I like to look at the National Snow and Ice Data Center Web Site. I was looking at the data presented and I noticed something that seemed odd to me. So much of the data was outside the shaded area of the chart which was labeled as a + or – 2 standard deviation area. In a normal distribution, over 95% of the data should lie within the shaded area.
Then I looked at a second chart that shows recent data
Then I finally figured out what was going on…. the standard deviation data was old data. Why did they do that? Yes I know it was right in front of me the whole time, but I wasn’t paying close attention. I hadn’t noticed that the data set that calculated standard deviation and average was based on data that ended in 2000.
So when I made a post touting April as just about average for the 34 year history of the data, I was wrong. The Arctic Ice sheet was bigger than average in April of 2012. OK, a 34 year ice sheet data set doesn’t mean much, the data set is too small…but it’s bigger than a 21 year data set….60% bigger!
We know that the Arctic Ice Sheet summer melt peaked in 2007. We know data between 2000 and 2007 would have been less than average, and all data since 2007 has been warmer (less ice) than the average. It’s safe to assume that if a 34 year data set had been used instead of a 21 year data set, the average would have been a much lower number, and the shaded area that was 2 standard deviations would have looked different too.
So April of 2012 was not average for the data set, is was average for the last 21 years of the 20th century and much colder (more ice) than the average of the entire data set.
I am used to people making grandiose statements about Arctic Ice from small data sets. And really a 21 year data set or a 34 year data set doesn’t mean much in the grand scope of climate cycles that last millions of years…but it is an interesting manipulation of statistical data.