Basic questions and discussions about melting and freezing physics

[Feeltheburn]

This very interesting discussion regarding the heat losses and gains when ice melts and refreezes got me to thinking about a different heat flow issue involving water: Liquid water and water vapor.

I can't seem to find much in the climate science literature about it, but it is in physics books and an old popular science archive from 1891. Yes, that long ago!

When water evaporates from a body of water, the remaining thermal energy in the water is reduced by the amount of heat required to cause the phase change, or the latent heat of vaporization, which is enormous (540 calories per gram of water). It takes 100 calories to heat water from 0 degrees C to 100 degrees C, but 5-1/2 times more heat to just change liquid water to water vapor at constant temperature. This explains why evaporative coolers are very efficient at lowering air temperature in hot, dry climates.

When water evaporates from the ocean, heat energy (540 cal/gram) is transferred from the water to the water vapor, which then rises up into the upper atmosphere, where it meets extremely cold dry air, forms clouds, and then precipitates as rain or snow. When water vapor turns into water or ice, an enormous amount of heat energy is released into the surrounding upper atmosphere equal to the latent heats of vaporization and fusion.

Thus, there is a constant cooling cycle in which heat energy in the ocean is continuously being transferred to the upper atmosphere, where extremely cool rarified air absorbs this energy, causing water to precipitate and return to the earth as cold water or ice. The energy emitted from the precipitating water into the rarified air is then emitted back to outer space. I wonder if anyone has ever done the mathematical calculations to determine the absolute quantity of heat energy that is released from the oceans and sent on a one way ticket to the upper atmosphere where it is mostly sent back into space by the process of evaporation and precipitation cycles.

I read people saying there is unaccounted for heat hiding somewhere in the ocean, just waiting to strike the climate with a vengeance. But what if this energy isn't missing or hiding at all but slipping out the back door right before our very eyes by rising water vapor that is a conduit that pumps heat energy from the ocean back into space through continuous cycles of surface evaporation (picking up heat), rising 20,000 to 40,000 miles above the earth, and then releasing this energy into the freezing upper atmosphere when water precipitates and falls back to the earth. The heat cannot come back to earth because the precipitated water by definition gave up all its heat of vaporization and fusion to the upper atmosphere.

In short, rising water vapor is a one way conduit that removes heat from the ocean and all bodies of water and sends it up to the upper atmosphere, where it is released from the water, which falls back to earth depleted of heat, forming a heat gradient in the upper atmosphere. Because of entropy, hotter air always moves in the direction of colder air, which is air that is further from earth and closer to space. That means the heat that is released by precipitation is mostly lost to outer space. There is no mechanism to return it to earth.

Thus, the ocean is one giant heat pump that sends heat back to space through a process that is remarkably similar to how air conditioners work through a working fluid that evaporates to cool locally in one place and condenses to heat locally at another place. The cooling through evaporation is at the earth's surface, and the heating through condensation to release heat is at 20,000 to 40,000 feet where the heat is essentially given up by the earth for good.

I welcome all opinions and counterpoints.

[oren]

I read people saying there is unaccounted for heat hiding somewhere in the ocean, just waiting to strike the climate with a vengeance. But what if this energy isn't missing or hiding at all but slipping out the back door right before our very eyes by rising water vapor that is a conduit that pumps heat energy from the ocean back into space 

I doubt you are describing it correctly, and I doubt "most energy is immediately lost to outer space", but regardless this mechanism has been in place forever, while ocean heat content growth is measurable and very real, so what is there to doubt?

[binntho]

I read people saying there is unaccounted for heat hiding somewhere in the ocean, just waiting to strike the climate with a vengeance. But what if this energy isn't missing or hiding at all but slipping out the back door right before our very eyes by rising water vapor that is a conduit that pumps heat energy from the ocean back into space

I doubt you are describing it correctly, and I doubt "most energy is immediately lost to outer space", but regardless this mechanism has been in place forever, while ocean heat content growth is measurable and very real, so what is there to doubt?

I think in general FTB's essay has it right, but the particulars that oren points out are what needs closer scrutiny.

For one, the heat content of the oceans is something that can be (and is) measured directly, and it is rapidly increasing (causing most of the current sea level rise - the oceans are acting as their own old-fashioned thermometer).

As oren points out, the evaporative cycle is nothing new but it's there and it is probably conducting more heat the hotter the oceans get.

As for condensation, not all the water vapor condenses in the upper atmosphere - much of it condenses back to the surface (e.g. during nighttime) or close to the surface as fog. Not to forget the Foehn effect where heat released by condensation is brought down from the mountains into the lowlands, significantly raising near-surface temperatures.

And the heat released by cloud condensation is going to be radiative heat to a large degree, and given both greenhouse gases and the fact that this is happening inside a cloud, I don't think it's possible to claim that most of it gets lost to space. Much of it certainly, but how much? Does anybody know?

But in the end this is all just part of the heat balance, the flow of energy in the atmosphere (including the oceans), all very well measured and accounted for in total although the details may be a bit trickier.

[P-maker]

Felltheburn,

To some extent, you are thinking along the right lines. Alternatively, you are messing up some of the basic physics.

As long as we are in the Tropics, everything is honkydory. Most physical processes take place well above 0C (or 273K). It is when you start using calories and F, your problems begin.

When in the Sub-tropics, some of the precipitation comes down as hail. In recent years, we have seen an increasing hail diameter and more devastating hailstorms spreading north. This may be a sign of the kind of physics you focus on. More evaporative cooling in combination with more humid air should eventually lead to bigger hail, if condensation nuclei are present.

Moving to Mid-latitudes, we begin to have the change-over from freezing rain to warm rain. Apparently, you have forgot to include the temperature of the falling rain. All heat evaporated is not simply sent out to outer space. Some of it will return to Earth as a mild, warm rain, where you would even enjoy getting your underpants wet from time to time.

Now, finally to the physical processes in the Arctic. Your idea that all evaporative cooling will eventually leave the Arctic cold and the World in thermal balance (had it not been for the wicked ocean heat storage), is simply wrong. The moisture advected from southerly latitudes (as well as the minor part of the moisture evaporated from open Arctic waters), will eventually have to come down again as either rain (above 0C), or as snow. In the latter case, we may see the opposite of "sublimation" - that is water vapour going directly into the frozen phase - which of course releases a lot more energy, than just converting water vapour directly into rain.

