I have been in a discussion with a determined global warming advocate. During our discussion I pointed out that the temperature difference between Brookhaven City, MS and Monticello, MS during the 1960s showed a seasonal variation. You can see this at this place.. But below is the pertinant picture.
The above photo is a chart of the temperature difference between these two towns. I wondered if the average temperature difference on, say Oct 30 was bigger or smaller than the average temperature difference on June 1. So....
I decided to align ALL the years from 1948 to 2002 and average the temperature difference for each day in the year. That gives me 366 daily averages and shows up any seasonal bias that lasts throughout the entire temperature series. The thing I am interested in is are there any days or periods of time that are systematically warmer or colder in one of the two towns.
In the chart below, the first value is the average temperature difference between Brookhaven City, MS and Monticello, MS for January 1. The second point is for January 2, and so on and so forth.
If the temperature is being measured properly, one should see a random distribution about the x-axis. But that isn't what we see.
Notice that in the fall (after day 260, which is about Sept 16, there is a strong bias on the days, in which Brookhaven is warmer than Monticello. That difference lasts until about Dec 1. This is the average difference for the ENTIRE temperature record.
If one looks at 10 year periods one gets a different picture. The biases change from decade to decade, but they are still biases.
In the 1960s the bias not only included the Fall but the Spring as well.
This view of the data is quite disturbing. It means that the data going into global warming studies is totally screwed up.
Let's look at the same thing for Okemah, OK vs Okmulgee OK. It looks like:
One can see that the bias here rises in the Fall and Winter but is least in the summer--even though the spread is noisy.