Clouds are important in the Earth's climate system because of their effects on solar radiation, terrestrial radiation and precipitation. Different cloud types contribute to total cloud amount and are associated with a wide variety of thermal and dynamic processes in the climate system (see Chapter 7, Section 7.2.2). Therefore knowing the variations in total cloud amount and different cloud types would significantly contribute to improving our understanding of the role of clouds in contemporary climate change. Several analyses of cloud amounts for regions of the world have been performed since the SAR. Problems with data homogeneity, particularly concerning biases with changing times of observation (Sun and Groisman, 2000; Sun et al., 2001) have been addressed in several studies, but other issues continue to be a source of uncertainty.
Dai et al. (1997a, 1999) and Kaiser (1998) examined cloud cover changes over the former USSR and China during the last four to five decades, to add to earlier analyses for Europe, the United States, Canada, and Australia by Henderson-Sellers (1992) and Karl and Steurer (1990). These studies show 20th century increases in cloud cover over much of the United States (mostly confined to the first 80 years) and the former USSR, which are significantly negatively correlated with changes in the diurnal range of surface air temperature (DTR) (as shown earlier in Figure 2.3). Sun and Groisman (2000) showed that in the former USSR low-level cloud cover significantly decreased during the period 1936 to 1990. However, this was more than offset by a significant increase in cumulus and cirrus clouds during the past several decades. Over much of China, however, daytime and night-time total cloud cover exhibited significant decreasing trends of 1 to 2% sky cover/decade for both day and night observations between 1951 and 1994 (Kaiser, 1998, 2000), which the DTR failed to follow (Figure 2.3). This discrepancy may result from the increasing effect of industrial aerosols on the DTR since the late 1970s (Dai et al., 1999). Tuomenvirta et al. (2000) show increasing trends in cloud cover during the period 1910 to 1995 for northern Europe, which are consistent with decreases in the DTR. A new analysis (Neff, 1999) reveals a dramatic increase (15 to 20%) of spring and summer cloud amount at the South Pole during the past four decades in this region. This appears to be related to the observed delay in the breakdown of the spring polar vortex and is believed to be related to decreases in stratospheric temperatures.
There are few analyses of the amounts of various cloud types or changes over the tropics and sub-tropics. Correlations with observed precipitation and clouds observed by satellites suggest that much of the increase in the total cloud amount is likely to have resulted from increases in thick, precipitating clouds (Dai et al., 1997a). Hahn et al. (1996) show decreasing decadal scale trends in cloud cover over much of China, as well as over most of South America and Africa for the period 1971 to 1991. The latter two areas have little surface-based information.
The SAR presented analyses of inter-decadal changes in marine cloud coverage. The data have now been re-examined and doubt has been cast on some of the previous findings (Bajuk and Leovy, 1998a; Norris, 1999). Additional data have also reversed some of the previous trends. In the SAR a 3% increase in cumulonimbus clouds was reported for the period 1952 to 1981. An update of this analysis showed a gradual rise in cumulonimbus cloud amount from the mid-1950s to the mid-1970s, with a gradual decline thereafter (Bajuk and Leovy, 1998a). Bajuk and Leovy (1998b) cast doubt on the homogeneity of the cloud amounts derived from ship data. They find that inter-decadal variations of the frequency of occurrence of cloud amount for a given cloud type are generally unrelated to similar time-scale variations in SST and large-scale divergence of the surface winds. Nonetheless, some regional changes and variations based on ship reports of low and middle clouds are likely to be rather robust. Variations in these categories of cloud are consistent with variations of other climate system variables. Examples include: (1) a long-term upward trend in altostratus and nimbostratus across the mid-latitude North Pacific and North Atlantic Oceans (Parungo et al., 1994; Norris and Leovy, 1995); (2) ENSO related variations in the frequency of low cloud types across the Pacific and Indian Oceans (Bajuk and Leovy, 1998b); and (3) interannual variations in summer season stratiform clouds across the North Pacific (Norris et al., 1998). Norris (1999) found an increase in total sky cover of approximately 2%, and an increase of approximately 4% in low cloud cover in his analyses of ship reports between 1952 and 1995. He finds no evidence for changes in observation practices that may have affected these trends. The trends are dominated by a globally consistent mode and are as large or larger in the tropics and Southern Hemisphere as in the Northern Hemisphere. This argues against attribution to increased anthropogenic aerosol amounts.
Although satellite estimates of changes and variations in cloud amount and type contain systematic biases, Rossow and Schiffer (1999) showed improved calibration and cloud detection sensitivities for the International Satellite Cloud Climatology Program (ISCCP) data set. Using data from 1983 to 1994, a globally increasing trend in monthly mean cloudiness reversed during the late 1980s and early 1990s. There now appears to be an overall trend toward reduced total cloud amounts over both land and ocean during this period. An estimate for aircraft-induced cirrus cover for the late 1990s ranges from 0 to 0.2% of the surface of the Earth (IPCC, 1999).
Since the SAR, land surface precipitation has continued to increase in the Northern Hemisphere mid- and high latitudes; over the sub-tropics, the drying trend has been ameliorated somewhat. Where data are available, changes in annual streamflow relate well to changes in total precipitation. Over the Southern Hemisphere land areas no pronounced changes in total precipitation are evident since the SAR. The changes in precip-itation in mid- and high latitudes over land have a strong correl-ation with long-term changes in total cloud amount. Little can be said about changes in ocean precipitation as satellite data sets have not yet been adequately tested for time-dependent biases. Changes in water vapour have been analysed most for selected Northern Hemisphere regions, and show an emerging pattern of surface and tropospheric water vapour increases over the past few decades, although there are still untested or uncorrected biases in these data. Limited data from the stratosphere also suggest increases in water vapour but this result must be viewed with great caution. Over land, an increase in cloud cover of a few per cent since the turn of the century is observed, which is shown to closely relate to changes in the diurnal temperature range. Changes in ocean cloud amount and type show systematic increases of a few per cent since the 1950s, but these relate poorly to SST or surface wind divergence changes, casting some doubt on the integrity of the trends. No changes in observing practices can be identified, however, that might have led to time-dependent biases in the ocean cloud amount and frequency statistics.
Other reports in this collection