Support the Arctic Sea Ice Forum and Blog

Author Topic: Volume? Extent?  (Read 7877 times)

Dundee

  • New ice
  • Posts: 44
    • View Profile
  • Liked: 6
  • Likes Given: 0
Volume? Extent?
« on: August 04, 2016, 06:50:45 PM »
From the IJIS thread -

Quote
Now that sea ice extent has strung together a string of century drops, I wonder about the thoughts of those who have been claiming extent is a meaningless/useless metric.
Off to make a cherry pie.

I don't think extent is meaningless....but it is CERTAINLY LESS MEANINGFUL than volume.  Volume has ALWAYS been the most important measure in my mind.
 
No need to shout.

Your speculation and arguments are not unreasonable, however.

Volume? Extent? I don't think anyone is going to win this argument but I also doubt that it intrinsically is an either/or question.

Getting beyond beliefs (I am always wary of those) some fundamentals apply.

No one currently measures either concentration or extent over the whole region. Folks measure microwave emissions at some combination of wavelengths, polarities, and resolution. Based on correlations (comparatively crude ones, if you measure their success against high resolution imagery) they publish rough estimates of extent. Based on a longer list of physical measures, input into a computer model, other folks publish even rougher estimates of volume. Any one of these estimates can be compared against itself over time, or (after cross comparison of years of data) with other estimates, giving some indication if the status quo has changed or not. Nobody is counting ice flows and subtracting those less than 15% over the entire Arctic. If you try it in any particular spot, comparison with the pantheon of published numbers are likely to be controversial. Used carefully, these comparisons can always provide insight but the further conditions on any given day are from the conditions on which the correlation or model was designed, the less the resulting numbers have to do with physical reality.

Back to reality (which we cannot currently measure), ice volume is a exact indicator of the locked in latent heat of fusion that would be needed to return the Arctic to liquid. The heat imbalance in the Arctic always operates against ice volume - a given number of joules results in a given amount of ice gained or lost. Extent (or 2D measurement based on whatever ice fraction cutoff you select) is the driver of the effect of ice. At its thickest, the depth of the pack is small compared to the ocean layer it influences. Ice sets the temperature of nearby water to the freezing (melting) point - 2D measures of ice drive what area of the ocean is stable at the local freezing point, and what area is free to store sensible heat as rising water temperature. Presence of the thinnest frazil or the deepest keel both mean nearby water is at the local freezing point (for the local salinity). Solar Albedo is also primarily a function of 2D area of ice. There are lots of second order effects but in all but thin ice, solar albedo is not strongly affected by thickness. In the winter season thickness does strongly affect net heat flow from the ocean (and over-ice air temperature), but that effect changes little as thickness moves into the multi-year range.

Heat added to or removed from ice always acts on its volume, but (particularly during the melting season) the most significant physical effects (including most negative feedback loops) of having ice or not are generally a function of its physical 2D presence.

Lots of strange things happen in the space between mature, unbroken ice and open water (the marginal ice zone - the part of the Arctic where ice is present but where wave action still occurs). The changing fraction and behavior of part of the pack which can be considered MIZ is hugely interesting. Both obviously exert influence but changes in the MIZ are not related in a simple or direct way to either extent or volume. It doesn't help that in the MIZ, both volume and extent are harder to measure with certainty. The MIZ is a function of local mechanical weather energy available, and local mechanical strength of ice available to resist it. A mild storm acting on 50cm ice can break up the same area as a huge storm over cold 3m ice. The effect (on ocean temperature and albedo) of a given concentration and size distribution of floes is not very sensitive to whether they are 50cm or 3m thick. Ice is there until it is not (which may help explain the lack of dramatic secondary effects from the huge decrease in volume in 2012).

I would hope that we don't get so wrapped up in details that we lose sight of the many (disproportionately influential) people in the world who claim nothing is changing at all, and certainly not because of CO2 that we have been returning to the atmosphere at an ever greater rate.

magnamentis

  • Guest
Re: Volume? Extent?
« Reply #1 on: August 04, 2016, 07:20:33 PM »
hat off for the great read provided :-)

binntho

  • Nilas ice
  • Posts: 2193
    • View Profile
  • Liked: 878
  • Likes Given: 235
Re: Volume? Extent?
« Reply #2 on: August 05, 2016, 07:20:45 AM »
hat off for the great read provided :-)
Second that!
because a thing is eloquently expressed it should not be taken to be as necessarily true
St. Augustine, Confessions V, 6

johnm33

  • Guest
Re: Volume? Extent?
« Reply #3 on: August 05, 2016, 10:57:46 AM »
" ice volume is a exact indicator of the locked in latent heat of fusion that would be needed to return the Arctic to liquid"
Only if it's safe to assume there's no significant fraction of snow present.

