Ice thickness models

One of the most intruiging and elusive aspects of Arctic sea ice is its overall thickness, and by implication the total volume of sea ice. The bigger the volume, the longer it will take for the Arctic Ocean to become effectively ice-free in summer. Unfortunately, despite many different kinds of data, no one can tell for sure what the current status is, except that volume has decreased significantly over the past decades.

Satellites could provide more definite answers, but we are currently in a gap between ICESat and CryoSat-2. ESA and NASA are doing a fascinating job in the Arctic gathering data which will be used to calibrate the latter as we speak. In the meantime we fill the time by speculating, and a big part of that speculation is fueled by ice thickness models.

The best known models in Arctic sea ice amateur circles are PIPS 2.0 (Polar Ice Prediction System), PIOMAS (Pan-Arctic Ice-Ocean Modeling and Assimilation System) and TOPAZ (Towards an Operational Prediction system for the North Atlantic European coastal Zones). Although model output is not as good an approximation of reality as actual data - to put it mildly - it is still interesting to look at.

Ever since Bfraser wrote his first guest blog here, TOPAZ data by Cryosphere Today regions, my hands have been itching to make an animation comparing PIPS and TOPAZ sea ice thickness maps this year so far. Now, PIPS (which is run by the US navy) has a daily updated forecast for the next day, which is great. TOPAZ however hadn't been updated ever since February 24th. It was taking so long that I decided to ask someone when the next update was due.

I promptly received an answer and it turns out the TOPAZ system - delivered as part of the MyOcean project - has been upgraded last year and transitioned to the Norwegian Met Office (met.no). Sea ice thickness maps can now be viewed with the help of this dynamic viewer and I have to say the result is stunning. You derive sea ice thickness maps by clicking on 'TOPAZ4 model results' on the left and then 'sea_ice_thickness'. Click on the little plus-sign on the right and choose 'North polar stereographic'. Then finally adjust the numbers on the colour bar, and presto, you have a sea ice thickness map. Great stuff.

What is also great is that I now am able to make an animation of sea ice thickness progress according to the models. I start off with a comparison between PIPS and TOPAZ from January 1st to February 26th 2011 (1 week intervals):

From March 5th onwards I can add sea ice thickness maps produced by the PIOMAS model (there are no images prior to this date):

We see quite a few differences between the models. This isn't much of a surprise, as there was much ado over this last melting season when the ice thickness models received more attention for the first time, on both sides of the climate aisle. These differences won't be solved to everyone's satisfaction until CryoSat-2 data comes rolling in, and perhaps not even then.

PIOMAS and TOPAZ agree more on where all the thick ice is, mainly north of the Canadian Archipelago and Greenland, but PIPS is showing a big red blob over a much larger area north of Greenland. As for the ice that is less thick: TOPAZ shows much of the Arctic Basin to be around or below 2 metres (different shades of blue), PIOMAS indicates between 2 and 3 metres (shades of green), and PIPS seems to be slightly over that.

Which model is the best? It's hard to tell, and explaining the subtle differences between all three of them is beyond my means. What I can do, is compare this year's output of the models with previous years, which I'll do later this week.

Comments

[Artful Dodger]

Nice resource, Nevin. You've got friends in *cold* places... (pardon my Garth Brooks)

[Andrew Xnn]

My understanding (and please somebody correct me if I err) is that these systems measure the difference in time it takes radar signals to bounce off the surface of the ice compared to the cracks between the ice. From this, they can deduce how thick the ice is above the water and then in turn calculate how thick the ice is on average.

What hasn't been stated is if they are both looking at the same data and coming up with different answers or if they have different satellite sources.

I'll guess that they are looking at the same data and are coming up with different answers because there is more than 1 way to interpret the data.

For example, while the thickness of the ice obviously varies, the amount of open water between ice varies as well. In addition, melt ponds must be tricky to deal with since they may appear to be open water between ice. Of course melt ponds are summer thing. So, we also have the problem that in some seasons the same data could mean different things.

