Forecasting the Climate of the New Century.

Computer simulations are our best window into the climate of the future, but our view is obscured by uncertainties in the models based on an incomplete understanding of how the climate system works. Recent research has led to a greatly increased understanding of the uncertainties in today's climate models. In attempting to predict the climate of the twenty-first century, we must confront not only computer limitations on the affordable spatial resolution of global models, but also a lack of physical realism in attempting to model key climate processes. Until we are able to incorporate adequate treatments of critical elements of the entire biogeophysical climate system, our models will remain subject to these uncertainties, and our scenarios of future climate change, both anthropogenic and natural, will not fully meet the requirements of either policy makers or the public. The areas of most-needed model improvements are thought to include air-sea exchanges, land surface processes, ice and snow physics, hydrolog ic cycle elements, and especially the role of aerosols (small particles), and cloud-radiation interactions.

Of these areas, cloud-radiation interactions are known to be responsible for much of the inter-model differences in sensitivity to green-house gases. In current research, scientists in many countries are diagnostically evaluating current and proposed model cloud-radiation treatments against extensive field observations. Satellite remote sensing provides an indispensable component of the observational resources. The researchers find that newly developed advanced schemes with explicit cloud-water budgets and interactive cloud-radiative properties are potentially capable of matching observational data closely. However, it is difficult to evaluate the realism of model-produced fields of cloud-radiative properties, cloud-liquid water content and effective cloud-droplet radius until high quality measurements of these quantities become more widely available. Thus, further progress will require a combination of theoretical and modelling research, together with intensified emphasis on both in situ and space-based rem ote sensing observations.

From far away, the Earth appears as a mainly blue ocean planet flecked with constantly changing intricate patterns of white clouds. The clouds, which cover about half the surface of the planet, are critical to climate. They cool the Earth by reflecting away sunlight. At the same time, they warm the Earth by trapping it's heat, thus contributing to the natural greenhouse effect.

In computer modelling of the global climate system, which is our main technique for forecasting future climate, we first have to understand climate science well enough to build crucial physical processes into our models realistically. Among the key science areas, treating clouds accurately is an especially critical issue, because clouds have such important effects on climate. However, they are much too small and short-lived to be modelled explicitly in a global simulation, and the physical processes involved are still imperfectly understood. Nevertheless, clouds and their consequences are much too important to be ignored, and so their effects are treated by simple rules.

Of course, clouds are also a source of water in the form of rain and snow, critical to all aspects of life and central to agriculture and thus to the well-being of humankind. If climate changes, whether for natural or man-made reasons, clouds are likely to change as well and have important feedbacks on the climate.

Climate models are solved numerically on global grids with a typical resolution of a few hundred kilometers horizontally. When the model values of climate variables at this resolution are used to prescribe clouds, the resulting rules are often highly simplistic. For example, a typical rule might make the cloud amount in a model grid area proportional to relative humidity. However, the behaviour of rules of this sort often bears little resemblance to the way clouds actually vary in space and time in the present atmosphere, let alone as to how they will change and feed back on any future climate which differs from the present one. These rules are rarely well-founded theoretically and have almost never been tested observationally in any thorough and satisfying way.

It is crucial that observational and modelling research be done to show how clouds behave in the actual atmosphere. Only recently have appropriate observations begun to be available to tackle this task. These observations provide invaluable information for the development of improved models. Space-based data have been critical to progress. Supporting the research to obtain and use these data optimally is a key to further progress in making future climate projections reliable, detailed and useful to policy makers.

The results of global climate models (GCMs) are sensitive to small-scale physical processes. GCM representations of such key processes are held to a higher standard today than only a few years ago, because it has become widely recognized that the role of cloud feedbacks is crucial in determining the sensitivity of GCM climates to changes in the atmospheric concentration of carbon dioxide and other greenhouse gases, and to the direct and indirect effects of atmospheric aerosols. Aerosols are small particles which can reflect and absorb radiation (the direct effect) and can also serve as nuclei on which water can condense to form clouds (the indirect effect). Early GCMs had no aerosols and had fixed clouds, prescribed once and for all as functions of latitude and altitude only. These early models had a low sensitivity to prescribed carbon dioxide changes. When cloud amounts are allowed to vary, several positive feedbacks come into play. One is the reduction of cloud amount in a warmer climate, a typical GCM re sult. This change produces an augmented warming, because GCM clouds, like global average real clouds, have a larger contribution to the planetary albedo (reflectivity) than to the planetary greenhouse effect.

A separate class of positive feedbacks arises from an increase in average cloud altitude, which is also a characteristic GCM response to climatic warming. On average, higher clouds are less effective infrared-radiative emitters to space because they are colder, so they add to the greenhouse effect, trapping energy in the atmosphere. Furthermore, they also are less effective reflectors of incoming solar radiation because they are generally less bright than lower clouds. In combination with the reduction in cloud amount, the increase in cloud height has the effect of powerfully reinforcing the warming which gives rise to these changes in cloud amount. It is the GCMs which incorporate these effects most strongly and which do not include compensating negative feedbacks that produce the highest sensitivities to changes in green-house gas concentrations.

The GCMs which have the lowest sensitivities on the other hand, are those which are dominated by negative cloud feedbacks. One such feedbacks is due to hypothetical cloud micro-physical processes. Chief among these is a possible increase in cloud-water content accompanying a warmer climate. If, in fact a warmer climate is characterized by wetter clouds, then, other things being equal, the wetter clouds, then, other things being equal, the wetter clouds may produce a negative feedback by increasing the reflectivity (or albedo) of lower clouds, or a positive feedback by increasing the infrared-radiative emissivity of higher clouds such as cirrus.

Which of these conflicting effects will dominate? Is it scientifically justifiable to focus on global average cloud feedbacks to the neglect of regional ones? How do these cloud feedbacks depend on season, or weather regime or geographical locale? What are the dynamical and thermodynamic consequences of the physical processes which depend on cloud feedbacks? These questions form much of the research agenda of the climate community. Even if anthropogenic climate change were not such an important policy issue, cloud processes would play an important role in climate modelling, because they have such a pervasive influence on climate, as evidenced by recent GCM results.

In recent years, scientists throughout the world have diagnostically evaluated several current and proposed model cloud-radiation treatments against extensive field observations. Satellite remote sensing provides an indispensable component of the observational resources. The researchers find that newly developed advanced treatments of clouds with explicit cloud-water budgets and interactive cloud-radiative properties are potentially capable of matching observational data closely. However, it is difficult to evaluate the realism of model-produced fields such as cloud extinction, cloud emittance, cloud-liquid water content and effective cloud-droplet radius until high-quality measurements of these quantities become more widely available. Thus, further progress will require a combination of theoretical and modeling research, together with emphasis on both in situ and space-based remove sensing observations.

We may draw several far-reaching conclusions from current climate research results:

* Predicting the future of Earth's climate, which can changed for both natural and human-related reasons, requires an improved understanding of key physical processes such as those related to clouds. Satellite measurements have been essential to progress.

* The research needed to achieve this understanding is now under way and the needed measurements are beginning to be made, but much more effort is needed to make key observations and use them to improve computer simulations. Space-based measurements will continue to have a key role to play.

* Even after the research has been advanced and the models have become more trustworthy, producing climate predictions with sufficient realism for policy makers will require vastly increased computer power and continuing scientific and political leadership.

Richard C. J. Somerville is Professor of Meteorology at Scripps Institution of Oceanography, University of California, San Diego, United States.