guessing that by "no worse than fourth" you mean that SIE will end up in 4th, 5th, or 6th place... meaning that 3rd, 2nd, or 1st would be "worse".  Which makes sense, from the ice's perspective....

Yes, from the ice's perspective....more is better.
For someone from Enceladus' perspective there is no better or worse here, so perhaps you could find a less value laden vocabulary....for example, 3rd least ice.  (And the ice couldn't give a damn.)

jdallen

  • Young ice
  • Posts: 3024
    • View Profile
  • Liked: 190
  • Likes Given: 172
Re: IJIS
« Reply #4864 on: August 26, 2017, 02:07:11 AM »
That seems about right.  If melt picks up then 4th place is likely, otherwise 6th.

I think it is worth noting, that all of the years in question which will likely exceed 2017's melt, have occurred in the last decade.

If this were 2007, everyone would be fairly complete shock over how low extent is, much less volume or area or quality of ice.

Recent years have innured us to just how incredible these seasons are.
This space for Rent.

Espen

  • Young ice
  • Posts: 3251
    • View Profile
  • Liked: 177
  • Likes Given: 0
Re: IJIS
« Reply #4865 on: August 26, 2017, 08:45:58 AM »
IJIS:

4,846,064 km2(August 25, 2017)down 47,493 km2 and 5th lowest measured for the date.
Have a ice day!

Quantum

  • New ice
  • Posts: 98
    • View Profile
  • Liked: 0
  • Likes Given: 0
Re: IJIS
« Reply #4866 on: August 26, 2017, 01:09:31 PM »
Extent for 8/24:  4893557 (drop of 49695; 5th lowest for this date).

2017 is now only 9437 km2 above the 2014 minimum.

Live version of the predict-o-matic now says 4.28 (range 4.06-4.49).  There is a 41% chance of 4th place and a 51% chance of 6th place.  Complete list of odds:

1st place:  0%
2nd place:  1%
3rd place:  2%
4th place: 41%
5th place:  5%
6th place:  51%
7th place:  2%
8th place or higher: 0%

Extent losses similar to those from 2007-2016 for the remaining days of the season would result in a 2017 minimum of 4.17 to 4.44


Sorry, could you explain to me why 5th place is so much less likely than 4th or 6th? That just seems so counterintuitive.

dnem

  • Frazil ice
  • Posts: 328
    • View Profile
  • Liked: 130
  • Likes Given: 55
Re: IJIS
« Reply #4867 on: August 26, 2017, 01:21:16 PM »

Sorry, could you explain to me why 5th place is so much less likely than 4th or 6th? That just seems so counterintuitive.

It's because there is only a very small difference between those two years for this year to "fit into" while there is a broader slot between 5th and 6th. 

dnem

  • Frazil ice
  • Posts: 328
    • View Profile
  • Liked: 130
  • Likes Given: 55
Re: IJIS
« Reply #4868 on: August 26, 2017, 03:14:22 PM »
Year     Daily Min      “Slot” Prob   Cum Prob
2012     3.18                  0                0
2016     4.02                  1                1
2007     4.07                  2                3
2015     4.26                 41               44
2011     4.27                  5               49
2008     4.50                 51              100
2010     4.62                  2              102

The first row indicates that 2017 has a 0% chance of finishing below 2012. The second row indicates that 2017 has a 1% chance of finishing above 2012 but below 2016 (for a new 2nd place minimum). The third row indicates that 2017 has a 2% chance of finishing above 2016 but below 2015 (for a new 3rd place minimum) and so on. The cumulative probability column gives the likelihood of a 2017 finish equal to, or below, the minimum indicated in the second column.  I think this (or something similar) is what Bill F. had in mind. (I assume rounding errors leads to the 102% cumulative probability.)  Ned's data.

There is a 5% chance of 2017 slotting into the very small gap between the 2015 and 2011 finish yet only a 2% chance of fitting into the larger gap between 2007 and 2016.  This is because the most likely outcome is centered right around that 4.25 -4.50 value.

