Pixel Counting Teensy Weensy Alarmist Brains

For the past five years, I have been listening to alarmist trolls blabber mindlessly about pixel counting and comments Walt Meier made to The Register. The five year anniversary seems like a good time to set the record straight humiliate these morons.

ScreenHunter_644 Sep. 16 17.24

Steve Goddard  8/22/08  to walt

Dear Walt,

The attached image shows UIUC and NSIDC maps on top of each other from August 15, 2007. The interior land edges surrounding the Arctic are all registered, and as you can see, the UIUC ice extent is much smaller on the Canadian side. NSIDC is transparent white, and UIUC is colored.

BTW – when I do pixel counting on the NSIDC maps, I get an exact match vs. your graphs. The technique is excellent for making a decent estimation – if the base maps are correct.

Based on your last two E-mails I was surprised by your comments in The Register, as you had made it clear in your private messages that you understood what the problem was and were searching for the same answer. I absolutely do calculate the pixel area corrections for each date, and they make very little difference. On most days they /increase /the discrepancy by about 3% beyond what I reported. I am not making any attempt to measure ice area, and I am making an apples for apples measurement of extent.

The reason why UIUC maps show 30% increase in extent is because their August, 2007 maps appear to be not plotted accurately – as can be seen in the attached image. Their 2008 maps seem to show much closer agreement with NSIDC maps.

The UIUC maps show a striking increase in ice extent in 2008. Seeing is believing – but unfortunately in this case the images were not accurate.

– Steven Goddard


Walt Meier  8/22/08  to me

Yes, if the map is not too distorting, like NSIDC’s maps, pixel counting is a good rough estimate. In fact, we helped developed an educational module where students would estimate the extent from pixel counting. Scientifically, it’s not valid, but using our maps we found it was close enough to get the message across for students. That’s why I wasn’t sure it was the pixel counting that was the problem – it seemed like it shouldn’t have been such a large error. But since I wasn’t familiar with the UIUC images, I couldn’t say exactly what the problem was, but it had to be due to the images in some way.

As for the comments on The Register, I forget exactly what I said, but I think I also tried to address some of the other comments by readers who were misunderstanding things. Sorry if it sounded like I was incorrectly criticizing you.

Thanks again for being willing to work with us to correct the errors.



The discrepancy was due to two problems with the Cryosphere Today maps. They don’t show ice below 30% concentration, even though their legend makes it appear they do, and their eye altitude was incorrect. Both problems were resolved by Bill Chapman at UIUC.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s