In an article published in Energy & Environment, they examine the IPCC report of 2007 for its use of scientific forecasting principles. This is the latest U.N. report claiming that man is causing dramatic increases in world temperatures and what these increases will be over the next 92 years. Green and Armstrong conclude that these forecasts by the U.N. IPCC violate scientific forecasting principles badly and are not a good basis for developing scientific or public policy.
Chapter 8 of the report provides some information, albeit inadequate, on the forecasting methodology used by the IPCC. It turns out that a huge effort was made in the 1990s to establish what good scientific principles for forecasting were. Many forecasting scientists made a joint effort to elaborate these principles and established 140 principles. Green and Armstrong used these principles to judge the scientific methodology of the forecasts made in the IPCC report and found that there was enough information in the report to make that judgment on 89 of the 140 principles. The IPCC violated 72 of the 89 principles for which sufficient information was provided to make a judgment! Green and Armstrong say that many individual failures to follow a particular principle would of itself make nonsense of the 92 year predictions of the report.
This is a very damning evaluation of the IPCC report and the predictions it makes. "The forecasts in the Report are not the outcome of scientific procedures." "We have been unable to identify any scientific forecasts of global warming. Claims that the Earth will get warmer have no more credence than saying that it will get colder."
Some further comments of interest are:
- "Complex models (those involving nonlinearities and interactions) harm accuracy because their errors multiply."
- "Given even modest uncertainty, prediction intervals are enormous." This means that a prediction of a future value of a property will have very large errors. Predicting the sales of General Motors over the next five years is a much simpler problem than predicting average global climate over 92 years, but the error in predicting General Motors sales is still likely to be very large.
- "When there is uncertainty in forecasting, forecasts should be very conservative."
The IPCC reports are principally about pretense, not about good science.