Simultaneously Constraining Climate Sensitivity and Aerosol Radiative Forcing

Abstract

An energy balance climate model with latitudinal, surface–air, and land–sea resolution is coupled to a two-dimensional (latitude–depth) ocean model and used to simulate changes in surface and surface air temperature since 1765. The climate model sensitivity can be prescribed by adjusting the parameterization of infrared radiation to space, and sensitivities corresponding to an equilibrium, global average warming to a CO2 doubling (ΔT2×) of 1.0° to 5.0°C are used here. The model is driven with various combinations of greenhouse gas (GHG), fossil fuel aerosol, biomass aerosol, solar, and volcanic forcings. The fossil fuel aerosol forcing is concentrated in the NH, while the biomass aerosol forcing is centered near the equator. The variation in the global mean air temperature, and in the NH minus SH temperature, is examined over the period 1856–2000, in order to simultaneously constrain both climate sensitivity and aerosol forcing. The model performance, compared to observations, is evaluated using three statistical measures. It is possible to identify a group of experiments that performs better than other experiments, but it cannot be claimed that any member of the group is better than any other member in a statistically rigorous manner. The different statistical measures and temperature variables (global mean, NH − SH, NH, or SH temperature) give slightly different groups of “more accurate” experiments. Based on the statistical measures and examination of the time series of model-simulated global mean and NH − SH temperature variation, the following conclusions can be drawn: (i) The most likely ΔT2× is around 2°C, which is at the lower end of the range of 2.1°–4.8°C obtained by recent general circulation models; (ii) the fossil fuel aerosol forcing is unlikely to have exceeded −1.0 W m−2 in the global mean by 1990; and (iii) the net biomass plus soil dust aerosol forcing is unlikely to have exceeded −0.5 W m−2 in the global mean by 1990. As an independent check of these conclusions, it was found that the simulated change of ocean heat content (over the 0–3000-m depth interval, during the period 1948–98) agrees well with the observed change in ocean heat content for climate sensitivity and aerosol forcing combinations that produce a good simulation of the observed temperature change during this time period, thereby validating the model uptake of heat by the oceans. Although the preferred ΔT2× is 2°C in this study, it is possible to choose fossil and biomass aerosol forcing combinations (within the ranges given above) that produce comparable simulations of global mean and NH − SH temperature variation after the 1880s for any ΔT2× in the range 1.0°–5.0°C. However, and in common with other models, this model simulates much too large a drop in temperature during the 1880s (in response to the eruption of Mount Krakatau in 1883). As ΔT2× ranges from 1.0° to 5.0°C, the simulated drop ranges from about 0.3° to about 0.7°C, compared to an observed change of about 0.2°C. On this basis, a lower ΔT2× is preferred. Inasmuch as the model response to the 1991 eruption of Mount Pinatubo accords well with observations, especially for intermediate and high sensitivities, it may be that the estimated radiative forcing due to the eruption of Krakatau is too large or that there was a short-term negative feedback, dependent on conditions just before this eruption, which reduced the effective radiative forcing. If half the base case forcing is assumed for Krakatau only, the temperature decrease during the 1880s ranges from 0.2°C for ΔT2× = 1°C (matching observations) to 0.3°C for ΔT2× = 5°C (modestly in excess of observations). Thus, the volcanic radiative forcing during the 1880s, and the quality of the historical and proxy temperature records around this time, are critical data in discriminating between different climate sensitivities, inasmuch as a smaller volcanic forcing might permit ΔT2× at the high end of the 1°–5°C range.