Global Warming: Correcting the Data: Surface Temperatures Are Rising, but Probably Not as Quickly as Is Claimed
Michaels, Patrick J., Regulation
All historical temperature records agree: the planet is warmer than it was. But those histories are subject to a number of biases, some of which are obvious while others are very subtle. The most obvious bias is that weather stations in cities do not global warming" in order to report warmer temperatures. The city's ever-increasing amounts of bricks, buildings, and pavement retain heat, absorb more of the sun's energy, and impede the flow of ventilating winds. There is nothing new about this type of local warming, which is quite distinct from a wholesale warming of the planetary surface.
These two concepts, global and locally induced warming, were first elucidated in a landmark paper published in 1933 by J. B. Kincer, rifled "Is Our Climate Changing? A Study of Long-Term Temperature Trends." Kincer worked for the U.S. Weather Bureau (forerunner of today's National Weather Service) and published his paper in the journal Monthly Weather Review. Since then, scientists studying global temperature history have taken pains to remove the biases caused by "urban warming."
CORRECTING THE RECORD
In principle, this correction is simple. Compare two neighboring weather stations. Their temperatures should oscillate in unison from year-to-year. But if one station displays a warming trend when the other does not, then the first station is assumed to be contaminated by "urban bias."
Many attempts have been made to deal with this and other factors that can bias a temperature history. Thomas Karl of the U.S. National Climatic Data Center published a landmark paper relating population levels and "artificial" warming in the Journal of Climate in 1987. Karl then developed what he called the "Historical Climate Network" (HCN) to gather temperature data that would be free of bias.
Prior to the HCN, the (lower-48) U.S. average temperature was computed by averaging the temperature readings from 344 multi-county aggregates, known as "climatological divisions" (CDs). There are currently 11,000 individual weather stations run mainly by "cooperative observers" who are usually individuals from families with a long history of staying in one place. They are provided equipment by the National Weather Service and they report their data to the National Climatic Data Center in Asheville, N.C. Their data, along with a few National Weather Service and airport locations, populate the CD record.
The CD records were simple averages of all readings from all the stations within each division. The records were contaminated by many factors. For example, sometimes the instruments were moved, or a tree shaded the weather station, or a new, nearby parking lot was built. Aside from the urban effect, the HCN attempted to account for the other biasing factors by looking for "discontinuities" in histories, indicating some sudden site-related changes, as well as changes in the time of observation during the day.
How can the time of observation of the day's mean temperature (the average of the high and the low for the previous 24 hours) bias a long-term temperature history? Almost all cooperative observer stations now contain electronic thermometers that automatically record the high and low temperature for each 24-hour period beginning at midnight. But historically, highs and lows were recorded "manually" at approximately the same time each day. Thermometers had mechanical stops in them that displayed the high and low temperatures until they were manually reset. Observers would choose a time, usually in the morning or the late afternoon, to record and reset. The fact that most observers chose the early morning introduced a very subtle "time-of-day" bias into the histories. Mean daily temperatures taken the "old-fashioned" way turn out to be colder than they actually were. Imagine recording the temperature on a recordbreaking cold winter morning. The result? Two record lows, one recorded at 7:00 a. …