THERE'S a sure-fire way to find out what human activity is doing to our planet: We can wait and see what happens. But if we let the system perform the experiment - to use climatologist Stephen H. Schneider's metaphor - it may be too late to repair the environmental damage.
``We have become part of the Earth system and one of the forces for global change,'' according to NASA's Earth System Sciences Committee (ESSC). ``These changes seem irreversible,'' warns Francis P. Bretherton, ESSC chairman and Dr. Schneider's colleague at the National Center for Atmospheric Research (NCAR). He adds: ``Unless they are perceived and understood in a timely manner, all are likely to be stressful. We must start now to lay the foundation of knowledge and understanding for our successors to act upon.''
In short, humanity must prepare to actively manage the planet.
To do this, we need a way to foresee what effect human activity has on Earth's environment and to anticipate how changes in that activity would forestall dangerous results. Computer simulation - such as the climate modeling work of Schneider and his colleagues at NCAR - is the only way this can be done. The trick, Schneider says, is to use this computer guidance effectively in spite of its uncertainty.
HE explains: ``[There are] people in the policy business who are demanding that we give them answers. And they don't want the answers after the system performs the experiment. Of course, the truth is we can't give them answers before [then]. We can give them scenarios. And we can define the probabilities of those scenarios.'' He adds: ``I think the biggest problem in policy is people expect us to say `yes' or `no,' what's going to happen. ... One of the reasons we have policy difficulties ... is that people have the [wrong] idea that we're going to ultimately have exactly the answer. And I think that we're not. What we're going to do is find the ranges of possibilities and consequences.''
Schneider illustrates this with the problem of carbon dioxide (CO2) and climate. This heat-trapping gas slowly accumulates in the atmosphere as we burn fossil fuels. It slows the escape of heat to space, slightly raising Earth's average surface temperature. This is the so-called greenhouse effect.
As Schneider points out, there's nothing controversial about the greenhouse effect. It's one of the best established theories in planetary science.
`WHAT is controversial,'' he says, ``is whether the small [temperature] increment is going to make a large enough difference to be of social and environmental importance.'' This uncertainty - plus the fact that any significant effect is unlikely to come until well into the next century - leaves policymakers wondering what, if anything, they should do about CO2 emissions.
Schneider notes that computer-generated scenarios give no clear pictures of what might happen to regional water supplies as CO2 accumulates. ``Current betting is slight reductions in the US. But I have to warn that the current betting has only a 51 percent chance of being right, in my opinion. Two out of three models agreeing does not constitute proof, by any stretch, on hydrology. ... [But] models agreeing on temperature does [constitute proof] because, in fact, it's not two out of three but it's 22 out of 23.''
What the computer studies can do, Schneider explains, is show there's a reasonable chance of substantial climatic change. Then, he adds, ``[we] can re-ask the question and ask what are the vulnerabilities of the system. ... What biological systems are vulnerable to changes of what kind? And then ask the physical scientists if there's any probability to those changes.''
Given the uncertainties about CO2's climatic impact, Schneider says it's too soon to try to work out a world policy to curb use of fossil fuels. Among other things, use of these fuels is embedded in the global economy. Asking third-world countries, in particular, to drastically restrict the burning of coal and oil is to limit their economic development.
However, he adds, the simulations do suggest that some action be taken - action that is in the nature of insurance and that would be worth taking anyway.
FOR example, some computer-generated scenarios suggest there'll be longer growing seasons in high latitudes. Some agriculture could move up there if, say, American wheat lands dry out. So, Schneider notes, it would make sense to start developing the seeds for crops that can handle the variable weather of those latitudes.
In general, he says, there's a case to be made for more fuel conservation, especially in developed countries, to slow down CO2 accumulation. ``There's nothing new in that,'' he observes, `` ... you'd have less air pollution, less acid rain, less dependence on foreign supply.'' Also, he adds, ``You develop alternative crop strategies. You develop and test them and find out which ones can work better in wet or dry [conditions] and so forth. And my gosh, that's actually useful in dealing with regular climate variability.''