Zeug Gezeugt

  • New ice
  • Posts: 76
    • View Profile
  • Liked: 9
  • Likes Given: 16
Re: IJIS
« Reply #4869 on: August 26, 2017, 03:45:57 PM »
Recent years have innured us to just how incredible these seasons are.

Indeed master jdallen, the further volume collapses as extent increases the more I'm reminded of my favourite ice slushies, the flavour of which is becoming rather bitter however. Gods bless our children's children as I'm afraid they will not remember us too fondly. Espen's IJIS thread remains my favourite arctic info stream, keep it coming as we approach zero point in the years ahead.

jplotinus

  • Frazil ice
  • Posts: 126
    • View Profile
  • Liked: 13
  • Likes Given: 11
Re: IJIS
« Reply #4870 on: August 26, 2017, 04:05:11 PM »
I'd be wary of characterizing 2017 as having "0" chance of going below 2012 minimum. Arctic sea ice does what it pleases, when it pleases, with no regard for human calculus.

Shared Humanity

  • Young ice
  • Posts: 3980
    • View Profile
  • Liked: 430
  • Likes Given: 52
Re: IJIS
« Reply #4871 on: August 26, 2017, 04:24:02 PM »
I would be gobsmacked if 2017 ended in 1st place.

dnem

  • Frazil ice
  • Posts: 328
    • View Profile
  • Liked: 130
  • Likes Given: 55
Re: IJIS
« Reply #4872 on: August 26, 2017, 05:42:53 PM »
I'd be wary of characterizing 2017 as having "0" chance of going below 2012 minimum. Arctic sea ice does what it pleases, when it pleases, with no regard for human calculus.

FWIW I was just presenting Ned W's year by year probabilities in a cumulative manner - I didn't calculate the probabilities. 

SH: What IS the probability of being "gobsmacked"? :-)

Bill Fothergill

  • Frazil ice
  • Posts: 278
    • View Profile
  • Liked: 0
  • Likes Given: 0
Re: IJIS
« Reply #4873 on: August 26, 2017, 05:55:09 PM »
...  I think this (or something similar) is what Bill F. had in mind. (I assume rounding errors leads to the 102% cumulative probability.)  ...

Yep, spot on.

Being an old fart, my brain was unable to dredge up the term "cumulative probability" from what passes for a memory these days.

My excuse for this inexcusable behaviour is that this was something we did in Statistics (a subset of 3rd Year Maths) - and that was about 44 years ago.

Aw shit - that also probably explains why I can't run anywhere near as fast as I used to back then.  :'( :'( :'(

Ned W

  • Guest
Re: IJIS
« Reply #4874 on: August 26, 2017, 08:48:29 PM »
That seems about right.  If melt picks up then 4th place is likely, otherwise 6th.

I think it is worth noting, that all of the years in question which will likely exceed 2017's melt, have occurred in the last decade.

If this were 2007, everyone would be fairly complete shock over how low extent is, much less volume or area or quality of ice.

Recent years have innured us to just how incredible these seasons are.
This is in fact fundamentally true of almost all aspects of climate change.  It happens slowly and people adjust their ideas of what is "normal" so the change doesn't seem so big, because we've forgotten what it was like in the past.

The great XKCD cartoon about cold weather in St Louis illustrated this nicely:

« Last Edit: August 26, 2017, 08:57:02 PM by Ned W »

Ned W

  • Guest
Re: IJIS
« Reply #4875 on: August 26, 2017, 08:50:18 PM »

Sorry, could you explain to me why 5th place is so much less likely than 4th or 6th? That just seems so counterintuitive.

It's because there is only a very small difference between those two years for this year to "fit into" while there is a broader slot between 5th and 6th.
Yep, that's it.  All of us keep finding that counterintuitive, Quantum, so you're in good company.

But it's true.  The odds of 4th place or 6th place are much higher than the odds of 5th place.

Ned W

  • Guest
Re: IJIS
« Reply #4876 on: August 26, 2017, 08:54:31 PM »
I'd be wary of characterizing 2017 as having "0" chance of going below 2012 minimum. Arctic sea ice does what it pleases, when it pleases, with no regard for human calculus.
Well, it's not a zero chance, but it rounds to zero even at three or four decimal places. 

