F. Joos and M. Bruno
Climate and Environmental Physics
Sidlerstr. 5, CH-3012 Bern, Switzerland; email@example.com
11. September 1996
The Bern Model has been designed to study the relationship between anthropogenic carbon emissions and atmospheric CO2 levels as well as the transient response of the surface temperature signal to a perturbation in the radiative balance of the Earth. The Bern Model has been used as a standard for CO2 scenario calculations and the calculations of global warming potentials by the Intergovernmental Panel on Climate Change [e.g. Houghton et al.(1996)].
Atmospheric carbon dioxide has increased since pre-industrial time by almost 30 percent due to anthropogenic activities such as fossil fuel burning and land use changes. Carbon released into the atmosphere is redistributed within the climate system. For time scales of decades-to-centuries, anthropogenic carbon is partitioned between the three reservoirs atmosphere, ocean, and land biosphere. The increased atmospheric levels of carbon dioxide and other radiative forcing agents (methane, nitrous oxide, ozone, CFCs, aerosols, etc.) lead to an increased radiative forcing on Earth's surface and to a rise in global average surface temperature. It takes decades to centuries until surface temperature has approached its new equilibrium after a perturbation of the radiative balance. This thermal inertia is caused by the large oceanic heat capacity.
The Bern model is a spatially aggregated model and represents the relatively fast exchanging carbon reservoirs atmosphere, ocean and terrestrial biosphere. The ocean uptake of heat, anthropogenic CO2 and other tracers is modeled in a consistent way and the global budget of anthropogenic CO2 is balanced. To save CPU time, scenario calculations are often done by using a pulse substitute version of the model [Joos et al.(1996)].
Oceanic tracer uptake (carbon, heat, etc) is described by the HILDA model. It includes two well-mixed surface boxes, representing low and high-latitude surface water masses, a well-mixed high-latitude deep water box and a diffusive interior reservoir [Shaffer and Sarmiento(1995)]. Air-sea and tracer transport in the ocean is described by six parameters which are determined such that the model reproduces both the oceanic distribution of natural, steady-state radiocarbon as well as of bomb-produced radiocarbon at time of the Geochemical Ocean Section Study survey. [Joos et al. (1991a), Siegenthaler and Joos (1992)]. The model was validated with CFCs and Argon-39 [Joos et al. (1991b), Joos (1992)] and the parameterization of surface-to-deep transport by eddy diffusion was evaluated for anthropogenic CO2 and radiocarbon using the GFDL ocean transport model [Joos et al. (1996)]. The terrestrial component includes representation of ground vegetation, wood, detritus and soil [Siegenthaler and Oeschger (1987)]. A possible enhancement of plant growth due to elevated CO2 level is taken into account by a logarithmic dependency between additional photosynthesis and atmospheric CO2 [Enting et al.(1994)]. Model ocean and biosphere are coupled to a well-mixed atmosphere. Model predictions for the decrease in atmospheric delta C-13 and Delta C-14 (Suess Effect) as well as the global budget of bomb-produced radiocarbon agree with observations within their error limits [Siegenthaler and Joos (1992), Joos (1994)].
Ocean uptake of the Bern version of the HILDA model compares well with results of the Princeton general circulation ocean model, a 2-dimensional dynamical model and the box-diffusion model (Fig. 2).
Figure 2: Ocean uptake of anthropogenic CO2 as calculated with the box-diffusion (dotted line, [Oeschger et al.(1975)]), the HILDA (solid, [Siegenthaler and Joos(1992)]), a dynamical 2-D model (dashed, [Stocker et al.(1994)]) and the Princeton/GFDL ocean general circulation model [Sarmiento et al.(1992)]. All results are obtained by using the mixed-layer pulse response technique [Joos et al.(1996)].
The ocean component of the Bern model is also used to calculate global heat uptake by the ocean, its thermal expansion, and trends in global average surface temperature. The air-sea coupling is described following Siegenthaler and Oeschger(1984). A coupling constant of 6.3 W/m2 is used for the logarithmic relationship between CO2 and radiative forcing. The fraction covered by land is 0.29 and the heat exchange coefficient between land and continent is set to 7.2 W/(m2 K), corresponding to an atmospheric relaxation time of 8 days. The equilibrium response of the model for a given radiative forcing, say for a doubling of pre-industrial CO2 is not modeled but prescribed according to results of atmosphere general circulation models. The ratio of the climate sensitivities over land and ocean is chosen in order to obtain a 30 percent warmer equilibrium response over land than over the sea. As a standard, the global climate sensitivity is set to 2.5 K for an increase in radiative forcing corresponding to a doubling of preindustrial atmospheric CO2 (Delta-T(2xCO2)=2.5 K).
Figure 3: Global average surface temperature warming of the GFDL A/OGCM as compared with results of the Bern model.
Figure 4: Sea level rise due to thermal expansion of the GFDL A/OGCM as compared with results of the Bern model.
In the Second Scientific Assessment of IPCC the results of 10 atmosphere/ocean general circulation models (A/OGCMs) are compared for an increase in radiative forcing by 1 percent/yr [Houghton et al.(1996), Figure 6.4,]. We have also calculated the corresponding response of the Bern model applying climate sensitivities of 2.1 and 4.6 K for doubling of CO2 , thereby bracketing the sensitivities of the 10 A/OGCMs. After 80 years, the increase in global average surface temperature is 1.6 K and 2.4 K, respectively. This compares well with the results of nine A/OGCMs (excluding one outlier) which are in the range of 1.5 to 2.7 K. The Bern model was also compared to the 2xCO2 and 4xCO2 stabilisation simulations of Manabe and Stouffer(1994). Here, we applied the climate sensitivity of 3.5 K of the GFDL model. The Bern model lags the temperature reponse of the GFDL A/OGCM by a few tenths of a degree. Thermal expansion is also smaller in the Bern model. Differences between results of the two models are less than 15 percent for globally averaged surface temperature and less than 25 percent for ocenic thermal expansion (Fig. 3 and Fig. 4) This is within the range of uncertainties associated with such calculations. To conclude, results of the Bern model in general agree with results of A/OGCMs. Recall, that ocean mixing in the Bern model has not been tuned to match any other model, rather its parameters have been derived from the observed distribution of oceanic tracers.
Climate indicators such as the globally averaged surface temperature must be interpreted with caution and results of A/OGCMs must be taken into account to properly interpret a climate indicator.
Adequately designed spatially aggregated models can be used to calculate the relationship between anthropogenic carbon emissions, atmospheric CO2 levels and simple climate indicators such as global average surface warming and oceanic thermal expansion. This is essential to perform integrated assessment of global change and cost-efficient scenario calculations.