About that Arctic sea ice …

The Register, an occasionally accurate online IT newspaper, has been running a series of warming denial pieces, by one Steven Goddard. Goddard has been trying to cast on temperature and ice data. Unfortunately, he does a whole lot of cherry picking. For example:

A second important issue with NASA’s presentation is that they use the time period of 1951-1980 as their choice of baseline. This was a well known cold spell, as can be seen in the 1999 version of the NASA US temperature graph below.

Why use a graph of US temperatures instead of world temperatures? The “cold spell” is more pronounced in the US graph. In fact, the average for 1951-1980 is almost the same as for the 20th century so it is misleading to call it a cold spell. Goddard prefers to use satellite data, with a baseline that is significantly warmer than the 20th century average, to try to making warming seem less. He does some more cherry picking when he presents a map of GISS temperatures leaving out the sea data and using 250km smoothing (even though the NASA used 1200km smoothing for the temperature graphs) in order to make it look like there are significant gaps in NASA’s coverage. He compounds this by picking a month where this makes it look like there is a warming bias in NASA’s temperatures.

But his most recent effort was even worse, claiming that the NSIDC’s graph of Arctic sea ice extent was wrong and that there was 30% more sea ice than at the corresponding time last year.

i-11712ed7b31cf55b5724628fc6c08efe-20080827_Figure2.png

But it was Goddard who was wrong, as NSIDC’s Walt Meier explained to the Register:

He appears to derive his estimate by simply counting pixels in an image. He recognizes that this results in an error due to the distortion by the map projection, but does so anyway. Such an approach is simply not valid.

If you correct Goddard’s error, you get the same number as the NSIDC. Meier adds:

Besides this significant error, the rest of the article consists almost entirely of misleading, irrelevant, or erroneous information about Arctic sea ice that add nothing to the understanding of the significant long-term decline that is being observed.

Goddard admitted he was wrong but, as noted by Joseph Romm, Kevin Grandia and James Hrynyshyn, the numerous denialists who claimed that Goddard had shown that the ice wasn’t melting have mysteriously failed to correct things.

There are too many to list, so I’ll just point to the Australian bloggers: Jennifer Marohasy “Arctic Sea Ice Refuses to Melt”, Tim Blair “Arctic ice seems to be growing somehow.” and Andrew Bolt “Tim Blair rounds up the local anecdotes of coldening.”. No corrections from any of them.

Incidently, in the graph above, 2005 was the record melt until 2007. 2008 has already passed 2005 and whether or not it ends up setting a new record, it’s clear that the melt in 2008 is similar to 2007 rather than anything ever seen before.

Update: Bolt referred to Goddard more than once:

Steven Goddard checks those predictions that the North Pole could melt clear away this summer, and finds we can (yet again) relax.

Comments

  1. #1 Eli Rabett
    December 17, 2008

    Bernard Jr, it is not so much how accurate they are as the methodology used. There are a couple of groups that track this and they used different algorithms and get slightly different results. As in many such things the changes are in better agreement than the the absolute amount

  2. #2 sean egan
    December 27, 2008

    Bernard J and Eli Rabett,

    It turns out even the changes have little hicks. Tamino has just exposed a issue in how NSICD satellite area records are spliced in a debate with Jeff Id. Seems people, including NSICD forgot to adjust for the polar zone not covered by the satellite, which is not the same between satellites.

    Having said that, for area/extent the real changes are way outside measurement error and minor accounting errors.

    There is much scope for debate in the depth/volume. You have the Americian sub record which is accurate, but published slowly – not yet showing 2007, and does not go everywhere at the same time. Then you have a satellite which is faulty and is not doing the number of scans hoped for. So there are some good snapshots, but not a continuos film. Not the same part of the cycle each year.

    NSCID experts do pull this together. But, if you have to do modelling / make assumptions it leaves wiggle room which is much large than the straight measurement error. So basic questions like is there more or less ice volume today than the same day a year ago, two years ago etc, appear to get
    answers with rather limited certainty.

  3. #3 Eli Rabett
    December 27, 2008

    Sean, obviously the answers are good enough for hedge fund work (e.g. comparing each series with itself over time should be pretty good)

  4. #4 Dano
    December 27, 2008

    There is much scope for debate in the depth/volume. You have the Americian sub record which is accurate, but published slowly – not yet showing 2007, and does not go everywhere at the same time. Then you have a satellite which is faulty and is not doing the number of scans hoped for.

    Then you have the denialist dataset, gathered over many years that shows…

    Oh, wait. Never mind.

    Best,

    D

The site is currently under maintenance and will be back shortly. New comments have been disabled during this time, please check back soon.