The ice may not care about our calculations, but our calculations are based on the behavior of the ice.  So the two aren't independent. 

I feel pretty safe saying there won't be a new record this year.  Heck, I feel pretty safe now saying 2017 won't go below 4.00.

Pavel

  • Frazil ice
  • Posts: 204
    • View Profile
  • Liked: 20
  • Likes Given: 4
Re: IJIS
« Reply #4877 on: August 26, 2017, 09:25:30 PM »
Since the weather did everything to prohibit ice loss during the entire melting season I would'nt say we have a good result now on SIE. If this winter will be mild what quite likely due to warm SSTs, we might have even lower volume in April and Arctic can't dodge a cannonball each year

Shared Humanity

  • Young ice
  • Posts: 3980
    • View Profile
  • Liked: 430
  • Likes Given: 52
Re: IJIS
« Reply #4878 on: August 27, 2017, 02:17:30 AM »


SH: What IS the probability of being "gobsmacked"? :-)

There is 100% probability of me being gobsmacked if 2017 ends up in 1st place.  :o
« Last Edit: August 27, 2017, 04:51:14 PM by Shared Humanity »

Ned W

  • Guest
Re: IJIS
« Reply #4879 on: August 27, 2017, 02:57:37 PM »
26 August:  extent = 4,811,553.  Down 34,511 from previous day.  Now in 6th place.

Nine of the past ten days have seen below-average decreases in extent.  The total extent is still decreasing but it's been pretty slow. 

Espen

  • Young ice
  • Posts: 3251
    • View Profile
  • Liked: 177
  • Likes Given: 0
Re: IJIS
« Reply #4880 on: August 27, 2017, 03:44:30 PM »
IJIS:

4,811,553 km2(August 26, 2017)down 34,511 km2 and 6th lowest measured for the date.
Have a ice day!

jplotinus

  • Frazil ice
  • Posts: 126
    • View Profile
  • Liked: 13
  • Likes Given: 11
Re: IJIS
« Reply #4881 on: August 27, 2017, 05:27:27 PM »


SH: What IS the probability of being "gobsmacked"? :-)

There is 100% probability of me being gobsmacked if 2017 ends up in 1st place.  :o

I foresee a possible scenario that might bring on a gobsmack. Namely, a minimum after the  September equinox or possibly a new precedent for an October minimum. Smaller than average August extent losses are taking place in a context of low volume and of cooler temperatures. The low volume will continue. As it is already as cool as it has been in September in recent years, the only requirement here would be a continuation of the temperature level that is being experienced now, rather than a more normal decline. Put differently, because it is already cool above 80°N, it may simply continue to be the same in coming weeks as it is now. That could happen.
« Last Edit: August 28, 2017, 03:54:58 PM by jplotinus »

Dharma Rupa

  • Frazil ice
  • Posts: 493
    • View Profile
  • Liked: 55
  • Likes Given: 26
Re: IJIS
« Reply #4882 on: August 27, 2017, 07:14:35 PM »


SH: What IS the probability of being "gobsmacked"? :-)

There is 100% probability of me being gobsmacked if 2017 ends up in 1st place.  :o

I foresee a possible scenario that might bring on a gobsmack. Namely, a minimum after the  September equinox or possibly a new precedent for an October minimum. Smaller than average August extent losses are taking place in a context of low volumn and of cooler temperatures. The low volumn will continue. As it is already as cool as it has been in September in recent years, the only requirement here would be a continuation of the temperature level that is being experienced now, rather than a more normal decline. Put differently, because it is already cool above 80°N, it may simply continue to be the same in coming weeks as it is now. That could happen.

Like 2012:

Ned W

  • Guest
Re: IJIS
« Reply #4883 on: August 27, 2017, 07:53:12 PM »
Following 2012's trajectory for the remainder of the season would put us at 4.31 for the minimum.  Even an extent loss three times as large as 2012's would not be enough to set a new record.  Even two weeks of nonstop century breaks would not set a new record. 

It's not going to happen.

