9.5.4.2 Global Precipitation Changes

The increased atmospheric moisture content associated with warming might be expected to lead to increased global mean precipitation (Section 9.5.4.1). Global annual land mean precipitation showed a small, but uncertain, upward trend over the 20th century of approximately 1.1 mm per decade (Section 3.3.2.1 and Table 3.4). However, the record is characterised by large inter-decadal variability, and global annual land mean precipitation shows a non-significant decrease since 1950 (Figure 9.18; see also Table 3.4).

Figure 9.18. Simulated and observed anomalies (with respect to 1961-1990) in terrestrial mean precipitation (a), and zonal mean precipitation trends 1901-1998 (b). Observations (thick black line) are based on a gridded data set of terrestrial rain gauge measurements (Hulme et al., 1998). Model data are from 20th-century MMD integrations with anthropogenic, solar and volcanic forcing from the following coupled climate models (see Table 8.1 for model details): UKMO-HadCM3 (brown), CCSM3 (dark blue), GFDL-CM2.0 (pale green), GFDL-CM2.1 (pale blue), GISS-EH (red), GISS-ER (thin black), MIROC3.2(medres) (orange), MRI-CGCM2.3.2 (dark green) and PCM (pink). Coloured curves are ensemble means from individual models. In (a), a five-year running mean was applied to suppress other sources of natural variability, such as ENSO. In (b), the grey band indicates the range of trends simulated by individual ensemble members, and the thick dark blue line indicates the multi-model ensemble mean. External influence in observations on global terrestrial mean precipitation is detected with those precipitation simulations shown by continuous lines in the top panel. Adapted from Lambert et al. (2005).

9.5.4.2.1 Detection of external influence on precipitation

Mitchell et al. (1987) argue that global mean precipitation changes should be controlled primarily by the energy budget of the troposphere where the latent heat of condensation is balanced by radiative cooling. Warming the troposphere enhances the cooling rate, thereby increasing precipitation, but this may be partly offset by a decrease in the efficiency of radiative cooling due to an increase in atmospheric CO₂ (Allen and Ingram, 2002; Yang et al., 2003; Lambert et al., 2004; Sugi and Yoshimura, 2004). This suggests that global mean precipitation should respond more to changes in shortwave forcing than CO₂ forcing, since shortwave forcings, such as volcanic aerosol, alter the temperature of the troposphere without affecting the efficiency of radiative cooling. This is consistent with a simulated decrease in precipitation following large volcanic eruptions (Robock and Liu, 1994; Broccoli et al., 2003), and may explain why anthropogenic influence has not been detected in measurements of global land mean precipitation (Ziegler et al., 2003; Gillett et al., 2004b), although Lambert et al. (2004) urge caution in applying the energy budget argument to land-only data. Greenhouse-gas induced increases in global precipitation may have also been offset by decreases due to anthropogenic aerosols (Ramanathan et al., 2001).

Several studies have demonstrated that simulated land mean precipitation in climate model integrations including both natural and anthropogenic forcings is significantly correlated with that observed (Allen and Ingram, 2002; Gillett et al., 2004b; Lambert et al., 2004), thereby detecting external influence in observations of precipitation (see Section 8.3.1.2 for an evaluation of model-simulated precipitation). Lambert et al. (2005) examine precipitation changes in simulations of nine MMD 20C3M models including anthropogenic and natural forcing (Figure 9.18a), and find that the responses to combined anthropogenic and natural forcing simulated by five of the nine models are detectable in observed land mean precipitation (Figure 9.18a). Lambert et al. (2004) detect the response to shortwave forcing, but not longwave forcing, in land mean precipitation using HadCM3, and Gillett et al. (2004b) similarly detect the response to volcanic forcing using the PCM. Climate models appear to underestimate the variance of land mean precipitation compared to that observed (Gillett et al., 2004b; Lambert et al., 2004, 2005), but it is unclear whether this discrepancy results from an underestimated response to shortwave forcing (Gillett et al., 2004b), underestimated internal variability, errors in the observations, or a combination of these.

Greenhouse gas increases are also expected to cause enhanced horizontal transport of water vapour that is expected to lead to a drying of the subtropics and parts of the tropics (Kumar et al., 2004; Neelin et al., 2006), and a further increase in precipitation in the equatorial region and at high latitudes (Emori and Brown, 2005; Held and Soden, 2006). Simulations of 20th-century zonal mean land precipitation generally show an increase at high latitudes and near the equator, and a decrease in the subtropics of the NH (Hulme et al., 1998; Held and Soden, 2006; Figure 9.18b). Projections for the 21st century show a similar effect (Figure 10.12). This simulated drying of the northern subtropics and southward shift of the Inter-Tropical Convergence Zone may relate in part to the effects of sulphate aerosol (Rotstayn and Lohmann, 2002), although simulations without aerosol effects also show drying in the northern subtropics (Hulme et al., 1998). This pattern of zonal mean precipitation changes is broadly consistent with that observed over the 20th century (Figure 9.18b; Hulme et al., 1998; Allen and Ingram, 2002; Rotstayn and Lohmann, 2002), although the observed record is characterised by large inter-decadal variability (Figure 3.15). The agreement between the simulated and observed zonal mean precipitation trends is not sensitive to the inclusion of forcing by volcanic eruptions in the simulations, suggesting that anthropogenic influence may be evident in this diagnostic.

