playing with numbers 2

The UK College of Agriculture source that provided me with monthly average temperatures for Kentucky, also included data on each month's deviation for "normal."

I thought "hey -- let's look for patterns of change over the past decade!" So I collected the deviation data for 2001 to 2008. What I found was that from January 2001 through May 2008, the average deviation from normal temperatures was +1.5 degrees. Looking at each year independently, all but one year had higher than normal deviations (ranging from +.979 to +1.94) and only one year had a lower than normal deviation, and it was only -.18. This strongly suggested that the current decade was substantially warmer than "normal."

Of course it matters what is used as "normal" to make the comparisons. I looked in vain for someplace on the UK College of Agriculture website for their definition of normal. Ultimately I found a link to a NOAA page that defined terms, which said that normal was based on a thirty year average, currently (by NOAA) 1971 to 2000.

But I'm not one to just accept numbers. I decided to go back to the original pages where the data was and see if I could find those places where the actual monthly "normal" temperature was listed, and not just the "deviation from normal." What I discovered was a tremendous lack of consistency in the UK data. In some years the average monthly temperature for January was given as 33.1 degrees F, and in others it was given as 31.3 degrees F. I don't know if they were calculating normal based on a different set of years each time, or whether they were just sloppy in recording the data.

Now I'm looking for better data sets. NOAA has monthly average temperature data sets for each individual reporting stations across the U.S., and one can find very localized data. These data sets vary considerably in their detail, reliability, and the time periods that they cover. The one closest to me only goes back to 1981. So far, however, I haven't found anything to provide average monthly data for slightly larger geographic units.

If anyone out there knows of publicly accessible, reliable, data sets on monthly average temperatures for anything less than the whole U.S. but more than individual reporting stations, I'd like to know about it.

Like I said previously, I actually like doing this sort of thing!