In the presence of cloud condensation nuclei, we will see hailstorms spreading north, and we will see rain and showers entering the Arctic during winter. What we have not seen yet, is the physical reaction from/to a clean - basically CCN-free - Arctic atmosphere. My guess would be that initially, we will see ice pellets, but eventually we will see drizzle as the Arctic temperatures during winter goes above 0C.

Sitting in Svalbard through the dark "drizzle season" is no great joy I would presume. Maybe it will be a great time to reflect on the basic physics of our time.

Cheers P

[Klondike Kat]

Feeltheburn is not wrong in his understanding of the physics.  It may be a simplistic way of stating the processes.  In general, heat flows from a hot body to a colder body.  During evaporation, water vapor carries heat energy into the atmosphere, where it can dissipated to the surrounding air molecules.  This is many diverge in their analysis.  Feeltheburn stated that it is "mostly sent back to space."  P-maker states that, "some of it will return to earth."  The major question is how much?  Many modeler claim that a significant portion of the energy will simply return to the surface via the radiative effect due to the higher concentration of greenhouse gases.  This seems to be an overestimate as it involves heat flowing from a cooler to a warmer body.  Feeltheburn is probably close to the correct answer, and explains why so many have overestimated the warming effect of increasing atmospheric concentrations of greenhouse gases.  It would also help explain why the warmer waters entering the Arctic have not resulted in lower sea ice minima.

[binntho]

Feeltheburn is probably close to the correct answer, and explains why so many have overestimated the warming effect of increasing atmospheric concentrations of greenhouse gases.  It would also help explain why the warmer waters entering the Arctic have not resulted in lower sea ice minima.

This got me confused. Who have been overestimting the warming effect of increased greenhous gases?

And am I missing something or haven't the lower sea ice minima been coming in at a regular rate as the oceans get warmer?

In the last few decades sea ice has declined drastically - so what is there to explain? Feeltheburn is, I feel, barking up the wrong tree altogether.

[Klondike Kat]

Feeltheburn is probably close to the correct answer, and explains why so many have overestimated the warming effect of increasing atmospheric concentrations of greenhouse gases.  It would also help explain why the warmer waters entering the Arctic have not resulted in lower sea ice minima.

This got me confused. Who have been overestimting the warming effect of increased greenhous gases?

And am I missing something or haven't the lower sea ice minima been coming in at a regular rate as the oceans get warmer?

In the last few decades sea ice has declined drastically - so what is there to explain? Feeltheburn is, I feel, barking up the wrong tree altogether.

Based on the NSIDC data, sea ice minima were occurring roughly every three years until 2007.  Since then, there has been only one - in 2012.  Nine of the past eleven years have seen a higher minimum, and this year is likely to make ten.  The other year, 2016, barely edged out 2007 for a lower minimum.  All this has occurred while the oceans have been getting warmer.  So no, lower sea ice minima have NOT been coming at a regular rate as the oceans get warmer.

[binntho]

Feeltheburn is probably close to the correct answer, and explains why so many have overestimated the warming effect of increasing atmospheric concentrations of greenhouse gases.  It would also help explain why the warmer waters entering the Arctic have not resulted in lower sea ice minima.

This got me confused. Who have been overestimting the warming effect of increased greenhous gases?

And am I missing something or haven't the lower sea ice minima been coming in at a regular rate as the oceans get warmer?

In the last few decades sea ice has declined drastically - so what is there to explain? Feeltheburn is, I feel, barking up the wrong tree altogether.

Based on the NSIDC data, sea ice minima were occurring roughly every three years until 2007.  Since then, there has been only one - in 2012.  Nine of the past eleven years have seen a higher minimum, and this year is likely to make ten.  The other year, 2016, barely edged out 2007 for a lower minimum.  All this has occurred while the oceans have been getting warmer.  So no, lower sea ice minima have NOT been coming at a regular rate as the oceans get warmer.

I think you are confusing random variability with trend. On a decadal trend, it is very obvious that sea ice has declined as ocean temps have gone up. Looking for that to happen every year is perhaps a bit simplistic.

[Klondike Kat]

Feeltheburn is probably close to the correct answer, and explains why so many have overestimated the warming effect of increasing atmospheric concentrations of greenhouse gases.  It would also help explain why the warmer waters entering the Arctic have not resulted in lower sea ice minima.

This got me confused. Who have been overestimting the warming effect of increased greenhous gases?

And am I missing something or haven't the lower sea ice minima been coming in at a regular rate as the oceans get warmer?

In the last few decades sea ice has declined drastically - so what is there to explain? Feeltheburn is, I feel, barking up the wrong tree altogether.

Based on the NSIDC data, sea ice minima were occurring roughly every three years until 2007.  Since then, there has been only one - in 2012.  Nine of the past eleven years have seen a higher minimum, and this year is likely to make ten.  The other year, 2016, barely edged out 2007 for a lower minimum.  All this has occurred while the oceans have been getting warmer.  So no, lower sea ice minima have NOT been coming at a regular rate as the oceans get warmer.

I think you are confusing random variability with trend. On a decadal trend, it is very obvious that sea ice has declined as ocean temps have gone up. Looking for that to happen every year is perhaps a bit simplistic.

The decadal trend from the mid 90s to the mid 00s was definitively and largely downward.  The decadal trend since has been flat - a very slight increase starting at 2007, a similar decrease starting at 06 or 08.  I am not looking for it to happen every year - just more than once a decade.

[binntho]

The decadal trend from the mid 90s to the mid 00s was definitively and largely downward.  The decadal trend since has been flat - a very slight increase starting at 2007, a similar decrease starting at 06 or 08.  I am not looking for it to happen every year - just more than once a decade.

I think you'll find that there is nowhere near enough datapoints to make statistically valid claims about changes in trends from one decade to another. In other words, it is not valid scientifically to say that there are differences in trend between one decade and another.

Besides just looking at a graph of minimums since the start of the satellite era shows a growing downwards trend, to match a growing warmin oceans trend.

But it's interesting that you think that Feeltheburn's hypothesis of having found a major missing piece in the whole field of meteorology can explain why your once-a-decade minimums aren't coming in (besides, I thought 2012 was this decade's minimum?)