Schneider believes that, in the long run, climate change implies redestributing resources. With advance warning and a range of alternatives from which to choose, this need not be disastrous. In fact, it might turn out to be a good thing if, for example, the fertilizing influence of increased CO2 concentration in the air led to a more vigorous agriculture. But, Schneider warns, ``if it happens out of ignorance and ... you have to abandon present practices without knowing what to go to, then you probably hurt in the transition.''
What this comes down to is a series of small investments to make it easier to adapt to possible climate changes. Where computer modeling comes in, Schneider explains, ``is evaluating the probabilities and consequences of future scenarios and the efficacy of various policy choices - whether they'll, in fact, achieve certain stated goals.''
While such modeling is admittedly a ``dirty crystal ball,'' he says it raises a significant question: ``How long do we clean the glass before acting on the things we see inside? It's a great clich'e. But it's true.''
Computer models: peering into a climatologist's crystal ball
FOR climatologist Warren Washington, computer modeling is the next best thing to time travel for studying Earth's past and future climates.
Dr. Washington explains that, as used in his work at the National Center for Atmospheric Research and elsewhere, such computational clairvoyance ``allows you to try to make quantitative estimations of what factors affect the climate - whether or not those are naturally caused or caused by man.''
In fact, he adds, ``to put it into a comprehensive global picture, where all the interactions are taking place simultaneously - the only real tool that you've got is the computer models of the atmosphere and the oceans and so forth.'' Tangle of processes
What, then, is this marvelous tool?
It isn't any particular computer system, although the capability of the hardware limits what modeling can do. The key to understanding it lies in Dr. Washington's words ``quantitative,'' ``comprehensive,'' and ``interactions.''
For Washington and other climate modelers, these words characterize the difference between confronting an incomprehensible jumble of data that feed an impenetrable tangle of processes, and the ability to construct meaningful scenarios of what the climate probably was like in the distant past and what it may become in the future. Underlying physical laws
Climate models are computer programs that simulate the action of the atmosphere, the oceans, or the great ice fields - the major parts of the climate system.
These models embody the underlying physical laws, which are well understood. They use these laws to simulate winds and weather, ocean currents, heating and cooling, rainfall and cloudiness - processes that are sometimes not well understood at all. The effect of clouds and aerosols, volcanic particles and other kinds of dust, are especially hard to simulate. Combining simulations
Atmospheric programs, which are closely related to weather-forecasting programs, are the most advanced.
But ocean and ice-field simulations are now good enough for modelers to begin to put all three of these major climatic players together and study the climate system as a whole.
Looking at the state of this work today, Washington notes two aspects:
``One is you have to deal with unexpected events.
``For example, I'm not sure we'll have down to a science the prediction of the number of volcanoes. ... That is going to be a kind of unpredictable [factor]. ...
``Then there's going to be a class of [climatic] forcings which we will probably be able to make some pretty good estimates on.
``For example, the increases of carbon dioxide, trace gases, and so forth. We'll probably be able to make some reasonable estimates what the growth of those things are.
``And also, for example, the growth of deforestation. If we know those things - and those are fairly predictable - we can put those into a model ... and make a scenario.''
He warns that ``there's always going to be a certain amount of uncertainty in how you tell the planners and the economists what the climate of the future is.'' Future improvements
The resolution of the models contributes to this uncertainty.
Right now, they are too coarse-grained to make meaningful climate forecasts on a scale smaller than about half a continent. This resolution should sharpen as faster computers with more memory become available to scientists over the next 10 to 20 years.
In sum, Washington says, climate modeling offers ``an opportunity to look into the future in a more quantitative way than we have had in the past in terms of ... the Earth's environment.''
But, he says, ``it's not going to be a clear crystal ball.''
Last of three articles.