It is interesting to note that "Average temperature" will change consideraby based upon whether the daily low and daily high are averaged together, hourly readings are taken, and what time of day the readings are taken. These have all changed considerably over the last few years.
I've seen some initial tests suggesting a large difference in temperature readings between the testing stations painted with whitewash, and painted with white latex paint. This in addition to the stations that sit adjacent to parking lots that weren't there a decade before.
You might recall here in CA where our “official” rain fall meter was used. After weeks of rain it still measured minimal increase in annual rain measurement. After weeks of searching, it was determined that the unit was leaking.
OK. What does that do to statistics.