[MyACIsDying]

The AMOC has been weakening this decade to it's lowest strength, when the AMOC was a reason for arctic melt to increase, one would assume this weakening to minimum would mean increase in arctic ice extent? Clearly not.. so I wholly disagree on the flat lining trend of arctic ice melt. The 'Slow Transition' thread also explains reasons why extent is proportionally dropping less than volume is.

Klondike Kat I appreciate having your down to earth attitude on this forum but to say people are overestimating the effect GHG's?? I thought the science world was in general agreement that GHGs raising our radiation emittance roof is by far the largest reason for the change in thermodynamic equilibrium we know is happening by increases in heat content (.4Wms-2 imbalance to ~1200 Wms-2 incoming solar radiation may seem little but it accounts for all of the ~10^21 J /year OHC increase)

gerontocrat pointed out recently this is 130 times the energy used for our yearly losses to put that in perspective. Sure we can have localized feedback mechanisms that bring areas to cold extremes but I doubt any of these would be able overcome the global increase in temps in a permanent matter.

This is all assuming we keep the trend of GHG increase as it is.. I'm very curious what would happen if we lower the roof by taking CO2 out of the atmosphere? A cooldown faster than the warmup has taken I expect. Anyone know of models that ran such scenarios?

[Klondike Kat]

The AMOC has been weakening this decade to it's lowest strength, when the AMOC was a reason for arctic melt to increase, one would assume this weakening to minimum would mean increase in arctic ice extent? Clearly not.. so I wholly disagree on the flat lining trend of arctic ice melt. The 'Slow Transition' thread also explains reasons why extent is proportionally dropping less than volume is.

Klondike Kat I appreciate having your down to earth attitude on this forum but to say people are overestimating the effect GHG's?? I thought the science world was in general agreement that GHGs raising our radiation emittance roof is by far the largest reason for the change in thermodynamic equilibrium we know is happening by increases in heat content (.4Wms-2 imbalance to ~1200 Wms-2 incoming solar radiation may seem little but it accounts for all of the ~10^21 J /year OHC increase)

gerontocrat pointed out recently this is 130 times the energy used for our yearly losses to put that in perspective. Sure we can have localized feedback mechanisms that bring areas to cold extremes but I doubt any of these would be able overcome the global increase in temps in a permanent matter.

This is all assuming we keep the trend of GHG increase as it is.. I'm very curious what would happen if we lower the roof by taking CO2 out of the atmosphere? A cooldown faster than the warmup has taken I expect. Anyone know of models that ran such scenarios?

Thank you for your kind response.  The "people" to which I was referring are the posters here who claim that the current scientific thinking is too conservative and that the current warming is being masked by factor X, and once this is overcome, warming will skyrocket, causing all sorts of global catastrophes.  What if factor X for Arctic sea ice is just as FeeltheBurn has postulated, and will never be overcome?

[gerontocrat]

I am not even going to try to get into the physics of the melting of ice. What I am looking at is what are the best measures of trends in sea ice loss (or gain)

When I was looking at trends in sea ice, like most people I used to look mostly at the sea ice annual minimum, and a little bit at the maximum. But as I started to make graphs of the Arctic in total, and then sea by sea, it became apparent that minimal reductions or even increases in the minimum could conceal significant reductions in sea ice.

I now have data that looks at ice-free days, open water percentages at various times of the year, and finally 365 trailing averages for extent, area and volume, the latter which  I attach. These show that although the 2012 JAXA extent minimum was 0.84 million km2 below any minimum since, the 2016 year produced a lower 365 day average. This was because sea ice was low for much longer than the crash in 2012.

2019 may now be looking like a bit of an end-of-season anti-climax, but nevertheless if refreeze is slow, the 365 day trailing average could well be a new record low by Spring 2020.

After the minimum, and after the end of the calendar year I hope to be able to show a similar story when looking at ice-free days and increases in open water.

So what am I saying - simply that minima may not be falling - but sea ice is declining.

[Klondike Kat]

Geronocrat,
I agree that the ice has been generally in decline for the past four decades.  This occurred slowly at first, as the maximum sea ice started to fall.  The minimum was declining only slightly until the late 90s, when it plummeted.  This led to ab overall ice declined at a much faster rate until recently, when it has slowed considerably.  This occurred during the time that many pundits predicted an accelerated loss in Arctic sea ice to near ice-free conditions this decade.    All I am saying is that that his explanation is plausible explanation for the change.

[gerontocrat]

Geronocrat,
I agree that the ice has been generally in decline for the past four decades.  This occurred slowly at first, as the maximum sea ice started to fall.  The minimum was declining only slightly until the late 90s, when it plummeted.  This led to ab overall ice declined at a much faster rate until recently, when it has slowed considerably.  This occurred during the time that many pundits predicted an accelerated loss in Arctic sea ice to near ice-free conditions this decade.    All I am saying is that that his explanation is plausible explanation for the change.

I just think that the minima to some extent disguise the loss of sea ice over time, and that the graphs I posted support that to some extent. I make no firm predictions about when a BOE etc etc will happen, but
- CO2 ppm will increase significantly for a good few years more on the most optimistic assumptions,
- the atmosphere will heat up for many more years,
- the Arctic will heat up faster than the average.

Ice and heat don't make long-lasting partners. The cryosphere in the Arctic will diminish, and so my very unfirm prediction is for summer sea ice to be very very much less by 2030, for which I cannot offer any evidence other than the above.

[sailor]

Geronocrat,
I agree that the ice has been generally in decline for the past four decades.  This occurred slowly at first, as the maximum sea ice started to fall.  The minimum was declining only slightly until the late 90s, when it plummeted.  This led to ab overall ice declined at a much faster rate until recently, when it has slowed considerably.  This occurred during the time that many pundits predicted an accelerated loss in Arctic sea ice to near ice-free conditions this decade.    All I am saying is that that his explanation is plausible explanation for the change.

I just think that the minima to some extent disguise the loss of sea ice over time, and that the graphs I posted support that to some extent. I make no firm predictions about when a BOE etc etc will happen, but
- CO2 ppm will increase significantly for a good few years more on the most optimistic assumptions,
- the atmosphere will heat up for many more years,
- the Arctic will heat up faster than the average.

Ice and heat don't make long-lasting partners. The cryosphere in the Arctic will diminish, and so my very unfirm prediction is for summer sea ice to be very very much less by 2030, for which I cannot offer any evidence other than the above.

