Here is the problem. A cartoon by XKCD
..and of course it got a boost from the Bad Astronomer:
It’s not like my pal Randall needs the traffic from me, but his xkcd comic on the new normal from global warming is just perfect, so I’m putting it here
Now I followed the link given in the cartoon but Randall had not put in upper and lower case, and I got nowhere.
I went searching for temperature information for St Louis, Missouri and got this page and in that page there are spreadsheets to data about weather extremes for St Louis especially numbers of cold days below 32F and 0F for each year since 1893!
From that sheet I was able to reconstruct the number of days per year where the min temperature reached 0F and the maximum temperature reached 0F and a plot a much longer series of the number of days where the temperature reached 0F every year since 1895.
One of the interesting things is that in 1936, the height of the Dustbowl, St Louis had far and away the most 0F days in the entire record.
What IS interesting is that I did not reproduce Randall’s diagram in the cartoon. There are clearly days below 0F in 2010, 2011 and 2012.
There might be a climate signal in the noise, who knows? But I am very wary of drawing any sort of trend through that data because of the inherent noise of temperature data and the filter used.
The question is: is the distribution of days below 0F distinguishable from a random process?
More to follow…