As discussed in Chapter 2, there is less confidence in observed variations in hydrological indicators than for surface temperature, because of the difficulties in taking such measurements and the small-scale variations of precipitation. There is general consistency between the changes in mean precipitation in the tropics over the last few decades and changes in ENSO. There is no general consistency between observed changes in mean tropical precipitation and model simulations. In middle and high latitudes in the Northern Hemisphere, the observed increase in precipitation is consistent with most model simulations. Observed changes in ocean salinity in the Southern Ocean appear to be consistent with increased precipitation there, as expected from model simulations (Wong et al., 1999; Banks et al., 2000).
The observed increases in the intensity of heavy precipitation in the tropics and in convective weather systems described in Chapter 2 are consistent with moist thermodynamics in a warmer atmosphere and model simulations. Observed increases of water vapour in the lower troposphere in regions where there is adequate data coverage are also consistent with model simulations. As discussed in Chapter 7, different theories suggest opposite variations of water vapour in the upper troposphere associated with an increased greenhouse effect and surface warming. The quality, amount and coverage of water vapour data in the upper troposphere do not appear to be sufficient to resolve this issue.
In middle and high latitudes of both hemispheres, there has been a trend over the last few decades towards one phase of the North Atlantic Oscillation/Arctic Oscillation and of the Antarctic high latitude mode, sometimes also referred to as "annular modes", (Chapter 2; Thompson et al., 2000). These are approximately zonally symmetric modes of variability of the atmospheric circulation. Both trends have been associated with reduced surface pressure at high latitudes, stronger high latitude jets, a stronger polar vortex in the winter lower stratosphere and, in the Northern Hemisphere, winter warming over the western parts of the continents associated with increased warm advection from ocean regions. The trend is significant and cannot be explained by internal variability in some models (Gillett et al., 2000b). These dynamical changes explain only part of the observed Northern Hemisphere warming (Gillett et al., 2000b; Thompson et al., 2000). Modelling studies suggest a number of possible causes of these circulation changes, including greenhouse gas increases (Fyfe et al., 1999; Paeth et al., 1999; Shindell et al., 1999) and stratospheric ozone decreases (Graf et al., 1998; Volodin and Galin, 1999). Some studies have also shown that volcanic eruptions (Graf et al., 1998; Mao and Robock, 1998; Kirchner et al., 1999) can induce such changes in circulation on interannual time-scales. Shindell et al. (2001) show that both solar and volcanic forcing are unlikely to explain the recent trends in the annular modes.
The majority of models simulate the correct sign of the observed trend in the North Atlantic or Arctic Oscillation when forced with anthropogenic increases in greenhouse gases and sulphate aerosols, but almost all underestimate the magnitude of the trend (e.g., Osborn et al., 1999; Gillett et al., 2000b; Shindell et al., 1999). Some studies suggest that a better resolved stratosphere is necessary to simulate the correct magnitude of changes in dynamics involving the annular modes (e.g., Shindell et al., 2001).
The combination of independent but consistent evidence should strengthen our confidence in identifying a human influence on climate. The physical and dynamical consistency of most of the thermal and hydrological changes described above supports this conclusion. However, it is important to bear in mind that much of this evidence is associated with a global and regional pattern of warming and therefore cannot be considered to be completely independent evidence.
An elicitation of individual experts' subjective assessment of evidence for climate change detection and attribution is being carried out (Risbey et al., 2000). This will help to better understand the nature of the consensus amongst experts on the subject of climate change attribution.
Other reports in this collection