Working Group I: The Scientific Basis

Other reports in this collection

9.3.2 Patterns of Future Climate Change

For the change in annual mean surface air temperature in the various cases, the model experiments show the familiar pattern documented in the SAR with a maximum warming in the high latitudes of the Northern Hemisphere and a minimum in the Southern Ocean (due to ocean heat uptake) evident in the zonal mean for the CMIP2 models (Figure 9.8) and the geographical patterns for all categories of models (Figure 9.10). For the zonal means in Figure 9.8 there is consistent mid-tropospheric tropical warming and stratospheric cooling. The range tends to increase with height (Figure 9.8, middle) partly due to the variation in the level of the tropopause among the models. Ocean heat uptake also contributes to a minimum of warming in the North Atlantic, while land warms more rapidly than ocean almost everywhere (Figure 9.10). The large war

Figure 9.8: Multi-model annual mean zonal temperature change (top), zonal mean temperature change range (middle) and the zonal mean change divided by the multi-model standard deviation of the mean change (bottom) for the CMIP2 simulations. See text for scenario definitions and description of analysis technique. (Unit: °C).

Figure 9.9: Change in annual mean sea-ice thickness between the periods 1971 to 1990 and 2041 to 2060 as simulated by four of the most recent coupled models. The upper panels show thickness changes in the Northern Hemisphere, the lower panels show changes in the Southern Hemisphere. All models were run with similar forcing scenarios: historical greenhouse gas and aerosol loading, then future forcing as per the IS92a scenario. The colour bar indicates thickness change in metres - negative values indicate a decrease in future ice thickness.

ming in high latitudes of the Northern Hemisphere is connected with a reduction in the snow (not shown) and sea-ice cover (Figure 9.9).

The ensemble mean temperature divided by its standard deviation {T} / {T} provides a measure of the consistency of the climate change patterns (Section 9.2). Different types and different numbers of models enter the ensembles for the G, GS and SRES A2 and B2 cases and results will depend both on this and on the difference in forcing. Values greater than 1.0 are a conservative estimate of areas of consistent model response, as noted in Section 9.2.2 above.

There is relatively good agreement between the models for the lower latitude response, with larger range and less certain response at higher latitudes (Figure 9.10). For example, most models show a minimum of warming somewhere in the North Atlantic but the location is quite variable. There is a tendency for more warming (roughly a degree) in the tropical central and east Pacific than in the west, though this east-west difference in warming is generally less than a degree in the multi-model ensemble and is not evident with the contour interval in Figure 9.10 except in the B2 experiment in Figure 9.10e. This El Nino-like response is discussed further in Section

The biggest difference between the CMIP2 G (Figure 9.10a,b) and GS experiments (Figure 9.10c) is the regional moderating of the warming mainly over industrialised areas in GS where the negative forcing from sulphate aerosols is greatest at mid-21st century (note the regional changes discussed in Chapter 10). This regional effect was noted in the SAR for only two models, but Figure 9.10c shows this is a consistent response across the greater number of more recent models. The GS experiments only include the direct effect of sulphate aerosols, but two model studies have included the direct and indirect effect of sulphate aerosols and show roughly the same pattern (Meehl et al., 1996; Roeckner et al., 1999). The simulations performed with and without the direct sulphate effect (GS and G, respectively) with the same model are more similar to each other than to the other models, indicating that the individual response characteristics of the various models are dominating the response pattern rather than differences in the forcing. With greater CO2 forcing, the simulated patterns are more highly correlated in the G simulations than in the GS simulations (Table 9.2, 26 of 36 possible model combinations for temperature, 22 of 36 for precipitation).

Table 9.2: The pattern correlation of temperature and precipitation change for the years (2021 to 2050) relative to the years (1961 to 1990) for the simulations in the IPCC DDC. Above the diagonal: G experiments, below the diagonal: GS experiments. The diagonal is the correlation between G and GS patterns from the same model.
CGCM1 0.96 0.74 0.65 0.47 0.65 0.72 0.67 0.65 0.31
CCSR/NIES 0.75 0.97 0.77 0.45 0.72 0.77 0.73 0.80 0.49
CSIRO Mk2 0.61 0.71 0.96 0.40 0.75 0.72 0.67 0.75 0.63
ECHAM3/LSG 0.58 0.50 0.44 0.46 0.40 0.53 0.60 0.53 0.35
GFDL_R15_a 0.65 0.76 0.69 0.42 0.73 0.58 0.61 0.69 0.55
HadCM2 0.65 0.69 0.59 0.52 0.50 0.85 0.79 0.79 0.43
HadCM3 0.60 0.65 0.60 0.49 0.47 0.63 0.90 0.75 0.47
ECHAM4/OPYC 0.67 0.78 0.66 0.37 0.71 0.61 0.69 0.89 0.41
DOE PCM 0.30 0.38 0.63 0.24 0.36 0.40 0.44 0.37 0.91
CGCM1 0.88 0.14 0.08 0.05 0.05 0.23 -0.16 -0.03 0.02
CCSR/NIES 0.14 0.91 0.13 0.21 0.34 0.36 0.29 0.33 0.18
CSIRO Mk2 0.15 0.14 0.73 0.13 0.29 0.32 0.31 0.07 0.11
ECHAM3/LSG 0.20 0.23 0.13 0.39 0.28 0.19 0.11 0.11 0.29
GFDL_R15_a 0.18 0.20 0.28 0.28 0.41 0.28 0.20 0.22 0.21
HadCM2 0.34 0.34 0.23 0.37 0.24 0.73 0.19 0.24 0.17
HadCM3 -0.20 0.06 0.31 -0.05 0.11 -0.01 0.81 0.25 0.09
ECHAM4/OPYC 0.13 0.30 0.09 0.07 0.04 0.23 0.20 0.79 0.01
DOE PCM 0.02 0.08 0.12 -0.09 0.06 0.13 -0.06 -0.07 0.43

The SRES A2 and B2 integrations (Figure 9.10d,e) show a similar pattern of temperature change as the CMIP2 and G experiments. Since the positive radiative forcing from greenhouse gases overwhelms the sulphate aerosol forcing at the end of the 21st century in A2 and B2 compared to the GS experiments at mid-21st century, the patterns resemble more closely the G simulations in Figure 9.10a,b. The amplitude of the climate change patterns is weaker for the B2 than for the A2 simulations at the end of the 21st century (Figure 9.10d,e).

Figure 9.10: The multi-model ensemble annual mean change of the temperature (colour shading), its range (thin blue isolines) (Unit: °C) and the multi-model mean change divided by the multi-model standard deviation (solid green isolines, absolute values) for (a) the CMIP2 scenarios at the time of CO2-doubling; (b) the IPCC-DDC scenario IS92a (G: greenhouse gases only) for the years 2021 to 2050 relative to the period 1961 to 1990; (c) the IPCC-DDC scenario IS92a (GS: greenhouse gases and sulphate aerosols) for the years 2021 to 2050 relative to the period 1961 to 1990; (d) the SRES scenario A2 and (e) the SRES scenario B2. Both SRES scenarios show the period 2071 to 2100 relative to the period 1961 to 1990. See text for scenario definitions and description of analysis technique. In (b) and (d) the ratio mean change/standard deviation is increasing towards the low latitudes as well as in (a), (c) and (e), while the high latitudes around Antarctica show a minimum.

Continues on next page

Other reports in this collection

IPCC Homepage