Deeenngee

  • New ice
  • Posts: 58
    • View Profile
    • Baddadgags
  • Liked: 0
  • Likes Given: 0
Re: IJIS
« Reply #4884 on: August 27, 2017, 10:52:39 PM »
Updated comparison of 2017 vs previous years' minima here. It looks instinctively to me like a collision course with 2008, 2011 or 2015, but really, who knows...?
https://twitter.com/baddadgags

One liners, dad jokes, other jokes; all mine. Please follow for 2 to 3 moderately amusing instances of the above per week, guaranteed.

Cato

  • New ice
  • Posts: 62
    • View Profile
  • Liked: 2
  • Likes Given: 0
Re: IJIS
« Reply #4885 on: August 28, 2017, 12:08:15 AM »
It cannot happen. Refreezing will be starting very soon, if it is not already happening. Synoptic configuration is causing some compaction but also a significant temperature drop in the next days, with no significant wind and plenty of clear sky that with solstice approaching will help release significant quantity of heat by radiation. It won't happen in my view, no doubt about it.

gregcharles

  • New ice
  • Posts: 91
    • View Profile
  • Liked: 6
  • Likes Given: 2
Re: IJIS
« Reply #4886 on: August 28, 2017, 05:22:28 AM »
I think you mean the equinox is approaching, not the solstice. I mean, yes, the solstice is approaching, just not as soon as the equinox.  :)

Unmex Chingon

  • New ice
  • Posts: 27
    • View Profile
  • Liked: 0
  • Likes Given: 0
Re: IJIS
« Reply #4887 on: August 28, 2017, 05:53:46 AM »
Would you rather there is a change to weather and ice recuperates?  OR would you want it to go to ZERO just to prove a point?

Cato

  • New ice
  • Posts: 62
    • View Profile
  • Liked: 2
  • Likes Given: 0
Re: IJIS
« Reply #4888 on: August 28, 2017, 07:40:02 AM »
I think you mean the equinox is approaching, not the solstice. I mean, yes, the solstice is approaching, just not as soon as the equinox.  :)
Ahahah good point gregcharles! I meant equinox but solstice is acceptable as well I reckon  ;D

Ned W

  • Guest
Re: IJIS
« Reply #4889 on: August 28, 2017, 11:17:50 AM »
An uptick today:

Extent for 8/27:  4,812,329 (6th place)
Increase of 776 km2

By my count, there have been five previous days with upticks in August during the 2002-2017 extent record, only two of which (in 2002 and 2005) were this early.  2006 and 2014 had upticks on Aug 31, and 2010 had an uptick on Aug 29.

Ned W

  • Guest
Re: IJIS
« Reply #4890 on: August 28, 2017, 11:27:24 AM »
Updated comparison of 2017 vs previous years' minima here. It looks instinctively to me like a collision course with 2008, 2011 or 2015, but really, who knows...?

Quote from: Ned W
An uptick today:

Maybe the ice heard you, Deeenngee, and swerved to avoid a collision!

 :)

Ned W

  • Guest
Re: IJIS
« Reply #4891 on: August 28, 2017, 03:13:09 PM »
This ongoing slowdown has wreaked havoc on the predict-o-matic.  Recall that its original (June 10) prediction for the minimum was 4.21.  Now it's up to 4.36.  In terms of the net loss of extent since June 10, the predict-o-matic's accuracy, which had been hovering around 99%, is now down to 96.4%.  The big problem is that the original prediction (4.21) is now outside the 95% confidence interval as of today (4.22 - 4.51).  Well, it was bound to happen eventually. 

Actually, it's been an improbably successful run, for a lot longer than I expected:


A 6th place finish is now the overwhelming favorite:



Following the pattern of various previous years (2007-2016) would result in various minimums from 4.25 to 4.48:



I can't remember what I ended up picking in the August poll (I was waffling between 3.75-4.25 and 4.0-4.5) but the 3.75-to-4.25 bin is hovering on the brink of death -- the bins starting at 4.0 and 4.25 are vastly more likely:


Espen

  • Young ice
  • Posts: 3251
    • View Profile
  • Liked: 177
  • Likes Given: 0
Re: IJIS
« Reply #4892 on: August 28, 2017, 04:06:08 PM »
IJIS:

4,812,329 km2
(August 27, 2017)up 776 km2 and 6th lowest measured for the date.
Have a ice day!