jdallen

  • Young ice
  • Posts: 3410
    • View Profile
  • Liked: 650
  • Likes Given: 244
Re: Volume? Extent?
« Reply #4 on: August 05, 2016, 04:33:08 PM »
" ice volume is a exact indicator of the locked in latent heat of fusion that would be needed to return the Arctic to liquid"
Only if it's safe to assume there's no significant fraction of snow present.
Snow is trivial in that regard; on 2 meter thick ice, a full meter of snow would only increase the energy required by about 5% at most.
This space for Rent.

johnm33

  • Guest
Re: Volume? Extent?
« Reply #5 on: August 06, 2016, 12:17:15 AM »
" would only increase the energy required "
I may be wrong but I'm thinking that maybe 1/20 - 1/2 of volume was laid down as snow, it looks like ice to the sensors because it either got rained on, had freezing fog pass over it or caught the salt released from frozen spray. So volume possibly is not an accurate measure of mass, likely an overestimate and the ice is thus rather more vulnerable than assumed. Just food for thought.

Nick_Naylor

  • Frazil ice
  • Posts: 291
    • View Profile
  • Liked: 0
  • Likes Given: 0
Re: Volume? Extent?
« Reply #6 on: August 06, 2016, 02:02:20 AM »
That is an interesting question: do we know whether PIOMAS is estimating an uncorrected or an "ice-equivalent" volume? i.e., if PIOMAS knew we had 2 meters of snow on top of 1 meter of ice over a 1 km^2 area, what would they report as the volume?

Dundee

  • New ice
  • Posts: 44
    • View Profile
  • Liked: 6
  • Likes Given: 0
Re: Volume? Extent?
« Reply #7 on: August 08, 2016, 11:52:00 PM »
Snow is a bit of a red herring (arguably, one of those "details" I complained about in my opening post). Not because it doesn't matter (it is a prime driver of both albedo and heat conductivity between air and the ice-water interface) but because it is a minor contributor to the total mass/volume of ice in the Arctic. Snow matters to most models (including PIOMAS and altimetry approaches to thickness/volume) and where it matters most it is dealt with explicitly. All else being equal, the effect of snow is directly proportionate to ice extent .

My comments were directed to the contrast between extent and volume from first principles. I get the impression what many people are really interested in is which of the two gives a better idea of how the ice season is going.

A byproduct of a project I've been fooling with for a while sheds a bit of light on the question. The project involved developing detailed statistical projections for a whole bunch of stuff. By comparing the difference between the projections over time with real life (residuals) and the accuracy of predictions of NSIDC September extent, you get some idea of what has predictive value and what does not.

Below is a graph of selected indicators - I did a grunch more, but these give a fair idea of what is going on. The curves have been filtered to reduce daily chaos while keeping the basic trends.

Very early on in the year, it is a total crapshoot. By May, PIOMAS volume gives some indication how the season might go. In early June, NSIDC area becomes the best predictor, overtaken in late July by NSIDC extent. I expected the behavior of the core of the Arctic to be a quality indicator, but it is not. It looks plausible early in the year, but extent and area are very nearly 100% then, the curve reflects noise between very small residuals.

The standard deviation of error for a plain straight line fit for the NSIDC September SII is about 550k km2. Adjusting the estimate by NSIDC extent for August 7th (vs the long term trend) cuts the error to about 250k, which is not too shabby.

So, the answer to volume vs extent as a predictor depends on what time of year you are asking, and before May, you can take your pick - neither tells you anything about how September will play out.

 

Nick_Naylor

  • Frazil ice
  • Posts: 291
    • View Profile
  • Liked: 0
  • Likes Given: 0
Re: Volume? Extent?
« Reply #8 on: August 09, 2016, 01:04:58 AM »
Snow matters to most models (including PIOMAS and altimetry approaches to thickness/volume) and where it matters most it is dealt with explicitly.

Explicitly how? If we have 2 meters of snow on top of 1 meter of ice, what thickness does PIOMAS report?

