EXPLORING “THE FOUNDATIONS” (Part 2 in the series) PDF Print E-mail

5 November 2009: In our first post of the series (below), we quickly introduced the concepts of how temperature histories are created, either by instrument or by “proxy.” We also noted that proxies require certain assumptions and caveats.

The United States started tracking climate data in 1890, the year the U.S. Weather Bureau was formed. Many nations have instrument networks, but the US network has long been considered one of the best in the world; those in other nations have suffered in recent decades. Concerns are mounting that the US station network is not really as good or extensive as the public has been lead to believe. An important concern we will examine in a later post is “urban heat island effect” (UHI).

In the summer of 2007, the SurfaceStations.org project was founded by a 25-year broadcast meteorology veteran. Using volunteers, official weather service guidance (NOAA's Climate Reference Network Site Handbook) was used to assess the US temperature sensing network reliability. Little NOAA documentation was available on the trustworthiness of its own system. As of this writing, the survey project has explored about 90% of the stations in the USHCN (the United States Historical Climate Network), “1,221 high quality stations.” Findings include temperature-monitoring equipment located in the shade, near burn barrels and air conditioners, near large expanses of as phalt (many at airports), and near buildings. Some stations are physically missing (but generating data). To be fair, some stations are actually located in great sites, away from artiicial heat sources and things that could bias the temperature readings, but the results illustrate large-scale problems:

Station Site Quality by Rating
Rating Quality Error # Stations % of Total
CRN = 1 Best <1°C 26 3%
CRN = 2 Good <1°C 68  8%
CRN = 3 Fair >1°C 171  20%
CRN = 4 Poor >2°C 496  58%
CRN = 5 Worst >5°C 94  11%
In summary:
  • about 1 site in 10 have an expected error less than a degree Celsius
  • 70% (7 in 10) have errors more than 2 degrees
  • one in 10 could have readings up to 5 degrees from reality.

The entire record appears unreliable. Put another way, that’s a lot of error when we’re being told that temperatures by the year 2100 could be anywhere up to about 6 degrees warmer than today, and that the 20th Century saw a rise of less than 1 degree...

For more information, see:


Summary report on the project: (PDF)

Peer-reviewed papers are being prepared on the project.

Last Updated on Monday, 25 January 2010 19:28