Rob Dekker

  • Nilas ice
  • Posts: 2338
    • View Profile
  • Liked: 105
  • Likes Given: 114
Re: IJIS
« Reply #4893 on: August 29, 2017, 08:35:03 AM »
This ongoing slowdown has wreaked havoc on the predict-o-matic.  Recall that its original (June 10) prediction for the minimum was 4.21.  Now it's up to 4.36.  In terms of the net loss of extent since June 10, the predict-o-matic's accuracy, which had been hovering around 99%, is now down to 96.4%.  The big problem is that the original prediction (4.21) is now outside the 95% confidence interval as of today (4.22 - 4.51).  Well, it was bound to happen eventually. 

Hi Ned, that is still a remarkable track record.
Are you using only 'extent' as your predictor ?
And did the fact that 2017 'extent' closely followed the "2010's average" since May play a role in the remarkable accuracy of your "predict-o-matic" ?
Also, next year, do you want to enter your prediction for the SIPN ?
https://www.arcus.org/sipn
This is our planet. This is our time.
Let's not waste either.

Ned W

  • Guest
Re: IJIS
« Reply #4894 on: August 29, 2017, 02:08:21 PM »
Hi Ned, that is still a remarkable track record.
Are you using only 'extent' as your predictor ?
And did the fact that 2017 'extent' closely followed the "2010's average" since May play a role in the remarkable accuracy of your "predict-o-matic" ?
Also, next year, do you want to enter your prediction for the SIPN ?
https://www.arcus.org/sipn
Thanks for the note, Rob. 

The name "predict-o-matic" was kind of a joke, but I decided I liked it.  It is basically an overcomplicated way of doing something very simple.  People keep asking about it, so I guess I should explain.

The goal is to make predictions about extent during the melting season.   The core methods that it uses for this are ridiculously straightforward and other people here use them too, but in the predict-o-matic they're wrapped in a very sophisticated and fancy interface. 

There are two versions.  Both are based on the historical extent data from 2007-present (one could use pre-2007 data but I vaguely feel like some kind of radical threshold was crossed in 2007 so I start there).

Boring details inside the quote box, with more interesting discussion below...

Quote
V1: For any day t0, compute the extent change dE(year,t0,min) for every year from day t0 to that year's minimum, regardless of on what day that minimum occurs.  Calculate the mean, standard deviation, etc. of those extent changes dE.  Subtract the mean change from the current year's extent on day t0 to predict the minimum, and use the standard deviation to calculate various probabilities (see below).

V2: For any day t0, compute the average extent change dE(year,t0,tn) to day t1, t2, t3, ... tn in each previous year, and calculate the mean and standard deviation of dE on each day.  Subtract the mean change for each day t1 ... tn from the current year's extent on day t0 to predict the evolution of extent over time, and use the standard deviation to calculate various probabilities.

Basically, version 1 gives the average of the minimums, while version 2 gives the minimum of the averages.  Those are different.  Version 1 always gives a slightly lower minimum (e.g., 4.14 vs 4.21) but version 2 forecasts the day-by-day evolution of extent, which version 1 doesn't do. 

A not fully implemented version 3 combines the previous two, to give both the evolution over time plus the actual expected value and date of the minimum.

I know other people here do the same kind of calculations (hello, gerontocrat!).  The nice thing about the predict-o-matic is that it's fully automated and generates a large number of outputs, including:

* Expected value of the minimum, plus a 95% confidence interval
* Expected value on every day of the season, plus a 95% confidence interval
* Projections of the minimum if the current year followed the trajectory of each previous year
* Probability values for each of the "bins" in the monthly JAXA extent polls
* Probability values for "ranks" relative to previous years
* Live-updating graphs for all of the above
* Day-by-day accuracy statistics of the initial ("static") forecast from the start of the season

The live updating graphs are really helpful to me in visualizing the probabilities.  I've posted examples of the graphs here; scroll up-thread to see what they look like. 