Finally, last year Sept one of the systems came up with 4000km^3 of ice with a long term trend of losing between 350 to 1000km^3/year. Not sure if the other system had similar findings or not.

Anyhow, many many thanks Neven. Great job and also to all the knowledgeable com mentors out there in cyber land.

[L. Hamilton]

When I was looking at PIOMAS volume estimates for my guest post earlier this month, I checked out the relationship between PIOMAS September volume and NSIDC area. Not surprisingly, there's a fairly strong linear relationship (r = .94). Here's how they moved together:

http://img.photobucket.com/albums/v224/Chiloe/Climate/Volume_area_2.png

[L. Hamilton]

Step-back view, same data but with zero points shown for perspective:

http://img.photobucket.com/albums/v224/Chiloe/Climate/Volume_area_1.png

[MikeAinOz]

Neven it's good to you you covering ice volume. How data is collected for these models is interesting. They are based on ensemble data, with radar playing a very limited role at the moment. Most data comes from floats and other on-site mearurement. Here's a paper about TOPAZ data http://www.cawcr.gov.au/staff/pxs/wmoda5/Oral/Bertino_Sakov.pdf
, it's a little old but covers the subject nicely.

[Neven]

Thanks a lot for that PDF, MikeAinOz. Not too much text, not too complex, just the way I like it. ;-)

BTW, I forgot to properly attribute the picture at the top of the post. I snitched it from the ESA blog.

[R. Gates]

The differences in the models is very interesting to look at and you've done a great job at highlighting them. I do have very strong confidence that in a few years hence, because of CrySat 2 and eventually Icesat 2, that the models will begin to become both more refined, and then less important as it is replaced by real verified data on thickness. Within a few years we should be getting regular sea ice volume data that will provide a whole new way of considering the state of the arctic cryosphere. I'm suspecting the AGW skeptics will be rather quiet by then...

[Peter Ellis]

R. Gates: Provably wrong, I'm afraid.

Either current volume models like PIOMAS are wrong about the amount remaining and the rate of decline, or they're right.

If they're wrong - for example there's more there than expected, or the decline is slower - then the skeptics will be noisily cock-a hoop.

If they're right, them we won't be getting regular summer sea ice volume data in a few years' time, because there will be no summer sea ice.

:-(

[R. Gates]

Peter,

I think we'll find that PIOMAS is probably better than the others, but still off the mark a bit. Eventually CryoSat 2 and then IceSat 2 will minimize the need for these models as we'll have actual data. And as far as summer sea ice goes, we've definitely got a quite a few summers before we'll see an ice free Arctic (my guess is at least 15 or so) and it could very well be that it will linger as virtually ice-free in summer (less than 1 million sq. km.)for many years with small pockets clinging north of Greenland and/or the Canadian Archipelago. And of course, it will always freeze over again in winter to one degree or another, and so CryoSat 2 will be useful to monitor rates of thickness growth, etc.

[Kevin McKinney]

"And of course, it will always freeze over again in winter to one degree or another, and so CryoSat 2 will be useful to monitor rates of thickness growth, etc."

Not necessarily. Modeling work has shown that consistently ice-free summers will likely be enough to 'flip' the Arctic into a year-round ice-free state. That's not a definitive result, if I understand it correctly, but is a very real (and very unsettling) possibility. I don't have the details at hand--and don't have time just now for a search--but I'm sure one of the regulars can oblige.

[R. Gates]

Kevin,

Certain models indicate rather interesting behavior for the ice even after an ice free summer. This study for example:

http://www.agu.org/pubs/crossref/2011/2010GL045698.shtml

Shows that an ice free summer may in fact allow greater amounts of heat to escape during the fall and winter from the open arctic water. Leading to cooler water and ice formation.

What I'm getting at is that the ice may have some interesting behavior as we get near or at an ice free summer arctic, showing both positive and negative feedbacks effects in unexpected combinations as any chaotic system does when undergoing change.

This will all of course play havoc with the weather here at the lower latitudes, especially in the winter.