Global Warming Computer Models Are Still Unreliable, New Study Warns

Published October 1, 2004

Computerized models of the Earth’s climate are at the heart of the debate over how policymakers should respond to climate change. Global climate models (GCMs)–also called general circulation models–attempt to predict future climate conditions by starting with a set of assumptions about how the climate works and making guesses about what a future world might look like in terms of such factors as population, energy use, and technological development.

Numerous analysts have pointed out, however, that many of the assumptions used in modeling the climate are of dubious merit, with biases that tend to project catastrophic warming. As a consequence, these analysts argue, climate models have many limitations that make them unsuitable as the basis for developing public policy.

Study Documents Computer Limitations

Computerized climate models have very little usefulness in the formation of public policy toward climate change, particularly for policy decisions as critical as ratification of the Kyoto Protocol, according to a July 7 study, “The Science Isn’t Settled: The Limitations of Global Climate Models,” released by The Fraser Institute.

The study notes current global climate models have two significant limitations. They rely on observed data, including surface station readings, weather balloons, and satellites, which are of uncertain value and accuracy due to the short length of the record and the need for adjustments to correct for artificial discontinuities such as instrument and satellite changes. Moreover, the models project future climate trends not only by extrapolating from observed data, but by including “fudge factors” and other complex adjustments that make the projections very unreliable.

“Climate models oversimplify many poorly understood climate processes, and results from the models can be contradictory,” said Dr. Kenneth Green, author of the paper and director of risk, regulation, and environment studies at The Fraser Institute. “Clearly, the data generated do not provide a meaningful foundation on which to base sound public policy decisions, especially something as significant as the decision to ratify Kyoto.”

“Land surface temperature records are biased by the ‘urban heat island effect,'” the study notes. “Failure to account for local warming in cities led to some claims of dramatic warming in the 1980s and 1990s and, while adjustments are made today and the predictions of warming significantly reduced, some researchers believe the adjustments to be inadequate.”

Bizarre Assumptions about Economic Growth

“Scenarios of future concentrations of greenhouse gases are based on dubious assumptions about the future,” the study observes. “These scenarios depend on other models of projected growth of population, economies, and energy use. Some projections are so dubious that MIT’s Dr. Richard Lindzen, a lead author of one of the IPCC [Intergovernmental Panel on Climate Change] science reports, has referred to them as ‘children’s exercises.'”

The study continues, “As researchers Ian Castles, formerly the head of Australia’s national office of statistics, and David Henderson of the Westminster Business School and formerly the chief economist of the OECD, point out, when estimating potential future climate changes, IPCC’s modelers inappropriately compared future estimates of GDP in terms of exchange rates rather than purchasing-power parity. This produces GDP estimates that are significantly inflated, leading to estimates of greenhouse-gas producing activity that are similarly inflated. Castles observes that if such assumptions are correct, then the average income of South Africans will have overtaken that of Americans by a very wide margin by the end of the century. Because of this economic error, the IPCC scenarios of the future also suggest that relatively poor developing countries such as Algeria, Argentina, Libya, Turkey, and North Korea will all surpass the United States.”

Green notes, “Canada’s ratification of the Kyoto Protocol, which many Canadian legislators vow to reverse, relied largely on frightening scenarios generated by computer climate models that are simply not sophisticated enough to serve as meaningful guides to instituting public policy. Though politicians … claim that ‘the science is solid,’ even a cursory inspection of the many problems with computer climate models suggests it is anything but.”

Reality Check

Green makes several recommendations that he says would provide a “reality check” on the science of climate modeling:

  • Reexamine the science of climate change and stop grounding policy in the output of computer models of limited utility.
  • Redirect some resources from greenhouse gas reduction efforts toward research efforts to improve the state of weather and climate forecasting.
  • Acknowledge that published scenarios of future greenhouse gas concentrations are skewed toward improbably high growth in emissions and, therefore, climate models using those scenarios will tend to project unrealistically intense warming.
  • Acknowledge that models cannot accurately predict the absolute amount of warming (or other climate change) resulting from a particular scenario of greenhouse gas concentrations.
  • Recognize that some climate changes (both natural and human-caused) are climate surprises, events that are not anticipated in advance (and, by definition, are not properly incorporated into models).
  • Perform full and transparent economic and risk analyses of the costs and effectiveness of proposed greenhouse gas control actions, including alternatives.
  • Redirect some resources away from greenhouse gas controls and toward researching probabilities of different climate change outcomes.
  • Redirect some of the resources currently focused on greenhouse gas mitigation toward research programs that will help people adapt to climate change regardless of origin.

For more information …

The July 7 Fraser Institute study, “The Science Isn’t Settled: The Limitations of Global Climate Models,” is available online at
ScienceIsntSettled.pdf. Author Kenneth Green can be contacted by email at
[email protected].