There seems to be a problem with the location of the surface stations that are used to assess  (global) temperatures across a wide geographic area and over long periods (as “climatologists” see it) of time.  In particular, the US Historical Climatology Network, which has major contributions to the data sets used by “climatologists,” seems to have been giving invalid readings for quite a number of years, and at least one US agency involved in driving Federal climate policy seems to have badly “adjusted” the data these stations have been doing a bad job of providing. 

Specifically, as Anthony Watts, the lead author of the paper that investigates the implications of this error (“An area and distance weighted analysis of the impacts of station exposure on the U.S. Historical Climatology Network temperatures and temperature trends”), put it in an earlier paper,

[A]pproximately 90% of USHCN stations were compromised by encroachment of urbanity in the form of heat sinks and sources, such as concrete, asphalt, air conditioning system heat exchangers, roadways, airport tarmac, and other issues.

This is the result of the well-known urban heat island effect.  The cities grew out to surround the originally placed sensors, and nothing was done about those sitings.

In Watts’ present paper (that inspirationally titled “area and distance weighted analysis” paper), Watts used a better method of assessing the quality of the station locations, one developed by Michel Leroy of METEO-France and accepted for use by the World Meteorological Organization.

Watts’ findings:

…a spurious doubling of U.S. mean temperature trends in the 30 year data period covered by the study from 1979 – 2008.

Moreover,

Poorly sited station trends are adjusted sharply upward, and well sited stations are adjusted upward to match the already-adjusted poor stations.

Well sited rural stations show a warming nearly three times greater after NOAA adjustment is applied.

Urban sites warm more rapidly than semi-urban sites, which in turn warm more rapidly than rural sites.

And finally:

The new analysis demonstrates that reported 1979-2008 U.S. temperature trends are spuriously doubled, with 92% of that over-estimation resulting from erroneous NOAA adjustments of well-sited stations upward.

Hmm….