The previous blog post dealing with area and extent had a bit of a wordy introduction, so I'm doing a small follow-up showing the different graphs made by commenters and myself.
First of all the very telling graph by Larry Hamilton:
Update: Larry Hamilton has updated the graph with July 2010 data and says the following in the comment section below:
July or August, not September, comprise all of the low points in the area/extent curves (except for an odd double-dip in 1995). FWIW, in the past 3 years August has been the low point.
July 2010 looks similar to July 2007, by this metric.
As area and extent have been in decline over the past 30 years, one might have expected the percentage you get when dividing area by extent to stay roughly the same. But that isn't the case. Larry's graph shows that area/extent percentage has been in decline over the past 30 years, which in my view can only mean one thing. Or two things actually. First is that because of higher air temperatures in the Arctic in some years, like 2007 and this year, more melt ponds show up earlier in the first part of the melting season (check Tamino's post which has a graph that shows this). But more importantly around this time in the melting season - when the percentage reaches its bottom - area and extent diverge in such a manner that it says something about the interior of the ice pack. The ice seems to have become so thin that it gets pushed around more easily, causing 'holes' to appear in the ice pack's interior.
As it says in the graph, Larry has used monthly NSIDC data. Another commenter FrankD has used IJIS extent data, deduced IJIS area data from their area graph (no downloadable data yet) and combined them to produce this graph, showing area/extent percentage during the melting season:
I've decided to man up and make my own graphs, as I want to use the area/extent percentage graph for the SIE updates until the end of the season. I'm hoping IJIS will answer my mails asking them for area data, but in the meantime I'm using IJIS extent data and CT area data and combine them to make graphs such as these:
And here's the graph I'll be using for the SIE updates this month:
Here's the graph for the 6 months that are most interesting:
And here's the graph for the whole year:
You can click the graphs for bigger versions.
If you want, you can look at and download the spreadsheet I've made. I've ordered all the data, removed the -9999 cells with more realistic numbers, and did a very naughty thing by removing Feb. 29th 2008. I just can't handle those leap years. Don't tell Tamino!
I end with the PIOMAS graph that was updated a few days ago. Although extent seems to drop relatively fast in the past 4-5 days, which boosts up the area/extent percentage, the graphs above might be a first confirmation of the anomalously low volume the PIOMAS model is showing:
Update August 5th:
I have received an answer from someone at IJIS:
In our webpages, only the graph and csv data of Arctic sea-ice extent are
released to the public. Regarding the graph of Arctic sea ice area, it
is prepared
tentatively only for research purposes because the sea-ice area can have
much
more errors than the extent has due to the effects of melt pond etc. in
summer.
So releasing the area graph and csv data to the public might cause a
confusion.
This is our current policy for the sea-ice monitor web.
Thank you for your understanding.
This reminded me to insert a caveat: these graphs are interesting and fun to make, but not conclusive in any way. :-)
The FrankD graph (melt season area/extent) shows us that the last 3 years are exceptional in that the minimum was reached in late August while the previous 4 years were consistently reaching minimums in late July.
It is also interesting to see how consistent 2002 thru 2006 were during most of the melt season. The most significant anomaly among those years was 2004 which was high during May to June.
2010 is also something of a standout in how erratic it has been between late June thru July.
Thanks again for all the great work!!
Posted by: Andrew Xnn | August 04, 2010 at 22:37
Andrew - That shift is interesting; it suggests more open water between floes at a moderately fine scale - a looser pack, in essence - since the bad years of 2006 & 2007. I'm watching to see if this year repeats the downwards kick through the middle of August that we have seen for the last three.
The 2004 line through April-June is quite odd, and I double-checked it to make sure I hadn't muffed a formula somewhere. But it does make sense when you look at the graphs side-by-side. While 2004 April area was low, 2004 April extent was the lowest of the nine years charted. Over the next two months both measures rose to the highest in the nine years, but with area always leading by a little. Since both were declining, obviously, we can say that the decline in area was relatively slower than the decline in extent. So the area:extent ratio was always high, even though both were varying.
I haven't gone back to look at the pictures, I expect that extent declined because of melt at the fringes, but the pack remained comparatively well consolidated. CT's historical maps seem to be offline, but I'll have a poke through the archive when I stop getting 404 errors at see if thats what their maps show.
Neven - thanks again for running this great blog. I totally sympathise with IJIS's position, and I hasten to add that I always thought what I was doing was fairly crude and non-definitive. However it seems an interesting indicator of pack health, so I'll update it from time-to-time over the remainder of the season, and post links here.
Larry's long span graph is very interesting, though I must concede I don't have the patience for two data points per year - I want see where the August line is going to be in five years NOW, dammit!
Posted by: FrankD | August 05, 2010 at 12:23
FrandD, the NSIDC data is month means for Area and Extend, from 1979 to present.
Posted by: Artful Dodger | August 05, 2010 at 15:31
Neven, good work on the graph. Smoothing the curves is a nice touch. I would prefer to see at least a full season's data (preferably 6 mths) on the graph. The way it is drawn now with only a few days of 2010 data we can not compare this year to others. As long as you continue to update the spreadsheet, I don't think that a coarser resolution on the time axis matters.
With respect to the 'data hole' at the north pole, CT does add in the area to their daily posted Area numbers, so I believe 'ignoring' worked well here.
