Meehl, who's thin and tan and looks like a farmer turned scientist, is used to dealing with uncertainty. One of NCAR's thirty or so computer climate modelers, he received his Ph.D. in climate studies from the University of Colorado in 1987 while working on a Department of Energy grant to study the effect of burning fossil fuels. "We can't run the real system forward in time," he says, stating the obvious, "so we used computers to do it for us."
Today those computers--and often Meehl himself--are at the center of the storm surrounding global warming.
The prediction that man's contribution to greenhouse gases will raise the average world temperature is based largely on computer models. Essentially, climate modelers take the factors that contribute to weather patterns--energy from the sun, precipitation, land and ocean temperatures--and build a mathematical model that re-creates global climate processes as closely as possible. The scientists then plug in a doubling of the amount of carbon dioxide, a greenhouse gas that traps heat, and let the computer project the future shock.
Computers have been used in weather forecasting since the 1950s, mostly to look ahead a few days. Even today, most "long-range" forecasts on television newscasts generally stick to five days, because that's how far you can predict a specific weather pattern (for instance, will a storm off the Pacific Northwest cause snow in Colorado a few days later?). "And despite what people think, they've gotten pretty good," says Meehl, adding that the modelers are also making progress in the area of seasonal forecasts.
It's the long, multi-decade predictions that spark heated controversy.
Meehl admits that climate modeling has limitations. The largest, perhaps, involves clouds, which computer modelers have not been able to fit into their equations with any assurance of accuracy. Although clouds usually have a cooling effect on temperatures, some types of clouds hold in heat. Under NCAR's global-warming scenario, increased heat would cause more evaporation, which would in turn create more clouds. But would those clouds have a net cooling or warming effect?
Another problem is that the current computers, as fast and powerful as they are, deal with generalities over a large land area. Data fed into the computer comes from grid points on a map, but the grids are so large that one over Oregon, for instance, would include (moving west to east) the coast, the coastal range, the Willamette Valley, the Cascades and the desert of the Great Basin--which have widely varying climates.
Critics of climate modeling note that without fudging the data, computers have been unable to recreate present climate patterns--much less accurately predict future patterns. Computer models from various institutions agree more with each other than they do with the real world, Meehl acknowledges; in other words, they tend to make the same errors. For example, computer models have been able to predict El Nino occurrences, but for some reason the predictions are smaller than the real thing by as much as one-third.
But technological progress is being made, Meehl adds. Recently, NCAR modelers were able to match computer models to present conditions without fiddling with data. While that's not yet predicting the future accurately--most modelers agree the computers are ten years from getting that right--it's a step closer.
"The way I view this is that climate modelers are giving their best estimate with the tools we have," Meehl says. "The policy-makers can take that information, along with twenty other pieces of evidence from other sources, and use it as a whole to make their decisions."
The best decision they could make regarding the threat of global warming, Meehl says, is to adopt what has become known as a "no-regrets policy." Rather than making massive, economy-shaking commitments to cut emissions quickly, the no-regrets strategy calls for smaller steps to wean ourselves from fossil fuels--which will run out sooner or later, whether or not global warming is a real danger. That means focusing on more energy-efficient cars and buildings, and emphasizing research and development of other energy sources.
Meehl blames policy-makers and the media for exaggerating the accuracy claims of computer models--and for fanning the flames of debate. "There is a real danger in overselling the science," he says. Although the U.S. government spent about $1.7 billion in 1997 alone on computer-model development, federal funding doesn't fuel the alarming predictions about global warming, he insists. "If, by perfecting the models, I can 'prove' what's going to happen in fifty years," he adds, "then I've solved the problem. And that would be the end of my funding."
And the government isn't the only entity funding research. The energy industry has also joined the fray, sponsoring a campaign of "disinformation and personal attacks" to discredit global-warming proponents, Meehl says, adding, "I don't think anybody on the science side anticipated the response of the energy industry."