Completely true. A good example is 2016 Winter. It’s kinda predictable what will come, but how, it will amuse us or terrify us

[binntho]

When I was looking at trends in sea ice, like most people I used to look mostly at the sea ice annual minimum, and a little bit at the maximum. But as I started to make graphs of the Arctic in total, and then sea by sea, it became apparent that minimal reductions or even increases in the minimum could conceal significant reductions in sea ice.

I now have data that looks at ice-free days, open water percentages at various times of the year, and finally 365 trailing averages for extent, area and volume, the latter which  I attach. These show that although the 2012 JAXA extent minimum was 0.84 million km2 below any minimum since, the 2016 year produced a lower 365 day average. This was because sea ice was low for much longer than the crash in 2012.
...
So what am I saying - simply that minima may not be falling - but sea ice is declining.

I agree with you Gero, and I think that the three graphs you posted there are by far the best when it comes to looking at the real underlying trends.

But I am not so sure about the red polynomial lines on two of them - just from eyeballing I would have thought that a simple curve (i.e. second grade polynomial) with an increasing downwards curve would be the best match?

[binntho]

Geronocrat,
I agree that the ice has been generally in decline for the past four decades.  This occurred slowly at first, as the maximum sea ice started to fall.  The minimum was declining only slightly until the late 90s, when it plummeted.  This led to ab overall ice declined at a much faster rate until recently, when it has slowed considerably.  This occurred during the time that many pundits predicted an accelerated loss in Arctic sea ice to near ice-free conditions this decade.    All I am saying is that that his explanation is plausible explanation for the change.

I'm a bit surprised at your repeated claims here KK.

"The minimum was declining only slightly until the late 90s when it plummeted" is not correct, nor has it slowed "considerably" recently.

The attached is NSIDC daily minimum figures, up to and including 2019 at 4.3 4.6, which is appr. today's number. This graph does not allow any such statements, and there are no statistical models that will support anything other than a straight line.

The curve does jump up and down, but that cannot be shown to be other than random variability. So any and all guesses as to explanations for probably non-existent changes in trend are really a bit silly.

EDIT: The figure I used for 2019 was the Jaxa figure, NSIDC is apparently 0.3 MKm2 higher. Graph is corrected.

[Klondike Kat]

Geronocrat,
I agree that the ice has been generally in decline for the past four decades.  This occurred slowly at first, as the maximum sea ice started to fall.  The minimum was declining only slightly until the late 90s, when it plummeted.  This led to ab overall ice declined at a much faster rate until recently, when it has slowed considerably.  This occurred during the time that many pundits predicted an accelerated loss in Arctic sea ice to near ice-free conditions this decade.    All I am saying is that that his explanation is plausible explanation for the change.

I'm a bit surprised at your repeated claims here KK.

"The minimum was declining only slightly until the late 90s when it plummeted" is not correct, nor has it slowed "considerably" recently.

The attached is NSIDC daily minimum figures, up to and including 2019 at 4.3 4.6, which is appr. today's number. This graph does not allow any such statements, and there are no statistical models that will support anything other than a straight line.

The curve does jump up and down, but that cannot be shown to be other than random variability. So any and all guesses as to explanations for probably non-existent changes in trend are really a bit silly.

EDIT: The figure I used for 2019 was the Jaxa figure, NSIDC is apparently 0.3 MKm2 higher. Graph is corrected.

Your curve supports my claim as stated.  The minimum was decreasing slowly, until it took a bigger dive in the 90s.  That lasted about 15 years, depending on which curve fitting one chooses.  Since then, the minimum has flatlined.  Calling those significant chances random variability is quite the stretch.  Even the curve Gerontocrat presenter shows the changes.  My claim should come as no surprise to those who have been following the ice for the past years (decades),

[Archimid]

So basically your argument is that "there has been no warming since 1998" adapted poorly to "Sea Ice stopped melting in 2012".

For that you use a pseudo explanation, seconded by the climate change deniers and validated by  Bintho's polite argument.

The "people" to which I was referring are the posters here who claim that the current scientific thinking is too conservative and that the current warming is being masked by factor X, and once this is overcome, warming will skyrocket, causing all sorts of global catastrophes. 

This has already happened and will keep happening and getting worse. You can't see it because fear blinds you, but you will feel it soon regardless if you understand what is happening or not.

[gerontocrat]

I agree with you Gero, and I think that the three graphs you posted there are by far the best when it comes to looking at the real underlying trends.

But I am not so sure about the red polynomial lines on two of them - just from eyeballing I would have thought that a simple curve (i.e. second grade polynomial) with an increasing downwards curve would be the best match?

So did I, so I told it to do it, and it produced a lunatic U curve pointing UP, and a lousy R2. So I looked for a reasonable R2 and a result not showing ridiculous future increases or decreases. The ones on the graphs are the best I could do. But I don't like these red trend lines either. These periodic large rises and falls are just as well represented by an obviously untrue linear projection.

My own pure guess is that we are looking at 2 steps down, 1 step up as the years go by until the ice gets so thin and fragmented that in a year favourable for melting - poof ! (as posters have shown here and there on the melting season thread). But that is pure speculation (that now belongs to me since I stole it from others).

[binntho]

Calling those significant chances random variability is quite the stretch. 
<snip> 
My claim should come as no surprise to those who have been following the ice for the past years (decades),

Well yes, I've seen the same claim made over and over again over the last few years and every time I try to point out that it simply can't be done that way. The points are way too few, the variation far too big, to be able to claim with any statistical support whatsoever that the trend has changed during the satellite era.

This is not to say that there has not been a change in trend, but statistical analysis will not support this. It is not good enough to squint at the chart and say, "it seems to me that the trend is bigger here than there ..." It can be good fun, and it can be very interesting to try to figure out why the trend has changed (that is, of course, if it has changed and it's not just natural variability).

But going from "it looks to me like the trend has been changing" to taking Feeltheburns attempts at rewriting meteorology to support this feeling is a huge stretch, literally an immaginative jump. And as such, pretty silly.

And look at the graph again, and compare it to this famous graph, copied from tamino's webpage. It's obvious, isn't it, that there is a change in trend between the 90s, the noughts and the teens? The noughts are obviously stalling.



But no, Tamino has shown again and again that statistically there was no pause in global warming in the noughts - it just looks that way because of the randomness of natural variability.

