54
Chapter 4: The Instrumental Data Record
in the observed record of land and ocean surface temperature changes. No
trends in the time series of the statistic, R(t), were found.
4.2 Homogeneity
In this section we define homogeneity according to Conrad and Pollak (1962).
In this they state that a numerical series representing the variations of a
climatological element is called homogeneous if the variations are caused only
by variations of weather and climate. Although ob servers may take readings
with meticulous care, non-climatic influences can easily affect the readings.
The following sectio'hs discuss the most important causes of non-homogeneity.
• Changes in instrument, exposure and measurement technique
• Changes in station location
• Changes in observation times and the methods used to calculate monthly
averages
• Changes in the station environment with particular reference to urbanization.
4.2.1 Changes in Instrumentation, Exposure
and Measuring Techniques
For land-based temperature, the effects of changes in thermometers have been
slight so far this century. More important to the long-term homogeneity is the
exposure of the thermometers. Readings now are generally from thermometers located in louvred screened enclosures, which are generally painted white.
Earlier readings often were from shaded walliocations, sometimes under roof
awnings. These need corrections which can only be estimated by parallel
measurements or by reconstructing past locations (Chenoweth, 1992).
The greatest potential discontinuity in temperature time series is currently
being induced in first order stations in the United States. Here mercury-ingl ass thermometers have been replaced by a thermister installed in a small
louvred housing (model HO-83). The change was made by the National
Weather Service in the USA at most first order sites during early 1982. There
appears to have been little thought given concerning the climatic homogeneity of the record. Parallel measurements were not made, as is generally
recommended.
Marine temperature data are seriously affected by homogeneity problems.
SST data were generally taken before World War II by collecting some sea
water in an uninsulated canvas bucket and measuring the temperature. As
there was a short time between sampling and reading, the water in the bucket
cooled slightly by evaporative means. Since WWII most readings are made
Chapter 4: The Instrumental Data Record
in the observed record of land and ocean surface temperature changes. No
trends in the time series of the statistic, R(t), were found.
4.2 Homogeneity
In this section we define homogeneity according to Conrad and Pollak (1962).
In this they state that a numerical series representing the variations of a
climatological element is called homogeneous if the variations are caused only
by variations of weather and climate. Although ob servers may take readings
with meticulous care, non-climatic influences can easily affect the readings.
The following sectio'hs discuss the most important causes of non-homogeneity.
• Changes in instrument, exposure and measurement technique
• Changes in station location
• Changes in observation times and the methods used to calculate monthly
averages
• Changes in the station environment with particular reference to urbanization.
4.2.1 Changes in Instrumentation, Exposure
and Measuring Techniques
For land-based temperature, the effects of changes in thermometers have been
slight so far this century. More important to the long-term homogeneity is the
exposure of the thermometers. Readings now are generally from thermometers located in louvred screened enclosures, which are generally painted white.
Earlier readings often were from shaded walliocations, sometimes under roof
awnings. These need corrections which can only be estimated by parallel
measurements or by reconstructing past locations (Chenoweth, 1992).
The greatest potential discontinuity in temperature time series is currently
being induced in first order stations in the United States. Here mercury-ingl ass thermometers have been replaced by a thermister installed in a small
louvred housing (model HO-83). The change was made by the National
Weather Service in the USA at most first order sites during early 1982. There
appears to have been little thought given concerning the climatic homogeneity of the record. Parallel measurements were not made, as is generally
recommended.
Marine temperature data are seriously affected by homogeneity problems.
SST data were generally taken before World War II by collecting some sea
water in an uninsulated canvas bucket and measuring the temperature. As
there was a short time between sampling and reading, the water in the bucket
cooled slightly by evaporative means. Since WWII most readings are made