Sunday, June 12, 2016

More About Anomalies and Normal Distributions

Some followups to my post of last night.

First, I made a spreadsheet error on the HadCRUT4 monthly anomalies that slightly inflluenced the result. I replaced the plot with the corrected one.

Second, it was requested that I slso do a histogram for the annual anomalies, since that was being discussed. It's also clearly not normally distributed:


Third, a Twitter response to last night's post pointed to this post by Andy Cotgreave, who created this nice graph, which shows how the distribution has been changing over time, shifting to higher temperatures:


There are more good graphs on his post.

2 comments:

JoeT said...

The yearly Hadcrut4 pdf is great example of a right-skewed distribution. Just what you'd expect in a warming climate. Thanks David.

David Appell said...

Thanks Joe. I'm not going to have time to fit a normal distribution to it (given its mean and stdev). I can't get it to work out, and that's after a couple of hours spent learning that, for Excel

-2^2=4, not -4 as I thought.

I thought taking a power has precedence over any other operation, so was surprised to find this. But Excel interprets it as (-2)^2.

and tomorrow I have a busy day. But it's clear HadCRUT4 anomalies are not normally distributed (which shouldn't be a surprise).