And the same goes for the changes in SIE minimums over the last four decades - randomnes and natural variability are just as good an explanation as any hypothetical state changes leading to changes in melting rate.

https://tamino.wordpress.com/

[binntho]

I agree with you Gero, and I think that the three graphs you posted there are by far the best when it comes to looking at the real underlying trends.

But I am not so sure about the red polynomial lines on two of them - just from eyeballing I would have thought that a simple curve (i.e. second grade polynomial) with an increasing downwards curve would be the best match?

So did I, so I told it to do it, and it produced a lunatic U curve pointing UP, and a lousy R2.

I tried the same and got the same result. Doh!

[binntho]

For that you use a pseudo explanation, seconded by the climate change deniers and validated by  Bintho's polite argument.

*sigh* I keep forgetting that there are Americans reading this blog. What I did was what we in Europe call "a polite rejection", similar to the Chinese custom of helping your opponent maintain face.

As Mrs. Brown would say, "that's nice".

Besides, Feeltheburn's small essay was well written and started off fine, but it ended in a very strange conclusion, so dealing politely with it seemed to me to be the best thing. But did I validate his conclusion? I hope not.

[Archimid]

Your second reply completely invalidates his conclusion with a very appropriate language and tone. Thank you for that great reply and thank you for proving me wrong.

[Klondike Kat]

Calling those significant chances random variability is quite the stretch. 
<snip> 
My claim should come as no surprise to those who have been following the ice for the past years (decades),

Well yes, I've seen the same claim made over and over again over the last few years and every time I try to point out that it simply can't be done that way. The points are way too few, the variation far too big, to be able to claim with any statistical support whatsoever that the trend has changed during the satellite era.

This is not to say that there has not been a change in trend, but statistical analysis will not support this. It is not good enough to squint at the chart and say, "it seems to me that the trend is bigger here than there ..." It can be good fun, and it can be very interesting to try to figure out why the trend has changed (that is, of course, if it has changed and it's not just natural variability).

But going from "it looks to me like the trend has been changing" to taking Feeltheburns attempts at rewriting meteorology to support this feeling is a huge stretch, literally an immaginative jump. And as such, pretty silly.

And look at the graph again, and compare it to this famous graph, copied from tamino's webpage. It's obvious, isn't it, that there is a change in trend between the 90s, the noughts and the teens? The noughts are obviously stalling.



But no, Tamino has shown again and again that statistically there was no pause in global warming in the noughts - it just looks that way because of the randomness of natural variability.

And the same goes for the changes in SIE minimums over the last four decades - randomnes and natural variability are just as good an explanation as any hypothetical state changes leading to changes in melting rate.

https://tamino.wordpress.com/

That graph us an excellent example.  Would you use a simple linear curve for that data?  Of course not, it is polymeric in nature.  The R2 would tell you that.  Just like Gerontocrat has been trying to tell you with his graphs.  Using a simple linear regression does not tell the whole story.  Does the line in his graphs tell the whole story?  Probably not.  However, it is a better analysis than a straight line.  Selective curve fitting to match ones own beliefs, is one of the reasons that several people have erroneously predicted that the Arctic would be ice-free by now.  The Arctic sea ice minimum has not decreased recently.  If you refuse to see that from the data, there is nothing more that I can say.

[oren]

Geronocrat,
I agree that the ice has been generally in decline for the past four decades.  This occurred slowly at first, as the maximum sea ice started to fall.  The minimum was declining only slightly until the late 90s, when it plummeted.  This led to ab overall ice declined at a much faster rate until recently, when it has slowed considerably.  This occurred during the time that many pundits predicted an accelerated loss in Arctic sea ice to near ice-free conditions this decade.    All I am saying is that that his explanation is plausible explanation for the change.

At what point did FTB's description include anything about a change in the described mechanism? Evaporation from the ocean is nothing new.  Why would it suddenly halt sea ice loss? Regardless of whether there has been or hasn't been a hiatus in ice loss, his description doesn't explain anything at all.
KK, do you believe global ocean heat content has been rising, or is it stable maybe? If it is rising, FTB's description is irrelevant.

[Klondike Kat]

Geronocrat,
I agree that the ice has been generally in decline for the past four decades.  This occurred slowly at first, as the maximum sea ice started to fall.  The minimum was declining only slightly until the late 90s, when it plummeted.  This led to ab overall ice declined at a much faster rate until recently, when it has slowed considerably.  This occurred during the time that many pundits predicted an accelerated loss in Arctic sea ice to near ice-free conditions this decade.    All I am saying is that that his explanation is plausible explanation for the change.

At what point did FTB's description include anything about a change in the described mechanism? Evaporation from the ocean is nothing new.  Why would it suddenly halt sea ice loss? Regardless of whether there has been or hasn't been a hiatus in ice loss, his description doesn't explain anything at all.
KK, do you believe global ocean heat content has been rising, or is it stable maybe? If it is rising, FTB's description is irrelevant.

Globally, yes.  That may not be true in the Arctic.  The lessening of Arctic sea will cause more evaporative losses.  Heat is lost more readily at the poles than elsewhere.

[oren]

I wonder then how it is that seas which lose their sea ice cover earlier (such as the Chukchi) have warmer SSTs than in the past. You would think that this newfound evaporation scheme would cool them somehow.
Note: I don't have access to a ready-made chart for this claim, but am pretty sure it is true based on following the melting seasons.

[Glen Koehler]

  The Arctic sea ice minimum has not decreased recently.  If you refuse to see that from the data, there is nothing more that I can say.

KK - depends on what you mean by "recently."  Given the internal variability of the system, a 7-year or 10-year window (i.e. since 2012, 2008) is too short to draw conclusions.  To protect against cherry picking and our innate human tendency for "pattern matching" where there is none, a longer interval of at least 20 years is required. 

     The 20-year downward trend in annual minimum Arctic sea ice volume appears to be much larger than year-to-year variation, and thus appears to be statistically significant.  Maybe somebody with more time and skill will run the stats on 1998-2018 annual min. volume data.  Moreover, 2019 is likely to end up with less the 4M km3, and thus land right on and thus reinforce the PIOMAS trend shown at
http://psc.apl.uw.edu/wordpress/wp-content/uploads/schweiger/ice_volume/BPIOMASIceVolumeAprSepCurrent.png

    I think focusing so much on Extent leads us to false conclusions on both the high and low side.  2012 was bad for volume, but not nearly as extreme as it was for Extent.  To a substantial degree, Extent reflects temporary wind and current conditions, not the true status of the Arctic sea ice.  By focusing on Extent we can under-represent the cumulative progressive effects on overall condition of the ice. 

