Effect of snow cover on thickness of arctic ice

[mati]

A posting on another thread led me to wonder if the later open sea in the arctic has led to a deeper snow cover on the ice, leading to thinner ice formation during the winter.

Here is the latest snow depth:

http://www.weatheroffice.gc.ca/data/analysis/352_100.gif

I know from experience on lake nipissing that when there is low snow cover (as it was this year) trucks and cars are out on the ice earlier and the ice is thicker and takes longer to melt out.

[Vergent]

there seems to a serious disagreement about high arctic snow depth.

[mati]

I did find some raw data, but i'm not very good at working out trends etc.  unfort this data only goes up to 2008:

http://neptune.gsfc.nasa.gov/csb/index.php?section=53

[Wipneus]

Hmm, the PIOMAS gridded data includes snow thickness.

I will see what I can do, it is not much work but I have very little time in the next few days.

[Wipneus]

First attempt is attached: graphing the total volume of snow (it is expressed in volume of water, so needs to be multiplied by some factor).

No time yet for adding a legend yet:

black - monthly values
purple - Jan smoothed
blue - April smooth
green - July smooth
red - Okt smooth

So total snow decreases in all months. That is partly because the ice covered area is declining.
Does it compensate? Wait for more graphs later, no more time today.

[edit: y-scale off factor 1000]

[mati]

Thanks, this is what i have seen so far in the research.
Declining snow cover attached to declining ice extent because less ice to sit on means less snow cover.  That seems to be what most research is about.
What would be interesting is to see if the amount of snow on the ice that is present at the time measurement is increasing or decreasing.
If it is increasing, one would expect more insulation, and less ice depth for the current ice present.
Not sure if i had my thoughts expressed well... LOL im just a computer scientist, not an expert here.
But i have practical experience in this area

[Wipneus]

Thanks, this is what i have seen so far in the research.

Well, I haven't searched at all, except for clues in the PIOMAS data. If you have found anything else than the link you posted, please share.

[Wipneus]

Attached snow thickness according to PIOMAS:

- no thickening to be seen
- May has the thickest snow cover, but also the steepest decline.

[crandles]

Presumably, there is ice which cannot support snow until later in the year so there is less time to accumulate snow.

Just wondering, if you split the arctic into east and west: Would East be losing more rapidly due to above effect? Would West be gaining snow depth through more precipitation wherever there is not much effect from having less time when snow could be supported?

Perhaps restricting areas even more would be better e.g. between pole and Ellesmere Island for 'West' versus East Siberian and Leptev Seas for 'East'.

Where in the PIOMAS data is the snow information?

[Wipneus]

Where in the PIOMAS data is the snow information?

The snow.Hyyyy files , see http://psc.apl.washington.edu/zhang/IDAO/data_piomas.html

I agree that regional trends are worth looking into.

[Wipneus]

Gridded AMSR-E snow thickness data is available from NSIDC upto the failure in Oktober 2011 http://nsidc.org/data/docs/daac/ae_si12_12km_tb_sea_ice_and_snow.gd.html

[Wipneus]

I think this will be even better than regional trends. Trends calculated per gridcell, over last 10 years where the concentration is >50% in all of those years.

Not much snow increase to be seen.

[crandles]

Very impressive. Thank you Wipneus. 

[Wipneus]

Thank you, feel free to give your interpretations.

I expected actually more snow, not less. There is some increase on the edges around the Barentsz, Kara, Baffin and Bering sea's. But the decreases in the rest of the icepack occupy more surface and are stronger.

[crandles]

Yes I thought there might be larger areas with increases in snow depth.

Given that poor performance my interpretations may not be worth much but...

Nov trends are likely less time when able to support snow. i.e. strong downward trend where ice is later in reforming.

Aug trend - now virtually no snow left whereas previously with more ice cover there was a little. DMI temps haven't risen making explanation a little more awkward. Latent heat transfer by evaporation from ocean to atmosphere which then quickly loses heat to any remaining snow?

For Feb and May:
As you say more snow around some of the edges. I was expecting an increase in precipitation. I think what is shown here may indicate a bigger temperature difference between ice cover and ocean causing precipitation to fall more heavily regionally where cloud moves over ice cover.

That doesn't explain increase by New Siberian Islands. I don't know what to make of that area.

East Greenland loss of snow is also wrong for explanation given. The ice is only a little thinner so I doubt it is just being able to support less snow. Cold this is mainly be just warmer temperatures? Wind direction changes from the almost constant highs over Greenland we have been seeing since 2007 would tend to keep the area cooler but perhaps it brings dry air and less snowfall.


Effects of the snow thickness changes:
More snow around edges probably helps with faster loss of thin ice around edges up to May but once that is gone the ice is then thicker due to less snow and that then helps explain why the volume anomaly reduces after May.

Does that sound sensible interpretation? Anyone got more or different thoughts?

