Evidence-based Forecasting for Climate Change Policies

In order to meet policy makers’ need for climate forecasts, this paper extends the application of evidence-based forecasting of global mean temperatures. The extensions utilize more years of global mean temperature data and 34 years of better data. Forecasts from the no-trend model were compared with forecasts from six moresophisticated methods that take account of more than the most recent data point to generate forecasts on the basis of recent levels, trends, and patterns. The more-sophisticated methods provided inconsequential improvements in forecast accuracy for horizons up to 10 years, beyond which horizon the no-change model provided the most accurate forecasts. The findings reinforce earlier conclusions that the IPCC scenario of dangerous increases in global mean temperatures fails to pass basic validation tests, and that the no-trend model provides the only scientific long-term climate forecasts. Without a scientific forecast of a global warming, there is no rational basis for government regulations, subsidies, funding, or penalties. Furthermore, a scientific forecast of global warming is not, by itself, a basis for action. That would require scientific forecasts that the warming would be harmful and forecasts that cost-effective policies would reduce the harm.