     The ice year this year has looked weaker more fractured than in past years.  A subjective measure to be sure, but not to be fully discounted either.  It will be interesting to see how the MYI numbers look after the 2019 melt season.  My guess is that they took another downward hit such that we are getting very close to only 1st and 2nd year ice that is more vulnerable than the tough old 5- and even 10-year old ice of the recent past.  The final phase of decline in ASI Extent is slowed by the much more rapid regrowth of 1st year ice.  But the flip side of that is that 1st year ice also melts out faster.   In retrospect, the functional loss of the Beaufort Sea ice nursery may be seen as a crucial event marking the beginning of the end of ASI in Aug-Sept.   I suspect with only 1st-year ice to get rid of at the start of future melt seasons, it won't last much beyond July 31.

     That's how it looks to me anyway.  No formal training in Arctic or ice, just a little experience analyzing numbers and interpreting science added to lurking in the ASIF.  Thanks to those who provide interesting info and speculation to ASIF.  Esp. thanks to Neven, Gero, JCG, uniquorn, Alphabet and others. Less gratitude for those who engage in extended ego-defensive debates.  To those folks I say, let others disagree with you and get over it.  Nobody else cares if somebody insults you.  Now you can disagree with me! 

[gerontocrat]

I wonder then how it is that seas which lose their sea ice cover earlier (such as the Chukchi) have warmer SSTs than in the past. You would think that this newfound evaporation scheme would cool them somehow.
Note: I don't have access to a ready-made chart for this claim, but am pretty sure it is true based on following the melting seasons.

Google provides...
https://arctic.noaa.gov/Report-Card/Report-Card-2015/ArtMID/5037/ArticleID/220/Sea-Surface-Temperature

A bit out of date (only to 2015) but I am sure given recent data that the only way was up.

The Chukchi Sea and eastern Baffin Bay show the largest ocean surface warming trends; August SSTs are increasing at ~0.5°C/decade in these regions.

[gerontocrat]

Oren - you sent me on the hunt for data....

A bit more... from NOAA Report Card 2018
https://www.arctic.noaa.gov/Report-Card/Report-Card-2018/ArtMID/7878/ArticleID/779/Sea-Surface-Temperature

Mean August SSTs from 1982-2018 show warming trends over much of the Arctic Ocean; statistically significant (at the 95% confidence interval) linear warming trends of up to +1° C decade-1 are observed (Figs. 1d and 2). These warming trends coincide with declining trends in summer sea-ice extent (including late-season freeze-up and early melt, e.g., Parkinson, 2014), increased solar absorption (e.g., Timmermans et al., 2018), and increased vertical ocean heat transport (e.g., Lind et al., 2018).

[oren]

Thank you G.

[Klondike Kat]

  The Arctic sea ice minimum has not decreased recently.  If you refuse to see that from the data, there is nothing more that I can say.

KK - depends on what you mean by "recently."  Given the internal variability of the system, a 7-year or 10-year window (i.e. since 2012, 2008) is too short to draw conclusions.  To protect against cherry picking and our innate human tendency for "pattern matching" where there is none, a longer interval of at least 20 years is required.

Glen, I fear you may be the one engaging in cherry picking.  By choosing a start date of 1998, you are ensuring the largest possible decreasing trend.  Contrarily, I choose the entire 40/year dataset.   The 7- or 10-year window only appears too short to you, because it does give the trend you want to see.  That is one of the main reasons I prefer analyses by well-established scientists over others.

[nanning]

<snip>
Nobody else cares if somebody insults you.

Good post imo Glen but I disagree with this last bit because I do care.
If somebody uses nasty language, impolite insults, name calling etc. they don't get a like from me, even if I think the rest of the post is really good.
I don't think it has any influence but it is my way. In some cases I'll post a question about it.

OK back to topic, sorry.

[binntho]

That graph us an excellent example.  Would you use a simple linear curve for that data?  Of course not, it is polymeric in nature.  The R2 would tell you that.

I doubt very much if it is "polymeric" in "nature" - whatever that means. Can a graph have a "nature"? And if you are so sure of the R2, why not try yourself?

Me, I am nowhere near good enough at statistics to make such claims, but I have been following Tamino's blog for some years, and he is a true master of the art.

His latest post is actually about this very thing, i.e. how has the rate of warming changed over the last 150 years or so. And funnily enough, he doesn't even mention the "polymeric nature" of the graph.

Here, on the other hand, is that graph with some statistically valid change points and linear trends connecting them. This is apparently statistically the best fit to the data. (https://tamino.wordpress.com/2019/08/20/global-warming-how-fast/)

 Just like Gerontocrat has been trying to tell you with his graphs.  Using a simple linear regression does not tell the whole story.  Does the line in his graphs tell the whole story?  Probably not.  However, it is a better analysis than a straight line.

If Gerontocrat tried to tell me something, I'm sure that he would succeed. And I do not recall him trying to tell me that graphs have "polymeric" natures, or that his polymeric line is a "better analysis" than a straight line. So what does Gero himself have to say about his polymeric red lines?

But I don't like these red trend lines either. These periodic large rises and falls are just as well represented by an obviously untrue linear projection.

Which is perfectly reasonable. A linear trend is simply the only thing that matches the data we have in any statistically valid sense. And before somebody points this out, I am well aware that the R2 values for the polynomials are slightly better than for the linears, but that is still not valid since these graphs are running averages.

If you run linear and polynomial regressions on the raw data you get identical R2 values of c.a. 0.037 (because of the very large variance, the R2 is quite low), no visible trend changes in the polynomial within the period in question, and a downward trend of 73.000 km2 per annum for both.

Funnily enough, this graph plotting the NSIDC values shows a downward trend of 81.100 km2 per annum, while Gero's SIE 365 day running average linear regression shows a downward trend of 67.291 km2 per annum.


Missing from the graph is the formula y = -0.0811x + 168.03 and an R2 of 0.7803.


The problem arises when people try to extend the linear trend into a linear projection and get different answers from different graphs. So clearly, the linear trends cannot be used to project the zero point. That does not mean that other strangely looking swirly polynomials are going to do any better. To repeat: There are not enough data points in the minimum graph, and the variance is too big to justify anything other than a linear trend.