[ChrisReynolds]

Crandles,

PIOMAS volume anomaly.
http://farm9.staticflickr.com/8245/8551805629_34cd27296e_o.jpg

For the last 3 years the volume anomaly ceases to fall and starts to level out by the end of June, then the anomaly climbs into July, and levels (level suggests only constant offset from baseline).

For the last three years volume losses from July to August have been the lowest three years since 1978.

2000   -4.946
2001   -4.452
2002   -4.980
2003   -4.924
2004   -5.014
2005   -4.462
2006   -4.235
2007   -4.455
2008   -4.907
2009   -4.509
2010   -4.043
2011   -3.849
2012   -3.962

Is this because there is a lot less area from which to lose ice edge, and to a lesser degree, thickness? Here's August CT average monthly area, the same behaviour of the last three years holds for July.

2000   5.008
2001   5.187
2002   4.626
2003   4.771
2004   4.904
2005   4.512
2006   4.561
2007   3.465
2008   3.765
2009   4.146
2010   3.920
2011   3.451
2012   3.037

I'm still not sure.

Source data in spreadsheets here:
https://docs.google.com/folder/d/0B3pB-kdzoLU3R0wtNnBIMmFyX2s/edit?usp=sharing
Never shared a folder in Google docs, let's see how it works. The Sea Ice Public file contains monthly data.

[crandles]

I am just guessing but my guess is a combinations of things:

1. Reduced area from which to melt volume.
2. Less edge of pack which tends to melt quickly.
3. Declining snow cover pattern might be different than PIOMAS model assumes such that at edges of pack snow is thicker and ice is thinner and ice melts quickly but further into pack the ice is thicker.
4. FYI stuck to bottom of MYI melts off earlier but then the MYI is hard to melt.
5. Early in melt season temperatures fluctuate causing brine rejection, so that later in season the ice is fresher and harder to melt.
6. Deeper ocean could mean less upward heat flux.


There could well be lots of other factors that are more important.

I find it hard to imagine a factor that could be strong enough to more than counteract what you would expect from albedo feedback so that is part of the reason why I am guessing combination. It didn't seem difficult to arrive at list above.

Got any other factors to add to list or different interpretation?

[ChrisReynolds]

Thanks Crandles,

I agree with them all, except perhaps 4.

PIOMAS has a two layer ice model according to the papers I've been reading. Thickness however is modelled as a thickness distribution function that smears the nominal thickness. In such a scheme I can't see how FYI melting onto the base of MYI would appear as anything as a shift in the nominal thickness and knock on shift of the thickness distribution.

In fact I don't find much to suggest differentiation between MYI and FYI, except in that they would have differing thickness profiles, as there is also math to parameterise ridging processes.

I'm trying to get organised before writing to Dr Zhang, unless there's a more recent radical revision what I say seems to be the case - not sure where PIOMAS V2 comes in, and aside from Schweiger 2011, which doesn't cover the innards of PIOMAS in detail, I can't find a paper detailing PIOMAS V2. If you need I can link to the three papers.

[crandles]

Thanks Crandles,

I agree with them all, except perhaps 4.

PIOMAS has a two layer ice model according to the papers I've been reading. Thickness however is modelled as a thickness distribution function that smears the nominal thickness. In such a scheme I can't see how FYI melting onto the base of MYI would appear as anything as a shift in the nominal thickness and knock on shift of the thickness distribution.

OK, I am sure you know how PIOMAS works better than I do.

Out of interest if increased FYI freezing onto bottom of MYI was happening in the real world but not modeled in PIOMAS then could assimilation of data tend to cause corrections to quicken the volume decline early in the melt season and to slow volume decline later in the season? Perhaps only a slowing of volume decline in areas when and where MYI fully melts out so not much adjustment.

Of course if it isn't modeled, it probably isn't very important or isn't really happening.

If they aren't modeling FYI/MYI difference when we know MYI is harder to melt then assimilation of lack of melt out of MYI that would be expected to melt out if it was FYI could be a more important reason why the volume anomaly reduces after June?

[Vergent]

PIOMAS appears to overestimate thin ice thickness and underestimate thick ice, yielding a smaller downward trend than apparent in reconstructions from observations.

http://onlinelibrary.wiley.com/doi/10.1029/2011JC007084/abstract

A consequence of this is that PIOMAS would overestimate the early melt and underestimate the late melt. The underestimation would increase as you start melting more of the thicker ice.

Verg

[ChrisReynolds]

Crandles,

I don't think it is due to the adjustment procedure. What's notable about the three recent years is that previously the pattern of loss from Apr to Jun was complex with gains and losses. In the recent years losses dominate.

This morning I've just worked out average loss in volume and thickness from gridded PIOMAS. They show that post 2010 there's a jump in thinning and vol loss for 2 to 3m categories. ice 1m and under doesn't change and ice 4m and above shows massive volatility in recent years. It's hard to reconcile what I'm seeing in the data with something like adjustment. But I'll think about it.

Later this weekend I will email Dr Zhang to ask if he's looked into the spring melt.

[mati]

I did find a research project at University of Alaska, but the results are not yet in:

http://seaice.alaska.edu/gi/projects/snow