So the fundamental process is embarrassingly simple but the interface is nice.

You're absolutely 100% correct that the "success" (  ::) ) of the predict-o-matic this year is entirely due to the bizarrely average evolution of extent over the course of this melt season.  I wish I could claim credit for continuous, uninterrupted 95+ percent accuracy over the past nine weeks, but I can't.

Another secret is that it's all part of an amateur psychology experiment.  We on the ASIF have a really poor record of prediction on the polls (see this thread) -- like an iceberg, the center of mass of our predictions tends to be way below the waterline.  Time after time our monthly polls on daily or monthly extent or area end up much much too low.  I thought that providing a bunch of posts with eye-catching graphics explaining the probabilities, during the time when people are making decisions on the polls, might help bump the distribution of poll responses towards more realistic numbers.  Dunno if it had any effect.

Looking ahead, I have a bunch of ideas to improve the extent predict-o-matic, by incorporating other sources of information.  But my first priority is something Jim Hunt asked about at the top of this page -- a daily-updating PIOMAS volume predict-o-matic.  That would be much more useful, and I've got a plan for how to do it.

Here's one of the output graphs I haven't posted lately -- it shows how the V1 predictions (the "lower" one) have evolved over the past ten weeks:



The predicted minimum has ranged from a low of 4.02 to a high of 4.41 (that's today's prediction -- yes, it's the highest one of the season). 
« Last Edit: August 29, 2017, 02:48:09 PM by Ned W »

Ned W

  • Guest
Re: IJIS
« Reply #4895 on: August 29, 2017, 02:12:05 PM »
And speaking of that, we had another uptick on extent:

28 Aug 2017 - extent 4,812,739 (6th place)
Increase of 410 km2 since previous day
Two consecutive days of increases

The probability of 6th place is now overwhelming (87%), and 7th place (10%) is now much more likely than anything less than 6th place (3%).

gerontocrat

  • First-year ice
  • Posts: 6809
    • View Profile
  • Liked: 1705
  • Likes Given: 22
Re: IJIS
« Reply #4896 on: August 29, 2017, 02:42:34 PM »
And speaking of that, we had another uptick on extent:

28 Aug 2017 - extent 4,812,739 (6th place)
Increase of 410 km2 since previous day
Two consecutive days of increases

The probability of 6th place is now overwhelming (87%), and 7th place (10%) is now much more likely than anything less than 6th place (3%).
Since 14th August extent loss has been so low (see graph) that it has made outcomes that seemed eminently possible now look - daft (see table for my self-imposed humiliation).

The problem is that in an individual season a period of a few days or weeks, especially at end of season(?) is so variable that predictions become a crap-shoot.

An off-topic note - Antarctic is doing the reverse, three days of extent loss totalling 125,000 km2 when on average a gain of about the same - 125,000 km2 - happens. Another table starts to look like junk.
"Para a Causa do Povo a Luta Continua!"
"And that's all I'm going to say about that". Forrest Gump
"Damn, I wanted to see what happened next" (Epitaph)

Espen

  • Young ice
  • Posts: 3251
    • View Profile
  • Liked: 177
  • Likes Given: 0
Re: IJIS
« Reply #4897 on: August 29, 2017, 03:59:34 PM »
IJIS:

4,812,739 km2(August 28, 2017)up 410 km2 and 6th lowest measured for the date.
Have a ice day!

Espen

  • Young ice
  • Posts: 3251
    • View Profile
  • Liked: 177
  • Likes Given: 0
Re: IJIS
« Reply #4898 on: August 30, 2017, 12:35:06 PM »
IJIS:

4,775,338 km2(August 29, 2017)down 37,401 km2 and 6th lowest measured for the date.
Have a ice day!

pauldry600

  • Frazil ice
  • Posts: 201
    • View Profile
    • weathergossip
  • Liked: 9
  • Likes Given: 3
Re: IJIS
« Reply #4899 on: August 31, 2017, 01:20:28 PM »
Down to 4.72 now.