I'm proud to announce the first guest blog on the Arctic Sea Ice Blog, written by Bfraser. As my virtual credit card still hasn't been cleared and I thus haven't upgraded the blog to a version where I can add authors, Bill sent me his text, images and more, which are reproduced below:
TOPAZ data by Cryosphere Today regions
During last fall's break in the PIOMAS data (nowhere near as long as this one), I had a thought: "TOPAZ has all of the information needed to compute area, extent, and volume." I wrote to NERSC, asking whether they'd be willing to calculate that information and print it at the bottom of the maps. I still haven't heard back. Last month, during the long hiatus in the PIOMAS data, I decided to download the TOPAZ data and compute it myself.
Their model is really quite cool, they track all the way from the Bering Strait to Portugal at a resolution of 12.5 km and keep track of the ocean depth for each grid cell. In addition to ice concentration and thickness (the only ones I needed for my purposes) the model tracks temperature, salinity, and motion for the ice, the air, and for the water at 11 different depths.
Every week or so, they run the simulation against any new input data that they have, resulting in a prediction for the next several days. The result is data for almost every day since May 22, 2007 (with the last 10 days subject to revision). By downloading the entire dataset, I was able to calculate Area, Extent, and Volume:
Two big caveats to consider for data are that it doesn't include the entire area of interest (most notably, the Bering Sea is missing) and that it disagrees with PIPS regarding the thickness of the Central Basin -- though I am inclined to believe TOPAZ, as it seems closer to PIOMAS. Hopefully, CryoSat will help settle the dispute. Also, notice that there seem to be some glitches (for example early December, 2007). In addition, the model occasionally produces concentration values greater than 100% (though never quite as high as 101%). And finally, notice that the units are slightly weird as the area is measured in km^2, while the height is measured in meters.
Following the discussion of the Central Basin in Open Thread 5, I added a feature for editing regions which can then be used to restrict the calculations. I attempted to match the Cryosphere Today regions, resulting in the following distribution:
Which I then used to compute Area, Extent, and Volume for the Central Basin:
The files that I have, which I would like to make available, if I could find a place to host them, are as follows:
1. The data set from TOPAZ -- 860 MB zipped
2. The software used to analyze the dataset -- 33K zipped [here it is: Bfraser.20110303.icesource, N.]
3. The spreadsheet produced from my software -- 175K zipped [here it is: Bfraser.20110303.icedata, N.]
Finally, are there any other interesting analyses that people would like to see? To get the ball rolling, I thought I would create a graph of how much ice at least one meter thick there was by region, though I don't know how much such a graph actually means with less than 4 years of data.
William E Fraser
Wow! This is great stuff. The striking feature here is that wintertime volume numbers are reaally decreasing much more quickly than the smmer time numbers. This chart shows 12 more years to an ice free arctic winter.
Posted by: RunInCircles | March 05, 2011 at 13:36
I second that emotion - this looks like some fantastic grist to the mill.
Five cheers for bill!
(there goes my Sunday, though...)
Posted by: FrankD | March 05, 2011 at 15:06
Holy Schmoely; this is absolutely fantastic!!!
Many many many thanks.
Would like to point out that there may be something of an anomaly in Sept 2007 as well.
The minimums in most years are fairly well rounded, while there appears to be something of a spike in that month. However, as we all know, there is an indisputable trend and it's just a matter of time before it all melts.
Posted by: Andrew Xnn | March 05, 2011 at 17:13
Pheeewwwww !!!! That's NICE.
I have lot of room and power on a good nice server, so I can offer to setup a place for this. I had also thought about offering some area to play with R... perhaps by managing the programs on Github, so everyone can play with the thing... so, I think it's time to start this project.
I'll try to setup that stuff by Monday, should be done by next Friday.
Neven, do you have my email address so we can exchange on this ?
Posted by: fredt34 | March 05, 2011 at 21:57
Great job William. I think that there was discussion on the issue of aveage thickness earlier, but this graph really demonstates how the average thickness has declined. Through much of last fall/early winter, average thickness of the ice was under one meter. Your chart shows this to be a new trend.
Posted by: Ecojosh | March 05, 2011 at 23:33
Refresh my memory: at what thickness does sea ice readily fracture and become susceptible to wave action? Or does the age of the ice have more control over strength than the thickness does?
