Figure 8.15: Observed (Parker/Jones) and modelled global annual mean temperature anomalies (°C) from the 1901 to 1930 climatological average. The control and three independent simulations with the same greenhouse gas plus aerosol forcing and slightly different initial conditions are shown from CGCM1 (Boer et al., 2000). The greenhouse gas alone simulation is labelled GHG. The three greenhouse gas plus aerosol simulations are labelled GHG+Asol1, GHG+Asol2 and GHG+Asol3, respectively.

Figure 8.16: The geographical distribution of the linear trend in annual-mean precipitation (%/100 yr) for (a) the observations of Hulme (1992, 1994); (b) the ensemble of three greenhouse gas + aerosol integrations using the CCCma model. Taken from Boer et al. (2000).

Since the pioneer experiments conducted at the Hadley Centre for Climate Prediction and Research (Mitchell et al. 1995) and at the Deutsche Klimarechenzentrum (DKRZ) (Hasselmann et al. 1995), reported in the SAR, a number of other groups internationally have reproduced the trend in the surface air temperature instrumental record over the 20th century. These include the Canadian Center for Climate Modelling and Analysis (CCCma) (Boer et al. 2000), Centre for Climate System Research/National Institute for Environmental Studies (CCSR/NIES) (Emori et al. 1999), Commonwealth Scientific and Industrial Research Organization (CSIRO), Geophysical Fluid Dynamics Laboratory (GFDL) (Haywood et al. 1997) and the National Center for Atmospheric Research (NCAR) (Meehl et al. 2000b). Many of these new contributions, including recent experiments at the Hadley Centre and DKRZ, include an ensemble of projections over the 20th century (e.g., Figure 8.15). Such an ensemble allows for an estimate of intra-model variability, which in the case of the CCCma model (Figure 8.15), is larger than the possible anthropogenic signal through the early part of the 20th century (cf., inter-model variability shown in Figure 9.3).

Coupled models that have been used to simulate changes over the 20th century have all started with �control model� levels of atmospheric CO₂ (typically 330 ppm). This initial condition is then referred to as the �pre-industrial� initial condition. Changes in radiative forcing are then calculated by taking the observed atmospheric equivalent CO₂ level over the 20th century as a difference relative to the actual pre-industrial level (280 ppm), and adding this as a perturbation to the control model levels. Implicit in this approach is the assumption that the climate system responds linearly to small perturbations away from the present climate. Haywood et al. (1997) demonstrated the near linear response of the GFDL-coupled model to changes in radiative forcing associated with increases in atmospheric greenhouse gases and sulphate aerosols. When added together, experiments which included aerosol and greenhouse gas increases separately over the 20th century yielded a similar transient response (in terms of globally averaged and geographical distribution of surface air temperature and precipitation) to an experiment which included both aerosol and greenhouse gas increases. This analysis is particularly important as it validates the methodological approach used in coupled model simulations of the 20th century climate.

As noted in the SAR, the inclusion of the direct effect of sulphate aerosols is important since the radiative forcing associated with 20th century greenhouse gas increase alone tends to overestimate the 20th century warming in most models. Groups that have included a representation of the direct effects of sulphate aerosols have found that their model generally reproduces the observed trend in the instrumental surface air temperature warming, thereby suggesting that their combination of model climate sensitivity and oceanic heat uptake is not unrealistic (see Chapter 9, Section 9.2.1 and Figure 9.7). These same models have more difficulty representing variability observed within the 20th century instrumental record (Sections 8.6.2, 8.6.3). As mentioned in Section 8.6.3, some modelling studies suggest that the inclusion of additional forcings from solar variability and volcanic aerosols may improve aspects of this simulated variability. Delworth and Knutson (2000), on the other hand, note that one of their six 20th century integrations (using GFDL_R30_c) bears a striking resemblance to the observed 20th century warming which occurs primarily in two distinct periods (from 1925 to 1944 and from 1978 to the present), without the need for additional external forcing. In addition, all coupled models have shown a trend towards increasing global precipitation, with an intensification of the signal at the high northern latitudes, consistent with the observational record (Figure 8.16). Nevertheless, AOGCM simulations have yet to be systematically analysed for the occurrence of other key observed trends, such as the reduction in diurnal temperature range over the 20th century and the associated increase in cloud coverage.

The aforementioned studies all prescribed the temporal and geographical distribution of sulphate aerosols and included their radiative effects by perturbing the surface albedo according to the amount of sulphate loading in the atmospheric column above the surface (see Chapter 6, Sections 6.7, 6.8 and 6.14). This approach both ignores the indirect effect of these aerosols (i.e., their effects on cloud formation) as well as weather affects on aerosol redistribution and removal. Roeckner et al. (1999) made a major step forward by incorporating a sulphur cycle model into the ECHAM4(ECMWF/MPI AGCM)/OPYC3(Ocean isoPYCnal GCM) AOGCM to eliminate these shortcomings. In addition, they included the radiative forcing due to anthropogenic changes in tropospheric ozone by prescribing ozone levels obtained from an offline tropospheric chemistry model coupled to ECHAM4. The simulation of the 20th century climate obtained from this model (Bengtsson et al. 1999), which includes the indirect effect of aerosols, shows a good agreement with the general 20th century trend in warming (see Chapter 12, Section 12.4.3.3). The results of this study also suggest that the agreement between model and observed 20th century warming trends, achieved without the inclusion of the indirect aerosol effect, was probably accomplished with an overestimated direct effect or an overestimated transient oceanic heat uptake. Alternatively, since these studies include only idealised scenarios of sulphate radiative forcing alone (direct and/or indirect) that do not include the apparent effects of other aerosol types (Chapter 6), one might view the sulphate treatment as a surrogate, albeit with large uncertainty, for the radiative forcing associated with all anthropogenic aerosols.

As noted in Chapter 2, Sections 2.2.2 and 2.2.3, land surface temperatures show a greater rate of warming than do lower tropospheric air temperatures over the last 20 years (see also discussion in Chapter 2, Section 2.2.4). While noting uncertainties in the observational records (Chapter 2, Sections 2.2.2 and 2.2.3), the National Research Council (NRC) (2000) pointed out that models, which tend not to show such a differential trend, need to better capture the vertical and temporal profiles of the radiative forcing especially associated with water vapour and tropospheric and stratospheric ozone and aerosols, and the effects of the latter on clouds. Santer et al. (2000) provide further evidence to support this notion from integrations conducted with the ECHAM4/OPYC3 AOGCM (Bengtsson et al. 1999; Roeckner et al., 1999) that includes a representation of the direct and indirect effects of sulphate aerosols, as well as changes in tropospheric ozone. They showed that the further inclusion of stratospheric ozone depletion and stratospheric aerosols associated with the Pinatubo eruption lead to a better agreement with observed tropospheric temperature changes since 1979, although discrepancies still remain (see Chapter 12, Section 12.3.2 and Figure 12.4).

IPCC Homepage