The Pittsburgh Tribune-Review features a story about meteorologist Anthony Watts of Chico, CA who is going around the country checking the accuracy of 1221 weather sites used by National Oceanic & Atmospheric Administration (NOAA). Watts believes the accuracy of data from many stations has been compromised by their surroundings.
To assure accuracy, stations (essentially older thermometers in little four-legged wooden sheds or digital thermometers mounted on poles) should be 100 feet from buildings, not placed on hot concrete, etc. But as photos on Watts' site show, the station in Forest Grove, Ore., stands 10 feet from an air-conditioning exhaust vent. In Roseburg, Ore., it's on a rooftop near an AC unit. In Tahoe, Calif., it's next to a drum where trash is burned.
His findings can be found at surfacestations.org.
Here are a couple of pics from his homepage showing a well-placed weather site versus one that is almost certainly compromised by surrounding objects.
Notice the difference in trend lines between the uncompromised station and the one surrounded by asphalt, air conditioning exhausts and other buildings. Coincidence?