| 英文摘要: | Studies often assume that climate is equally sensitive to emissions of warming greenhouse gases and cooling sulphate aerosols. Now, research illustrates that this is not true in models and that without this assumption recent assessments would have produced higher estimates of future temperatures. How bad can it be? When discussing anthropogenic climate change this question stimulates heated debate — in the scientific literature as much as anywhere else. It is an important question for climate policy and climate economics, but at its heart it is a question of climate physics. Partly as an attempt to provide a generic answer, substantial research efforts are directed towards quantifying the sensitivity of the earth's climate to increasing greenhouse gases. Yet recent studies have painted conflicting pictures. Some conclude that the climate is unlikely to be highly sensitive1, 2 whereas others conclude that it is unlikely to have a low sensitivity3, 4. Which are right? Writing in Nature Climate Change, Drew Shindell reports5 that removing a key simplifying assumption would have led some recent studies to conclude that climate is more sensitive than they did. We are all familiar with the vast and beautiful complexity of climate. To facilitate the quantitative study of climate change, however, simplifications must be made. One such simplification uses the increase in average global temperature in response to doubling atmospheric carbon dioxide as a representation of the sensitivity of the whole system. The temperature change may be considered after the system has come to a new equilibrium — equilibrium climate sensitivity (ECS) — or at the point of CO2 doubling, following a 1% per year increase in CO2 concentrations — transient climate response (TCR). These sensitivities are useful as representations of the average global response resulting from some forcing (for example, increasing atmospheric CO2 concentrations) and its related feedbacks (for example, changes in clouds or in land surface characteristics). They provide valuable, if somewhat blunt, instruments for considering the consequences of increasing atmospheric greenhouse gas concentrations. They are widely used in evaluations of climate economics, policy and impacts. For more than a decade researchers have been producing probability distributions for climate sensitivity, often using observations to provide constraints on a variety of models. A plethora now exist. The credibility of complicated global climate models (GCMs) is often discussed in terms of where their sensitivities lie within such distributions. This is also important for policy evaluations because such models are central to many assessments of climate change — not least the IPCC, whose latest report includes an atlas of projections formed from GCM output. Recent studies of physical processes in GCMs have concluded that the GCMs that are more realistic in some significant features relating to clouds and atmospheric convection are also those that tend to show higher sensitivities3, 4. In contrast, recent studies based on statistical assessments of observations and simple models have concluded that the most sensitive GCMs are the least consistent with observations1, 2. Shindell's analysis demonstrates that the conclusions of the latter studies are a consequence of their assumption that the sensitivity of climate to sulphate aerosols is the same as that to well-mixed greenhouse gases (WMGHGs) such as CO2 — something that he shows not to be the case in the latest GCMs. Removing this assumption, he derives a value of 1.7 °C (95% confidence interval of 1.3–3.2 °C) for TCR, which is considerably higher than the 1.3 °C (90% confidence interval of 0.9–2.0 °C) reported in one recent study1. Of particular relevance for economic assessments and risk-centred perspectives of climate change is that the upper temperature bound has increased more than the central value. How does this change come about? Sulphate aerosols provide a characteristic geographical pattern of negative (cooling) forcing, mostly located over land in the Northern Hemisphere (Fig. 1). This pattern leads to stronger feedbacks than those resulting from the same amount of positive (warming) forcing from WMGHGs that are fairly uniformly distributed around the world. The effect is seen in all the GCMs studied — the model TCR for aerosols and ozone is on average about 50% greater than for WMGHGs. Now consider simple models. In these models sensitivity is prescribed rather than generated internally. The same value is used for all forcing types, whether cooling or warming. This means that if sensitivity were chosen to accurately reflect the twentieth century warming due to increasing WMGHGs, then the sulphate aerosol cooling would be underestimated and therefore the total warming in the model would be greater than that observed. Better agreement with observations is achieved with lower values of sensitivity, which also underestimate WMGHG warming.
|