Posted by: Artful Dodger | August 05, 2010 at 15:39
Thanks, AD. I tried to make a graph for the whole year, but somehow OpenOffice didn't accept something I did and made all the trend lines appear in red. I'll try making a 6 month graph this weekend.
I believe 'ignoring' worked well here.
I know. It is one of the best methods around. The whole pseudoskeptic community is using it on a continuous basis.
Posted by: Neven | August 05, 2010 at 15:50
July NSIDC means just came out, so I drew an updated "area as % of extent" graph. To highlight July I (and give Frank more than 2 data points/year to look at!) I added a third line for that month.
The result was a surprise: July or August, not September, comprise all of the low points in the area/extent curves (except for an odd double-dip in 1995). FWIW, in the past 3 years August has been the low point.
July 2010 looks similar to July 2007, by this metric.
http://img.photobucket.com/albums/v224/Chiloe/Climate/area_vs_extent2.png
Posted by: L. Hamilton | August 05, 2010 at 16:49
Larry, you're making it difficult to keep up with you. Thanks for the great graphs.
Posted by: Neven | August 05, 2010 at 17:25
AD, I've added two more graphs, for half-year and full-year.
The half-year one looks nice for the SIE updates, I think.
Posted by: Neven | August 05, 2010 at 23:36
Neven, top shelf. This story needs to be put up front, how about adding the 6-mth graph to the the Arctic Sea Ice Graphs page? Well done!
Posted by: Artful Dodger | August 07, 2010 at 16:08
Indeed, very interesting graphs! And now we are ready to answer Dodgers original question, if we can use this to forecast extent? I'm still sorely missing my old favourite plottingprogram SM, so i am only at the point at plotting CT area although i fint that interesting in itself, but maybe you could start to put the three graphs: area, extent and area/extent together and see if something emerges from that.
Posted by: siili | August 07, 2010 at 16:28
My original question was, does a significant drop in the Area/Extent ratio precede a dramatic drop in total Extent? It seems to me that, in the case where the pack is melting in from the edges, A/R will not change significantly in advance of the drop in Extent.
In the case where the pack is breaking up / collapsing by being inundated with warmer water, then A/R must go up just before the drop in Extent.
Again, without per basin data, there may be so much noise in Area & Ext. measurements that a significant signal may be drowned out (though CT clearly has per basin data...)
Posted by: Artful Dodger | August 07, 2010 at 17:45
AD, in the previous Area vs Extent post Axel/IceMan says that the area/extent ratio is called 'compactness' and gives as an example a paper by Lindsay and Zhang.
I've updated the Arctic sea ice graphs page with the webcam image from Healy. If I knew how to make a script that collects all data and automatically adjusts the ice compactness graph I would add it to the page as well. To do it manually though (OpenOffice won't let me export the graph, so I have to do it with PrintScrn and Photoshop).
I will be using the 6 month-graph for tomorrow's SIE update.
Posted by: Neven | August 07, 2010 at 18:11
Guys, it's really easy to suggest work for others to do, but i really think that we should do a bit more of the process ourselfs and hence have much better controll of the outcome.
Over at the other side in a more than usually stupid thread Julienne from NSIDC has advertized their gridded daily data with concentration per pixel given. With that we can define own areas of interest and determine area, extent, compactness whatever we like.
It's a very simple dataformat, that in principle only has to be read into a matrix, together with the areatransformation matrix, and then we just start adding things up.
I can try it myself, but since the same ole was dicussed at the other place on the other side some years age, i'm sure there must be plentifull of code arround.
Posted by: siili | August 07, 2010 at 18:32
siili, count me out! :-)
Posted by: Neven | August 07, 2010 at 18:49
Neven, there's no overwhelming need to automate the 'Compactness' (A/E ration) graph on the Arctic Sea Ice Extent page (although AJAX could do it). I suggest it's fine to simply link there to a static URL for the graph as updated every 3 days or so in your regular updates.
Two suggestions for the graph: 1. perhaps provide a table (or at least a label) equating Day of the Year (x-axis) to Date. 2. a format other than jpeg would be more appropriate, perhaps .png or even .gif
Finally, it should be fairly straightforward to code a javascript that gets the current GMT time and and converts it into the image URL for the current Healy AloftCam. If you request, I could provide some support with that. Cheers!
Posted by: Artful Dodger | August 07, 2010 at 19:28
Johann, can you provide a link to the NSIDC gridded data?
Posted by: Artful Dodger | August 07, 2010 at 19:30
Getting exited are we?
You could try theese to start with
http://nsidc.org/data/polar_stereo/ps_grids.html
http://nsidc.org/data/docs/noaa/g02135_seaice_index/index.html#doc_info
http://nsidc.org/data/polar_stereo/ps_grids.html
Posted by: siili | August 07, 2010 at 20:49
and the data: ftp://sidads.colorado.edu/pub/DATASETS/seaice/polar-stereo/nasateam/near-real-time/north/
and the tools: ftp://sidads.colorado.edu/pub/DATASETS/seaice/polar-stereo/tools/
Posted by: siili | August 07, 2010 at 21:11
If you request, I could provide some support with that. Cheers!
AD, thanks, I'll keep that in mind. I'm really busy at the moment, so this is low on my priority list for now.
Posted by: Neven | August 08, 2010 at 03:17
New version of the IJIS Area/Extent graph to 10th August can be found at:
http://img808.imageshack.us/img808/4606/areaoverextent100810.png
Posted by: FrankD | August 13, 2010 at 10:47