TCS Daily

An Assessment Built on Guesswork

By Sallie Baliunas - June 28, 2001 12:00 AM

Scientists ask "What if...?" questions in order to find out how the world operates. But the rules of science demand thorough testing and validation of the results. Otherwise it isn't science.

On the matter of global climate change, a few scientists are making fantastic prognostications without validating the basis for their findings. A clear example is an account on the possible effects of possibly human caused global warming on the United States. The National Assessment Synthesis Team, an advisory committee to the U.S. Global Change Research Program, put together the report.

The assessment was forwarded last year to then-President Clinton and Congress, as mandated by law. It concentrated on the U.S., but its problems reflect those of reports by the United Nations International Panel on Climate Change. The IPCC reports, in turn, have been the basis for restrictions on greenhouse-gas emissions sought under the Kyoto Protocol.

While the U.S. team's purpose was to "synthesize, evaluate, and report on what we presently know about the potential consequences of climate variability and change for the U.S. in the 21st Century," the assessment is mostly guesswork.

The national assessment team went to "state-of-the-science climate models" - General Circulation Models, or GCMs, calculated with supercomputers. The results of climate forecasts for carbon dioxide added to the air were then plugged into other computer models of ecological, hydrologic and socioeconomic systems to get the final results - the guessed-at impacts of human-made global warming. There are thus two layers of extremely complex calculations that go into the assessment. Those complexities require scientific testing to be judged reliable.

And it is the impacts of projected human-made global warming - the what-if games - that make the U.S. assessment and the IPCC reports so popular with the press. The U.S. team indicated that rising temperatures would likely destroy fragile ecosystems, increase droughts in some places and flooding in others, erode coasts and reduce fishery production, among other scary things.

But the impacts must first be based on reliable forecasts of climate. For the climate portion of the calculation, the GCMs were from the Hadley Centre in Britain and the Canadian Climate Center. Using two models rather than one, the assessment report thought, "helps capture a sense of the range of conditions that may be plausible in the future." May be plausible? Does that mean anything is possible? If so, science has been forgotten.

Comparability is good; so is the use of multiple models. However, the intercomparison or averaging of results is only useful if the models have been tested and proven correct. Can the models capture present reality? If not, how can they deliver any sense of future reality? As even human-made global-warming enthusiast, Kevin Trenberth, director of the Climate Analysis Section of the National Center for Atmospheric Research, noted: "The two models used (in the national assessment) are quite different and give different results, so how can they both verify against the observed record?"

The answer is they can't, and they didn't. Neither does a very good job when attempts were made to validate them by modeling past temperature changes. Ruth Doherty and Linda Mearns of NCAR, in a supplement report to the national assessment effort, found that the simulated outcomes of the models for actual, present-day conditions for the Great Lakes and Hudson Bay area, for example, were too warm in the winter by at least 10º F; for the southwestern U.S., too cold by almost 10° F in the winter. For the Central Plains, model temperatures were too warm by 2º - 6ºF in the summer. Exaggerated summer heat may overemphasize the possibility of drought. Also, the precipitation estimated by the models is 100 to 200 percent above observed values for broad areas of North America. It's especially problematic in the mountain regions like the Colorado Rockies.

David R. Legates, associate professor in Climatology and Computational Methods at the University of Delaware, examined other shortcomings in the Canadian model and found, among other things, that Louisiana has less rain than Colorado - the opposite of reality.

How significant is the assessment? Not very. John Christy, a noted climatologist with the University of Alabama at Huntsville and the National Aeronautics and Space Administration, criticized the U.S. assessment as "an evangelistic statement about a coming apocalypse, not a scientific statement about the evolution of a complicated system with significant uncertainties." He is right. Region by region across North America, the computer simulations get the current climate conditions wrong. And North America's climate measurements are among the best in the world, so the observations cannot be in error - the models are.

Perhaps asking for climate and impact simulations region by region is unfairly stretching the models' abilities. Perhaps globally the models do a better job. But the reported estimates of global warming are the averages of area-by-area results - just those types of outcomes that fail to reproduce North America's current climate. Severely biased area results cannot be averaged to produce accurate global outcomes. If the models cannot yet successfully simulate present climate conditions, surely the future forecasts and the overlying impact calculations cannot be reliable either. Neither climate nor impacts can be credibly forecast - regionally or globally.

Not only are the models and forecasts troubling, but so is the bias that shows up in the assessment. Just as U.N. advocates of the Kyoto Protocol have argued, the U.S. team also indicated that these effects could be mitigated if human-induced emissions of greenhouse gases, like carbon dioxide, were controlled.

The writers of the assessment's overview claimed, "Greenhouse gas emissions lower than those assumed in this assessment would result in reduced impacts." That position poses as one that could be effective in mitigating temperature increases and their damaging effects. But according to controls envisioned in the Kyoto Protocol the temperature difference at the end of the next century -- if the world did nothing as compared with holding per-capita emissions at present levels -- would amount to about 0.25° C. That's not nearly enough to have any environmental impact, and could prove very expensive if done abruptly.

Problems with the models and bias are not trivial. Asking "what if" questions is vital to understanding how natural processes work and improving the science in the models. But delivering "answers" that thwart true understanding cannot serve to solve human-made global warming, or any other environmental problem.


TCS Daily Archives