Forecasting experts say IPCC warming reports violate principles of good forecasting.

Dr. J. Scott Armstrong and Dr. Kersten Green are co-directors of the Forecasting Principles website. Armstrong is at the prestigious Wharton School, University of Pennsylvania and Green is at the Business and Economic Forecasting Unit at Monash University. Armstrong is founding editor of the Journal of Forecasting and the author of Long-Range Forecasting. Armstrong and Green, along with dozens of other experts on forecasting also produced a handbook, Principles on Forecasting.

Armstrong and Green recently looked (PDF) at the forecasting methods used by the Intergovernmental Panel on Climate Change, which has been the principle lobbying body for the claims of anthropogenic global warming. They were unhappy with what they saw. They argue that basic principles of accurate forecasting were ignored by the IPCC.

One argument that these forecasters make is that “complex models (those involving nonlinerarities and interactions) harm accuracy because their errors multiple.” They argue that the more complex the model the more opportunities for error there are. And the complexity alone makes it difficult to find those errors. The models have numerous parts, each of which rely on all the other parts being accurate. An error in one section can create a chain-reaction of errors. The full extent of the error is only known when in real time. A forecast of warming, in 100 years, of X degrees, is only verified, or fully disproved, in 100 years time.

Green and Armstrong note that models become even less reliable when the “prediction intervals are enormous”. “For example, prediction intervals (ranges outside which outcomes are unlikely to fall) expand rapidly as time horizons increase, so that one is faced with enormous intervals even when trying to forecast a straightforward thing such as automobile sales for General Motors over the next five years.” My weather predictions are pretty much spot on provided I am predicting the weather in the next few minutes and looking out my window. The longer range my prediction the lower my accuracy rate. Global warming theory is dependent on extremely complex models projected far into the future.

Green and Armstrong suggest that the more uncertain the forecasting the greater the reason to be conservative. They note that forecasters have found that predictions of no change are more likely to be right when there is substantial uncertainty.

There is a difference between a forecast by scientists and a scientific forecast. Routinely forecasters find that experts have only a slightly better chance, than non-experts, of accurately predicting a trend. Many warming predictions were based on surveys of the opinions of experts but, of late, the doomsayers have been using computer models. Armstrong and Green argue that these models are “mathematical ways for the experts to express their opinions” and that there is “no empirical evidence to suggest that presenting opinions in mathematical terms rather than in words will contribute to forecast accuracy.”

Neither Green nor Armstrong could find any references to the forecasting principles used by the IPCC. No literature on forecasting was referenced. “It is hard to understand how scientific forecasting could be conducted without any reference to the literature on how to make such forecasts. At a minimum, one would expect to see some empirical justification for the forecasting methods that were used.”

The Forecasting Audit has identified 140 principles for successful forecasting. Looking at the IPCC report, Armstrong and Green said 127 of those principles were relevant. But there was “insufficient information to rate the forecasting procedures that were used against 38 of these principles.” That left 89 principles with sufficient evidence to evaluate how well the IPCC did in following forecasting principles. They found that the IPCC “violated 72” of the 89 principles.

Green and Armstrong contend that the IPCC report violated so many principles of good forecasting that it is basically worthless.