Changes in runoff have been observed in many parts of the world, with increases or decreases corresponding to changes in precipitation (Section 3.3.4). Climate models suggest that runoff will increase in regions where precipitation increases faster than evaporation, such as at high northern latitudes (Section 10.3.2.3 and Figure 10.12; see also Milly et al., 2005; Wu et al., 2005). Gedney et al. (2006) attribute increased continental runoff in the latter decades of the 20th century in part to suppression of transpiration due to CO₂-induced stomatal closure. They find that observed climate changes (including precipitation changes) alone are insufficient to explain the increased runoff, although their result is subject to considerable uncertainty in the runoff data. In addition, Qian et al. (2006) simulate observed runoff changes in response to observed temperature and precipitation alone, and Milly et al. (2005) demonstrate that 20th-century runoff trends simulated by the MMD models are significantly correlated with observed runoff trends. Wu et al. (2005) demonstrate that observed increases in arctic river discharge are reproduced in coupled model simulations with anthropogenic forcing, but not in simulations with natural forcings only.

Mid-latitude summer drying is another anticipated response to greenhouse gas forcing (Section 10.3.6.1), and drying trends have been observed in the both the NH and SH since the 1950s (Section 3.3.4). Burke et al. (2006), using the HadCM3 model with all natural and anthropogenic external forcings and a global Palmer Drought Severity Index data set compiled from observations by Dai et al. (2004), are able to formally detect the observed global trend towards increased drought in the second half of the 20th century, although the model trend is weaker than observed and the relative contributions of natural external forcings and anthropogenic forcings are not assessed. The model also simulates some aspects of the spatial pattern of observed drought trends, such as the trends across much of Africa and southern Asia, but not others, such as the trend to wetter conditions in Brazil and northwest Australia.

9.5.4.2.2 Changes in extreme precipitation

Allen and Ingram (2002) suggest that while global annual mean precipitation is constrained by the energy budget of the troposphere, extreme precipitation is constrained by the atmospheric moisture content, as predicted by the Clausius-Clapeyron equation. For a given change in temperature, they therefore predict a larger change in extreme precipitation than in mean precipitation, which is consistent with the HadCM3 response. Consistent with these findings, Emori and Brown (2005) discuss physical mechanisms governing changes in the dynamic and thermodynamic components of mean and extreme precipitation and conclude that changes related to the dynamic component (i.e., that due to circulation change) are secondary factors in explaining the greater percentage increase in extreme precipitation than in mean precipitation that is seen in models. Meehl et al. (2005) demonstrate that tropical precipitation intensity increases are related to water vapour increases, while mid-latitude intensity increases are related to circulation changes that affect the distribution of increased water vapour.

Climatological data show that the most intense precipitation occurs in warm regions (Easterling et al., 2000) and diagnostic analyses have shown that even without any change in total precipitation, higher temperatures lead to a greater proportion of total precipitation in heavy and very heavy precipitation events (Karl and Trenberth, 2003). In addition, Groisman et al. (1999) demonstrate empirically, and Katz (1999) theoretically, that as total precipitation increases a greater proportion falls in heavy and very heavy events if the frequency remains constant. Similar characteristics are anticipated under global warming (Cubasch et al., 2001; Semenov and Bengtsson, 2002; Trenberth et al., 2003). Trenberth et al. (2005) point out that since the amount of moisture in the atmosphere is likely to rise much faster as a consequence of rising temperatures than the total precipitation, this should lead to an increase in the intensity of storms, offset by decreases in duration or frequency of events.

Model results also suggest that future changes in precipitation extremes will likely be greater than changes in mean precipitation (Section 10.3.6.1; see Section 8.5.2 for an evaluation of model-simulated precipitation extremes). Simulated changes in globally averaged annual mean and extreme precipitation appear to be quite consistent between models. The greater and spatially more uniform increases in heavy precipitation as compared to mean precipitation may allow extreme precipitation change to be more robustly detectable (Hegerl et al., 2004).

Evidence for changes in observations of short-duration precipitation extremes varies with the region considered (Alexander et al., 2006) and the analysis method employed (Folland et al., 2001; Section 3.8.2.2). Significant increases in observed extreme precipitation have been reported over some parts of the world, for example over the USA, where the increase is similar to changes expected under greenhouse warming (e.g., Karl and Knight, 1998; Semenov and Bengtsson, 2002; Groisman et al., 2005). However, a quantitative comparison between area-based extreme events simulated in models and station data remains difficult because of the different scales involved (Osborn and Hulme, 1997). A first attempt based on Frich et al. (2002) indices used fingerprints from atmospheric model simulations with prescribed SST (Kiktev et al., 2003) and found little similarity between patterns of simulated and observed rainfall extremes, in contrast to the qualitative similarity found in other studies (Semenov and Bengtsson, 2002; Groisman et al., 2005). Tebaldi et al. (2006) report that eight MMD 20C3M models show a general tendency towards a greater frequency of heavy precipitation events over the past four decades, most coherently at high latitudes of the NH, broadly consistent with observed changes (Groisman et al., 2005).