Dundee

  • New ice
  • Posts: 44
    • View Profile
  • Liked: 6
  • Likes Given: 0
Re: Volume? Extent?
« Reply #9 on: August 09, 2016, 05:29:24 AM »
PIOMAS treats snow explicitly, from the standpoint of thermodynamic effects and mass of frozen water. PIOMAS is primarily a thermodynamic and mechanical model - inputs include NSIDC gridded ice concentration, sea surface temperature and atmosphere information from NECP/NCAR reanalysis (National Environmental Prediction Center/National Center for Atmospheric Research). PIOMAS does not get ice thickness information, it derives it through heat calculations and mechanical treatment of ridging. More information is at the following links, with an email or two you should be able to get as many details of the model as you are willing to digest:

http://psc.apl.uw.edu/research/projects/arctic-sea-ice-volume-anomaly/

http://psc.apl.uw.edu/research/projects/projections-of-an-ice-diminished-arctic-ocean/model/

The most extensive source of ice thickness measurement is from submarines, measuring the depth of the ice base with an upward looking sonar - these measurements are stated in terms of equivalent thickness of ice - irrespective of whatever entrained air may be present (it is essentially a mass based measurement - the observed mass of seawater displaced is translated to an equivalent thickness of ice). Surface measurements usually report snow and ice thickness separately. Altimetric measurements of freeboard, however, need to know or assume the depth and density of snow cover when they estimate overall ice thickness. The issue is not the few centimeters of additional volume (entrained air) of the snow layer, but the many centimeters of additional ice that would be assumed to be below the surface, based on the observed freeboard.

The Arctic is essentially arid. The NASA data for snow over sea ice only finds it necessary to display to 50cm, overall Arctic snow tops out at about 35cm.  During the summer peak melting season, snow cover is all but zero. The water (ice) equivalent of even the deepest snow in the Arctic is something like 15cm - not enough to drive the uncertainty in PIOMAS. A good treatment of Arctic snow climatology is at:

http://www.atmos.washington.edu/sootinsnow/PDF_Documents/Warren%20et%20al%201999%20snowdepth.pdf

The short answer to your question is PIOMAS would never need to deal with two meters of snow. The model considers the thermal conductivity of snow and includes a component that properly accounts for mass changes in the snow layer. I believe that PIOMAS includes the mass of frozen water within its snow layer in the mass of ice that it reports as volume, but cannot say for certain and have no particular interest in digging the answer out of the source code. There may be another reader here who can clarify this detail. In any event, compared to the published uncertainty of PIOMAS estimates and the large changes occurring in ice volume, it makes little difference.

The biggest impacts of snow have nothing to do with the comparatively small amount of mass/volume involved. There is very little snow in late summer - it is wetted out and refrozen as proper ice, or drained away into the sea. About half of the year's snow falls in September and October. As minimum extent declines, more and more snow falls on open water, less and less overlays young peripheral ice formed after peak snowfall. With a less effective insulating layer, this ice freezes initially freezes faster but the freezing rate promptly falls off as the ice thickens, freezing through the entirety of the winter season does not increase a great deal. On the other hand, the water table reaches the top of the snow faster in the spring, exposing melt ponds and wetted surface to the sun sooner - the impact of this early albedo change is not self limiting but persists through the melting season.

jdallen

  • Young ice
  • Posts: 3410
    • View Profile
  • Liked: 650
  • Likes Given: 244
Re: Volume? Extent?
« Reply #10 on: August 09, 2016, 07:47:25 AM »
Snow matters to most models (including PIOMAS and altimetry approaches to thickness/volume) and where it matters most it is dealt with explicitly.

Explicitly how? If we have 2 meters of snow on top of 1 meter of ice, what thickness does PIOMAS report?
Hmmm. Can you show us an example of more than 50CM of snow on top of sea ice?
This space for Rent.

Nick_Naylor

  • Frazil ice
  • Posts: 291
    • View Profile
  • Liked: 0
  • Likes Given: 0
Re: Volume? Extent?
« Reply #11 on: August 09, 2016, 01:02:30 PM »
It was not intended to be a realistic scenario. It was intended to clearly frame the question: When snow represents a significant percentage of total volume, how does PIOMAS volume numbers represent that snow?

I'm quite sure it's not a burning question - especially in August - but it's hardly an esoteric thing to wonder about.