At any rate, I would like to see a map which uses both age and thickness to predict the likelihood of each discrete chunk of ice melting out in the coming summer. (Weather and currents are critical, but all other factors being equal, areas of stronger ice, older ice, edge ice, ridged ice, and so forth should be able to be teased out with a GIS.)
(I really should apologize if my comments and questions are Ice 101; my scientific background is strictly terrestrial.)
Posted by: Rlkittiwake | March 06, 2011 at 03:01
Thanks for a great post. And especially for offering
are there any other interesting analyses that people would like to see?
I've read Stoat's suggestion that increasing open water will provide a negative feedback that makes ice-free periods unlikely any time soon. At a glance, I don't see two volume growth regimes in these graphs (fast growth while open water lasts, followed by slower growth under ice cover), but - would proper stats analysis show it? And, as the open-water period for the central basin is very short, might other basins' data, with lower extent at minimum, show a change in rate of volume growth before/after the end of open water?
Thanks again for a thought-provoking piece.
Posted by: Simon | March 06, 2011 at 23:52
Bill, my webhost claims I have unlimited space and bandwidth at rhinohide.org. If you like, I would be happy to host the files there. You can ping me at gmail.com by the handle of ronald.broberg.
Posted by: Ron Broberg | March 07, 2011 at 00:07
Hi guys (and girls ?),
I'm happy to annouce that I managed to setup the stuff I was thinking about... Soon you'll be able to make your own "R" code on GitHub, or take others' code and play with it, and have it gotten and run by my server.
Stay tuned, more info to come soon !
Posted by: fredt34 | March 07, 2011 at 01:11
Bill wrote his code in java, it's quite easy to read : is there a volunteer to translate it to "R"? So we can play with it and share it, in a (probably) very efficient way.
Posted by: fredt34 | March 07, 2011 at 01:16
Ritikitte:
It is more complicated that just thickness and age. You also have to consider latitude. Ice closer to the pole melts slower than ice further south. NSIDC uses an algorithm that considers age and latitude for their forecasts of ice extent. (Does anyone know of a link to their data?). Soon age will not matter, all the ice will be first and second year. As the ice in the Canadian Archapeligo melts, new ice blows over from the higher latitudes. As the ice near Canada thins, more will blow over. That leaves more open water by Siberia.
Posted by: michael sweet | March 12, 2011 at 21:33
@ michael sweet
You might try looking here:
http://nsidc.org/data/seaice/data_summaries.html
The Yooper
Posted by: Daniel Bailey | March 12, 2011 at 22:19
Hi, I appreciate Your work very much.
I'm interested in the data table linked to above. If you could give a hint, why for each region not only colums "extent", "area" and "volume" are given but also 3 colums with the area title on top (e.g. "baffin", "baffin" and "baffin"). Also a word about the units would be helpful.
Posted by: dominik lenné | March 15, 2011 at 03:59
Your welcome, dlen.
My intention for the column headers is
TimeStamp Date central extent central area central volume baffin extent baffin area .....
but I realize that whitespace doesn't always travel well :[
As to units:
timestamp is seconds since 1970, and is used because that's the way the data files are named -- allowing me to debug things more easily.
Date should be self-explanatory
Extent and area are in km^2 (notice that the extent is always a multiple of 156.25, since the measurements are made on a grid whose spacing is 12.5 km).
Volume is in km^2 * m (kind of an awkward unit, but I chose it because km is natural horizontal unit, while m is the natural vertical unit -- plus it graphs better than the alternatives)
Posted by: Bfraser | March 15, 2011 at 08:04
Ack, whitespace really doesn't travel well I guess I need whitespace so the:
TimeStamp < tab > Date < tab > central extent < tab > central area < tab > central volume < tab > baffin extent < tab > ....
Posted by: Bfraser | March 15, 2011 at 08:06
Ah, yess, I imported the data in excel and said to excel: "use blanks as separator". Somehow I guess it took the blank in the first line and split title *and* the numerical data into 2 columns. Something like that.
Thanks for the remarks.
Posted by: dominik lenné | March 18, 2011 at 01:30
Does anyone know when the TOPAZ thickness map will be updated? It seems to be stuck at Feb 24, and my hands are itching to make a PIPS vs TOPAZ animation.
Posted by: Neven | March 18, 2011 at 15:43