Section 4.2: Homogeneity
55
in the intake pipes which take water on to ships to cool the engines. The
change appears to have been quite abrupt, occurring within a month during
December 1941. The sources of ship's observations in the two marine data
banks (COADS and UKMO) changed from principally European to mainly
American at this time (for a discussion of marine data, see Bottomley et al.,
1990 and Woodruff et al., 1987).
Studies of the differences between the two methods indicate that bucket
temperatures are cooler than intake ones by 0.3 to 0.7°C (James and Fox,
1972). A correction technique for SST bucket measurements has been derived based on the physical principles related to the causes of the cooling
(Bottomleyet al. 1990; Folland and Parker, 1994). The cooling depends on
the prevailing meteorological conditions, and so depends on the time of year
and on location. Although a day-to-day phenomenon, the various influences
on the bucket are basically linear, so cooling amounts can be estimated on
a monthly basis. Assuming that there has been no major change in the seasonal cycle of SSTs over the last 100 years, the time between sampling and
reading can be estimated as that which best minimises the residual seasonal
cycle in the pre-WWII data. The estimate of between 3 and 4 minutes is
both quite consistent spatially and agrees with instructions given to marine
observers (Bottomleyet al. 1990; Folland and Parker, 1991, 1994).
4.2.2 Changes in Station Locations
It is rare that a temperature recording site will have remained in exactly the
same position throughout its lifetime. Information concerning station moves
is of primary importance to homogenization. Such information about a station's history is referred to as metadata. Assessment of station homogeneity
requires nearby station data and appropriate metadata.
4.2.3
Changes in Observation Time and the Methods
U sed to Calculate Monthly A verages
In the late 19th and early 20th centuries, there was considerable discussion in
the climatologicalliterature concerning the best method of calculating true
daily and, hence, monthly average temperatures (e.g. EIlis, 1890; Donnell,
1912; Hartzell, 1919; Rumbaugh, 1934). Many schools of thought existed,
and unfortunately, no one system prevailed in all regions. At present, there
is no uniform system, although the majority of nations use the average of
the daily maximum and minimum temperatures to estimate monthly mean
temperatures. The remainder use a variety of formulae based on observations at fixed hours, some of which are designed to simulate the true daily
mean that would be estimated from readings every hour. There would be no
need to correct readings to a common standard, however, provided the observing systems had remained internally consistent. Unfortunately, only in a
55
in the intake pipes which take water on to ships to cool the engines. The
change appears to have been quite abrupt, occurring within a month during
December 1941. The sources of ship's observations in the two marine data
banks (COADS and UKMO) changed from principally European to mainly
American at this time (for a discussion of marine data, see Bottomley et al.,
1990 and Woodruff et al., 1987).
Studies of the differences between the two methods indicate that bucket
temperatures are cooler than intake ones by 0.3 to 0.7°C (James and Fox,
1972). A correction technique for SST bucket measurements has been derived based on the physical principles related to the causes of the cooling
(Bottomleyet al. 1990; Folland and Parker, 1994). The cooling depends on
the prevailing meteorological conditions, and so depends on the time of year
and on location. Although a day-to-day phenomenon, the various influences
on the bucket are basically linear, so cooling amounts can be estimated on
a monthly basis. Assuming that there has been no major change in the seasonal cycle of SSTs over the last 100 years, the time between sampling and
reading can be estimated as that which best minimises the residual seasonal
cycle in the pre-WWII data. The estimate of between 3 and 4 minutes is
both quite consistent spatially and agrees with instructions given to marine
observers (Bottomleyet al. 1990; Folland and Parker, 1991, 1994).
4.2.2 Changes in Station Locations
It is rare that a temperature recording site will have remained in exactly the
same position throughout its lifetime. Information concerning station moves
is of primary importance to homogenization. Such information about a station's history is referred to as metadata. Assessment of station homogeneity
requires nearby station data and appropriate metadata.
4.2.3
Changes in Observation Time and the Methods
U sed to Calculate Monthly A verages
In the late 19th and early 20th centuries, there was considerable discussion in
the climatologicalliterature concerning the best method of calculating true
daily and, hence, monthly average temperatures (e.g. EIlis, 1890; Donnell,
1912; Hartzell, 1919; Rumbaugh, 1934). Many schools of thought existed,
and unfortunately, no one system prevailed in all regions. At present, there
is no uniform system, although the majority of nations use the average of
the daily maximum and minimum temperatures to estimate monthly mean
temperatures. The remainder use a variety of formulae based on observations at fixed hours, some of which are designed to simulate the true daily
mean that would be estimated from readings every hour. There would be no
need to correct readings to a common standard, however, provided the observing systems had remained internally consistent. Unfortunately, only in a