Dundee

  • New ice
  • Posts: 44
    • View Profile
  • Liked: 6
  • Likes Given: 0
Re: Volume? Extent?
« Reply #12 on: August 09, 2016, 07:02:36 PM »
If your question was not intended to represent a realistic scenario, what was its intent? Not esoteric? Perhaps you could describe the circumstances in which a significant fraction of sea ice could be in the form of snow? Or think through your original question. Long before PIOMAS has the opportunity to analyze it, what would physically happen to one meter of floating ice if you placed two meters of snow on top of it? I'll offer a hint - the low range of observed Arctic snow density runs something upwards of 0.25gm/cc, trending above 0.3 by Spring. Even better, if you have a deep freeze, salt, and something to grate ice with, you can physically perform an appropriately scaled experiment.
 
It would be interesting to frame up a set of NECP/NCAR reanalysis data representing the biblical flood and push it through PIOMAS. It would not tell us anything about the real world, but it would certainly be interesting.

To review volume and extent, the actual physical volume of crystalline ice on Arctic waters responds linearly to the heat balance at the boundary between water and ice (whatever its form). This balance is forced by the heat balance of the Arctic generally. The environmental effects of ice (including feedbacks) are, for the most part, a function of the actual geographic area of the sea/atmosphere interface that is interrupted by ice. PIOMAS models the former, sea ice extent is a convenient convention for characterizing the latter. Both metrics are subject to very significant local errors as well as overall uncertainty that is less significant and reasonably well documented. These measures are most useful for their consistency and thus their ability to reliably show that a long term change is occurring. More recent high resolution measurements will not be able to do this until they also have 37 years of consistently collected data. Of note, as improvements are made in the PIOMAS model, the entire NSIDC/NECP/NCAR record is reprocessed to provide continuity in the revised record, and NSIDC is much more concerned with matching new instruments with  results from the previous record of radiometry than with any new objective measure of performance vs actual ice conditions.

Statistically, published measures of ice from October through April have little correlation with the next September minimum. In May, PIOMAS volume begins to correlate meaningfully to the upcoming minimum. Through June and July, overall area is the most useful leading indicator and beginning in August extent itself is the best indicator (among daily ice measurements) for how the SII minimum will turn out. (I found this result interesting, even if it apparently does not rise to the level of improbably snowy hypotheticals).

Nick_Naylor

  • Frazil ice
  • Posts: 291
    • View Profile
  • Liked: 0
  • Likes Given: 0
Re: Volume? Extent?
« Reply #13 on: August 09, 2016, 11:41:28 PM »
Are you saying you can't answer my question? If so, fine, but I'm quite certain there is nothing wrong with trying to understand what is meant by "volume".

Nick_Naylor

  • Frazil ice
  • Posts: 291
    • View Profile
  • Liked: 0
  • Likes Given: 0
Re: Volume? Extent?
« Reply #14 on: August 10, 2016, 02:12:52 AM »
If your question was not intended to represent a realistic scenario, what was its intent? Not esoteric? Perhaps you could describe the circumstances in which a significant fraction of sea ice could be in the form of snow? Or think through your original question.

It appears that snow depth averaged across the Arctic reaches a maximum of 30cm around the end of May. This represents about 16% of PIOMAS reported thickness at that time, which I do believe justifies taking the question at least somewhat seriously. Not that there was any reason for anyone to be combative about it even if the ratio was smaller. 

http://journals.ametsoc.org/doi/pdf/10.1175/1520-0442(1999)012%3C1814%3ASDOASI%3E2.0.CO%3B2

johnm33

  • Guest
Re: Volume? Extent?
« Reply #15 on: August 10, 2016, 08:59:56 AM »
Some peices of ice seem to be have more 'sail' and move around in the wind much faster. I think it was first noticed on one of JayW s animations but the last animation in this post shows it very clearly, this calls for some sort of explanation and I think the simplest is an accumulation of snow on a thin bed of ice.
http://forum.arctic-sea-ice.net/index.php?action=post;quote=84751;topic=382.2400;last_msg=85886
In this post Wayne comments on how fast some of the residual ice moves which could be another hint that snow is present. http://eh2r.blogspot.co.uk/2016/08/massise-number-of-goodbye-waves-dont.html
With so much more open water and it being warmer we're likely getting more snow than previously too. http://eh2r.blogspot.co.uk/2016/08/thicker-sea-ice-look-for-less-snow.html