Passage IV

NATURAL SCIENCE: This passage is adapted from the book The Change in the Weather: People, Weather, and the Science of Climate by William K. Stevens (@1999 by William K. Stevens) .

Basically, a general circulation model is a computer program incorporating the fundamental equations expressing the atmosphere's behavior. The physical laws embodied in the equations govern the interlinked workings of the sun, the atmosphere, the oceans, the land, and all other elements of the climate system. Temperature, wind, pressure, clouds, snow, ice, water vapor, convection, turbulence, soil moisture, vegetation's effect on climate, ocean salinity, ocean heat, ocean currents, and other characteristics—plus their interaction—can all be simulated using these equations. One can enter into the computer any given set of initial conditions and set the model running. The model calculates the extent to which the basic physical laws create fluctuations in such things as temperature, precipitation, and wind over a specified time period. One then can introduce any kind of external forcing to see how the climate would change. This is what makes models useful, within limits, as tools for understanding how greenhouse gas emissions are likely to alter the global climate.

If the problem were a straightforward one of translating the warming potential of greenhouse gases to some numerical value, it would be a simple matter to calculate changes in the climate. One could almost do it on the back of an envelope. Mainstream scientists are confident that a doubling of atmospheric carbon dioxide, by itself and independently of other influences, would raise the average global temperature by about 2 degrees Fahrenheit. But as we know, there are many other influences on climate, and the system behaves in perverse and complex ways that can either strengthen or lessen the warming and also produce different effects in different parts of the world. Landmasses, snow and ice, the differential response of oceans and land to heating, the very shape of the globe in producing uneven heating and driving the general circulation—these and many other things complicate the task of prediction.

Everything is complicated further by various feedbacks, some of which enhance warming and some of which lessen it. One of these, for instance, is the role of ice and snow: Both cool the planet by reflecting solar energy. If the atmosphere warms, ice and snow melt, and their disappearance means that less of the sun's heat is reflected; the planet warms even more. This is called a positive feedback. One of the most critical feedbacks involves water vapor. When a warmer atmosphere causes more water to evaporate, the resulting vapor traps even more heat in the atmosphere than does carbon dioxide. This amplification is central to the greenhouse phenomenon; if it is less than most experts believe, the overall warming of the earth from greenhouse gases will be less than they think. There are also negative feedbacks, those that tend to counteract the warming. For example, warming increases low-level cloudiness, cooling the planet. For a model to be realistic, it must reflect with some accuracy the welter of interacting and often conflicting feedbacks.

Since the atmosphere has so many properties, and limited computer capacities make it impossible to include every detail, the art of modeling is to try to include as many climatic factors in the model as possible while at the same time excluding those that are not absolutely essential. One of the most important simplifications involves the very structure of the model itself. Models cannot possibly compute climatic changes at every point in the atmosphere. So they make calculations at widely separated points, producing a more or less crude approximation of how the climate really behaves. Conceptually, the points form a three-dimensional grid typically rising ten to twelve miles above the earth's surface. A typical spacing between grid points today is about 150 miles horizontally and less than half a mile vertically. This "resolution," as scientists call it, is about twice as fine as a decade ago. But it still misses many processes, like cloud formation, for instance, that take place between grid points. So modelers estimate them from conditions, such as temperature and humidity, within the grid space as a whole.

But the fastest computer imaginable cannot overcome another deficiency: incomplete knowledge about the workings of the atmosphere. This more than anything is what ultimately limits the models' realism.