Al, Yeah -- you're right :) Since this is polar integration there's that extra factor of r I dropped. Luckily, people under estimate area relationships for larger circles around small ones, so it sort of compensates for that ;) No, seriously, thanks for the catch. v*v*t it is!

And, ps: It's scary to look from the summer segments of the recent orbits to the origin and back. How tiny a sliver of the pack remains! I get the feeling we don't have many more seasons of watching melt pools freeze over on the drifiting cams...

Jim -- I really like the radially plotted charts. Beyond their immediate visual impact, they are an extremely effective way to accurately convey the scope of the decline. The area enclosed by each period's line is the time integral of the volume, in Volume-Days. The relative areas of the plotted regions (and hence the volume-days of each period) are much easier to discern in this format than in a time series. Of course, a big reason that they work so well for this is the magnitude of the decline relative to the noise, which means we don't have to visually untangle overlap from period to period :(

James -- Indeed. http://www.nature.com/nature/journal/v487/n7406/fig_tab/nature11257_F3.html may be the money shot re saddle collapse. 18 meter / year thickness loss across 1000s of square kilometers of ice sheet... 1.5km of ice thickness lost in 200 years.

Greg - The wikipedia greenland bedrock topography map linked above is based on the ETOPO1 global relief data set (http://www.ngdc.noaa.gov/mgg/global/etopo1sources.html). ETOPO1 bases its Greenland bedrock topography on the NSIDC 5KM Greenland Bedrock grid, hosted at http://nsidc.org/data/nsidc-0092.html and based on Bamber et al 2001. The Bamber paper is worth a read, but suffice to say neither the processed grid nor the sampling frequency over the ice really have the resolution to address questions of continuous below-sea-level paths to the interior. (http://www.cpom.org/research/jlb-ag.pdf) CReSIS at KU hosts the raw data sets the 5KM grid was produced with and a lot of data from subsequent field seasons: https://data.cresis.ku.edu/#RDS but they state, "We do not currently have plans to update the Greenland grid." So, there is room for some data crunching here :) As for well-substantiated sea level channels, I don't know :)

I created a version of Steve's animation with the frames registered to each other and a delay at the beginning and end:

Timothy, I believe that the page you linked only contains information about emissions, and not concentration/mole fraction. I have found reference to Kitt Peak spectrographic estimation of column methane starting in 1977, but I don't know of any continuous time series measurements of ground-level methane from that time period. There were individual measurement campaigns carried out, however, but I believe most of the data for those is to be found in published papers and not in most of the online carbon cycle data repositories. Law Dome ice core and "CH4 in archive tanks from Cape Grim Station, Tasmania" data are however in strong agreement and highly supportive of the assumption that the 70s had a high growth rate: CH4 in archive tanks from Cape Grim Station, Tasmania, deseasonalised: (year A.D.) Ratio (ppb) 1978.3 1468.3 1979.1 1490.9 1980.7 1507.7 1984.4 1572.9 Law Dome: 1965 1261.3 1967 1285.2 1967 1278.4 1967 1288.5 1970 1351.7 1971 1357.2 1972 1380.3 1972 1379.9 1973 1385.8 1974 1421.5 1975 1420.4 1977 1458.1 1980 1479.7 1980 1479.3 (http://cdiac.ornl.gov/ftp/trends/atm_meth/EthCH498A.txt)

Daniel -- Take a look at http://www.esrl.noaa.gov/gmd/webdata/iadv/ccgg/graphs/ccgg.MLO.ch4.1.none.discrete.all.png The Mauna Loa data only goes back to 1983, but you can see CH4 levels rose from about 1.63ppm to about 1.76ppm in the decade from '83 to '93, but flatlined or even decreased slightly from '99 to '06. Since then, however, the rate of increase has returned to just below that of the 80s. To what degree this is attributable to a clathrate gun scenario is unclear to me.

And, as for why mid 2010 beats out 2007 in the one-year median extent plot despite a higher max and a higher min: 2010 spent a lot more time hanging out between 6x10e6 and 1x10e7 than 2007 did.

I should add that the extent graph above very closely follows the inverse of the 12-month moving median of the UAH North Polar Region temperature anomaly for the same period, but with an additional downward trend in the extent data...

Re: Annual Averages -- the moving median with a window of n*period is extremely useful for detecting secular trends in periodic, slightly noisy data. I whipped this up from the IJIS AMSR-E Extent time series: The last plotted date is from January, since the window can't extend past the current date :)