58 Climates ofthe past
A number of research groups in different countries have made careful
and independent analyses of these observations. In somewhat different
ways they have made allowances for factors that could have introduced
artificial changes in the records. For instance, the record at some land
stations could have been affected by changes in their surroundings as
these have become more urban. In the case of ships, the standard method
of observation used to be to insert a thermometer into a bucket of water
taken from the sea. Small changes of temperature have been shown to
occur during this process; the size of the changes varies between day and
night and is also dependent on several other factors including the material
from which the bucket is made – over the years wooden, canvas and metal
buckets have been variously employed. Nowadays, a large proportion of
the observations are made by measuring the temperature of the water
entering the engine cooling system. Careful analysis of the effects of
these details on observations both on land and from ships has enabled
appropriate corrections to be made to the record, and good agreement
has been achieved between analyses carried out at different centres.
Confidence that the observed variations arereal is increased by notic-
ing that the trend and the shape of the changes are similar when different
selections of the total observations are made. For instance, the separate
records from the land and sea surface and from the northern and south-
ern hemispheres are closely in accord. Further indirect indicators such
as changes in borehole temperatures and sub-surface ocean tempera-
tures, decrease in snow cover and glacier shrinkage provide independent
support for the observed warming.
During the last thirty years or so observations have been available
from satellites orbiting around the Earth. Their great advantage is that
they automatically provide data with global coverage, which are often
lacking in other data sets. The length of the record from satellites, how-
ever, is generally less than twenty years, a comparatively short period in
climate terms. It has been suggested that satellite measurements of lower
atmospheric temperature since 1979 are not consistent with the trend of
rising temperatures in surface observations. The satellite observations
do not cast doubt on the accuracy of the surface measurements – the two
measurements are of different quantities. However, it is expected that
trends in lower atmosphere measurements and surface measurements
should be related and much careful work has been carried out studying
the two records and interpreting the differences (see box).
The most obvious feature of the climate record illustrated in
Figure 4.1 is thatof considerable variability, not just from year to year,
but from decade to decade. Some of this variability will have arisen
through causes external to the atmosphere and the oceans, for instance
as a result of volcanic eruptions such as those of Krakatoa in 1883 or of