Selective curve fitting to match ones own beliefs, is one of the reasons that several people have erroneously predicted that the Arctic would be ice-free by now.  The Arctic sea ice minimum has not decreased recently.  If you refuse to see that from the data, there is nothing more that I can say.

We can agree that selective curve fitting is a no-go. But I can only agree with your statement that the minimum has not "decreased recently" if we always define "recently" as "since the last record minimum".

The problem with your statements, KK, is that you think you can see trend changes in the graph. I claim that the graph has too few data points and too much variance to justify such claims.  So the "not decreased recently" is a real thing for now, but has no actual meaning - it is not possible, based on the data, to make claims about a statistically valid change in trend.



That is not to say that there isn't a change in the underlying forces that interact in the high Arctic and give us minimums every year. Perhaps something has changed that makes new record minimums very unlikely, or perhaps that was always built into the system, that the graph would eventually flatline around this level.

And yes, just looking at the graph of SIE minimums can well make you think that this is what is happening. But there is no statistically valid basis to such claims, and probably won't be for another 10 years or so. If, in 2029, we are still hovering around the 4M mark, a change in trend will be statistically valid. Today it is not.

And when you seek weird explanations (such as Feeltheburn's erraneous understanding of meteorology) to explain your non-statistically-valid claims, then we are way into the land of silliness.

[blumenkraft]

Kat, what you see as a stalling is most likely variance. Variance is expected in a highly complex/chaotic system. Given the context, it's extremely unlikely that it is indeed a stalling.

[Klondike Kat]

Kat, what you see as a stalling is most likely variance. Variance is expected in a highly complex/chaotic system. Given the context, it's extremely unlikely that it is indeed a stalling.

If it were truly variance, it should be random.  Look at the graph and trend posted by binntho.  From 96 to 06, all the variance was above the trend.  Then, there were six straight years below the trend, some substantially.  There is no physical reason that the ice should follow a linear line.  Perhaps it is just variance, but it has shown variance in a flattening direction for over a decade now.  How long must it occur, before it becomes acceptable?

[binntho]

Kat, what you see as a stalling is most likely variance. Variance is expected in a highly complex/chaotic system. Given the context, it's extremely unlikely that it is indeed a stalling.

If it were truly variance, it should be random.  Look at the graph and trend posted by binntho.  From 96 to 06, all the variance was above the trend.  Then, there were six straight years below the trend, some substantially. 

Even if it is possible for the human mind to see patterns doesn't mean that they aren't the product of random variance. Co-incidence and randomness could indeed well produce the pattern we all see in the graphs.

This whole discussion really revolves around the number of datapoints. With only 40 datapoints it's not possible to claim anythying statistically other than that a linear trend is a fairly good fit.

There is no physical reason that the ice should follow a linear line.  Perhaps it is just variance, but it has shown variance in a flattening direction for over a decade now.  How long must it occur, before it becomes acceptable?

There is a very good physical reason for the ice to follow what seems to be a linear trend over the 40 year period - global temperatures have been rising with a linear trend during the same time, and lower SIE is a direct consequence of higher temperatures.

As can be seen in my earlier graph, the closest we can say statistically about the temperature change over the 40 year period is that it has risen linearly. Therefore we should a priori expect sea ice extent to fall linearly.

On the other hand, we can all come up with lots of reasons for why the rate of diminishing SIE or rate of falling minimums etc. should not be linear, at least not always and under all circumstances. The drastic loss of MYI is certainly one of the most obvious ones, the shape of the Arctic ocean another.

But there are lots of other factors at play, such as the relentlessly rising ocean temperatures, the increasing storminess, and the essential randomness of weather coupled with the apparent sensitivity of the melting season to the "right" type of weather at the "right" time.

So at least from my point of view, declaring the linear trend to be dead and void is much too premature. I'll repeat what I said earlier - let's wait another 10 years, and then we'll (perhaps) be able to tell if there really was a change in trend 10 years ago.

[Klondike Kat]

Kat, what you see as a stalling is most likely variance. Variance is expected in a highly complex/chaotic system. Given the context, it's extremely unlikely that it is indeed a stalling.

If it were truly variance, it should be random.  Look at the graph and trend posted by binntho.  From 96 to 06, all the variance was above the trend.  Then, there were six straight years below the trend, some substantially. 

Even if it is possible for the human mind to see patterns doesn't mean that they aren't the product of random variance. Co-incidence and randomness could indeed well produce the pattern we all see in the graphs.

This whole discussion really revolves around the number of datapoints. With only 40 datapoints it's not possible to claim anythying statistically other than that a linear trend is a fairly good fit.

There is no physical reason that the ice should follow a linear line.  Perhaps it is just variance, but it has shown variance in a flattening direction for over a decade now.  How long must it occur, before it becomes acceptable?

There is a very good physical reason for the ice to follow what seems to be a linear trend over the 40 year period - global temperatures have been rising with a linear trend during the same time, and lower SIE is a direct consequence of higher temperatures.

As can be seen in my earlier graph, the closest we can say statistically about the temperature change over the 40 year period is that it has risen linearly. Therefore we should a priori expect sea ice extent to fall linearly.

On the other hand, we can all come up with lots of reasons for why the rate of diminishing SIE or rate of falling minimums etc. should not be linear, at least not always and under all circumstances. The drastic loss of MYI is certainly one of the most obvious ones, the shape of the Arctic ocean another.

But there are lots of other factors at play, such as the relentlessly rising ocean temperatures, the increasing storminess, and the essential randomness of weather coupled with the apparent sensitivity of the melting season to the "right" type of weather at the "right" time.

So at least from my point of view, declaring the linear trend to be dead and void is much too premature. I'll repeat what I said earlier - let's wait another 10 years, and then we'll (perhaps) be able to tell if there really was a change in trend 10 years ago.

I will agree with much of what you says.  40 data points is not possible to claim anything statistically.  Although, we can say with much confidence that today is lower that 40 years ago.  Focusing on just the data from the annual minimum is not the best method of analyzing the sea ice, but it tends to get the most attention.   Yes, many factors are in play, such that correlating sea ice directly to global temperatures may not be the best practice.  I have no reason to believe that the sea ice will show a flat trend for another 10 years, but if it does, than I would expect most to accept the change in trend.  The sharp decline in sea ice has only shown a ~15-year trend, so they would be on par. 

