Most of the warming in the 20th century occurred in two
distinct periods separated by several decades of little overall globally averaged
change, as objectively identified by Karl et al. (2000) and discussed in IPCC
(1990, 1992, 1996) and several references quoted therein. Figures
2.9 and 2.10 highlight the worldwide behaviour
of temperature change in the three periods. These linear trends have been calculated
from the Jones et al. (2001) gridded combination of UKMO SST and CRU land-surface
air temperature, from which the trends in Table 2.2
were calculated. Optimum averaging has not been used for Figures
2.9 and 2.10, and only trends for grid boxes
where reasonably complete time-series of data exist are shown. The periods chosen
are 1910 to 1945 (first warming period), 1946 to 1975 (period of little global
temperature change), 1976 to 2000 (second warming period, where all four seasons
are shown in Figure 2.10) and the 20th century,
1901 to 2000. It can be seen that there is a high degree of local consistency
between the SST and land air temperature across the land-ocean boundary, noting
that the corrections to SST (Folland and Parker, 1995) are independent of the
land data. The consistency with which this should be true locally is not known
physically, but is consistent with the similarity of larger-scale coastal land
and ocean surface temperature anomalies on decadal time-scales found by Parker
et al. (1995). The warming observed in the period from 1910 to 1945 was greatest
in the Northern Hemisphere high latitudes, as discussed in Parker et al. (1994).
By contrast, the period from 1946 to 1975 shows widespread cooling in the Northern
Hemisphere relative to much of the Southern, consistent with Tables
2.1 and 2.2 and Parker et al. (1994). Much of
the cooling was seen in the Northern Hemisphere regions that showed most warming
in 1910 to 1945 (Figure 2.9 and Parker et al., 1994).
In accord with the results in the SAR, recent warming (1976 to 2000) has been
greatest over the mid-latitude Northern Hemisphere continents in winter. However,
the updated data shows only very limited areas of year-round cooling in the
north-west North Atlantic and mid-latitude North Pacific. Over 1901 to 2000
as a whole, noting the strong consistency across the land-ocean boundary, most
warming is observed over mid- and high latitude Asia and parts of western Canada.
The only large areas of observed cooling are just south and east of Greenland
and in a few scattered continental regions in the tropics and sub-tropics.
Faster warming of the land-surface temperature than the ocean surface temperature
in the last two decades, evident in Figure 2.6,
could in part be a signal of anthropogenic warming (see Chapters
9 and 12). However, a component, at least in the Northern
Hemisphere north of 40 to 45°N, may result from the sharp increase in the
positive phase of the winter half year North Atlantic Oscillation (NAO)/Arctic
Oscillation (AO) since about 1970 (Section 2.6.5),
though this itself might have an anthropogenic component (Chapter
7). There has also been a strong bias to the warm phase of El Niño
since about 1976 (Section 2.6.2). In particular, Hurrell
and van Loon (1997) and Thompson et al. (2000a) show that the positive phase
of the NAO advects additional warm air over extra-tropical Eurasia north of
about 45°N. The positive phase of the NAO or AO is therefore likely to be
a major cause of the winter half-year warming in Siberia and northern Europe
in Figure 2.10, as also quantified by Hurrell (1996).
Cooling over the western North Atlantic Ocean also occurs, partly due to advection
of cold air in an enhanced north to north-west airflow. Hurrell (1996) also
shows that the warm phase of El Niño is associated with widespread extra-tropical
continental warming, particularly over North America and parts of Siberia, with
cooling over the North Pacific Ocean. Both effects are consistent with the strong
warming over Siberia in winter in 1976 to 2000 (Figure
2.10), warming over much of North America and cooling over the Davis Strait
region. Note that some regional details of the seasonal trends for 1976 to 2000
in Figure 2.10 may be sensitive to small changes
in record length. A test for the shorter period 1980 to 1997 showed the same
general worldwide pattern of (generally somewhat reduced) seasonal warming trends
as in Figure 2.10, but with some regional changes,
particularly over North America, almost certainly related to atmospheric circulation
fluctuations. However, Siberian trends were considerably more robust.
We conclude that in the 20th century we have seen a consistent large-scale
warming of the land and ocean surface. Some regional details can be explained
from accompanying atmospheric circulation changes.
Other reports in this collection |