One final point.  Sea ice analyses, no matter which metric, treats the entire ice as one.  In reality, it is made up of several small seas, bays, etc., and one large central ocean.  Much of the previous changes occurred largely in the peripheral seas, and less so in the central Arctic basin.  We may just be seeing a shift to the new regime.

[binntho]

I will agree with much of what you says.  40 data points is not possible to claim anything statistically.  Although, we can say with much confidence that today is lower that 40 years ago. 

To be specific: 40 datapoints are not enough to decide whether there has been a change in trend. But it's plenty sufficient to show that there is a downward trend!

Focusing on just the data from the annual minimum is not the best method of analyzing the sea ice, but it tends to get the most attention.   

Absolutely, and the graphs that Gerontocrat has made from the 365 day averages are vastly to be preferred.

Yes, many factors are in play, such that correlating sea ice directly to global temperatures may not be the best practice.  I have no reason to believe that the sea ice will show a flat trend for another 10 years, but if it does, than I would expect most to accept the change in trend.  The sharp decline in sea ice has only shown a ~15-year trend, so they would be on par. 

If the trend flatlines for the next 10 years then I think that it's going to be pretty obvious, and statistically valid as well. And if there is an underlying dynamic that has cause a stall the last 10 years, why shouldn't it continue?

On the other hand, if the trend does not flatline then a hypothetical change in trend becomes unproven - but the underlying dynamics may still have been at work, flatlining for a period, but too hidden by the noisiness of the signal to be statistically detectable.

And your last comment, that the sharp decline in sea ice has shown a ~15 year trend, is again not something that cannot be stated with any statistical validity. There is no 10 or 15 or 20 year trends that can be read out of the minimum graphs, or the 365 day average graphs.

The only trend that can be read out of the graphs is that sea ice has been declining, and as far as the statistics go, basically linearly for the last 40 years.

One final point.  Sea ice analyses, no matter which metric, treats the entire ice as one.  In reality, it is made up of several small seas, bays, etc., and one large central ocean.  Much of the previous changes occurred largely in the peripheral seas, and less so in the central Arctic basin.  We may just be seeing a shift to the new regime.

Yes that's absolutely true, as I indicated earlier by mentioning the shape of the Arctic ocean. And a hypothetical "shift to a new regime" could well be an explanation for a hypothetical "stall". And if there is indeed such a shift then I would expect the stall to easily last the next 10 years.

[blumenkraft]

If it were truly variance, it should be random.

OK, you don't understand what randomness is. That's a common problem with humans.

If you roll a fair, 6-sided die, there is an equal probability that the die will land on any given side.

Meaning, if you roll a dice, you have always the chance to roll a 6. So, if you roll 20 times a 6 in a row, it is ... well ... completely random!

[crandles]

To be specific: 40 datapoints are not enough to decide whether there has been a change in trend. But it's plenty sufficient to show that there is a downward trend!

The number of datapoints needed does of course depend on how much the trend changes, the level of noise in the data, and perhaps other factors.

Also 40 datapoints is not all we have got. We have other months data which also show similar shapes. How much use that is may depend on autocorrelation, but I think there isn't much autocorrelation between maximum and minimum.

So anyone want to venture a calculation of when we should be able to say change of trend is statistically significant rather than just plucking 10 years out of the air? 10 years seems a bit long to me if the difference in trend rate is noticeably different. If new data points tend to be just above long term linear rate then it will take a long time, probably longer than 10 years. So the answer may need to be in terms of 'if x difference then y years but if z different then w years. So maybe a better question is at what statistical level of confidence are we at now? (Where not reasonably established until 95% confidence reached.)

[Klondike Kat]

If it were truly variance, it should be random.

OK, you don't understand what randomness is. That's a common problem with humans.

If you roll a fair, 6-sided die, there is an equal probability that the die will land on any given side.

Meaning, if you roll a dice, you have always the chance to roll a 6. So, if you roll 20 times a 6 in a row, it is ... well ... completely random!

Really?  I would question your die.  That odds of that occurring is less than 1 in a trillion.  You call that random, and have the audacity to question my understanding of randomness?!

[blumenkraft]

That odds of that occurring is less than 1 in a trillion.

And? What has odds to do with how the dice will fall?

The dice can't remember it's recent states and it has no understanding of the concept of 'odds'. It just obeys physics.

I'm having a bitcoin mining background which is the ultimate lesson in understanding randomness. I got this, Kat, believe me.

[Klondike Kat]

That odds of that occurring is less than 1 in a trillion.

And? What has odds to do with how the dice will fall?

The dice can't remember it's recent states and it has no understanding of the concept of 'odds'. It just obeys physics.

I'm having a bitcoin mining background which is the ultimate lesson in understanding randomness. I got this, Kat, believe me.

What does the odds have to do with how the dice fall?  Everything!  While each individual throw has the same odds, over time, they will even out, each number occurring nearly equally.  Straight-forward mathematics.  Should one number dominate against all odds, I would seriously question the integrity of the die.  This has nothing to do with physics, unless the die is loaded, causing the number 6 to occur repeatedly.

[blumenkraft]

each individual throw has the same odds

Stop right there. This is really all there is to it.

It's not math. Math is a tool to calculate the odds. The math doesn't make the odds.

[Klondike Kat]

each individual throw has the same odds

Stop right there. This is really all there is to it.

It's not math. Math is a tool to calculate the odds. The math doesn't make the odds.

If you truly understand the odds, then why are you having such a hard time grasping the improbability of 20 straight sixes?

[blumenkraft]

If you truly understand the odds, then why are you having such a hard time grasping the improbability of 20 straight sixes?

I'm not talking about the odds, Kat. You do. I'm talking about randomness.

You can't know the odds concerning the arctic area/extent numbers. And here we go full circle.

You can though, given the context, make assumptions about the odds. And they tell me this is most likely variance.

[Klondike Kat]

If you truly understand the odds, then why are you having such a hard time grasping the improbability of 20 straight sixes?

I'm not talking about the odds, Kat. You do. I'm talking about randomness.

You can't know the odds concerning the arctic area/extent numbers. And here we go full circle.

You can though, given the context, make assumptions about the odds. And they tell me this is most likely variance.

So, depending on ones assumptions, determined ones assessment.  Hence, they is little agreement.