Giant natural fluctuation models and anthropogenic warming

Abstract

Explanations for the industrial epoch warming are polarized around the hypotheses of anthropogenic warming (AW) and giant natural fluctuations (GNFs). While climate sceptics have systematically attacked AW, up until now they have only invoked GNFs. This has now changed with the publication by D. Keenan of a sample of 1000 series from stochastic processes purporting to emulate the global annual temperature since 1880. While Keenan's objective was to criticize the International Panel on Climate Change's trend uncertainty analysis (their assumption that residuals are only weakly correlated), for the first time it is possible to compare a stochastic GNF model with real data. Using Haar fluctuations, probability distributions, and other techniques of time series analysis, we show that his model has unrealistically strong low-frequency variability so that even mild extrapolations imply ice ages every ≈1000 years. Helped by statistics, the GNF model can easily be scientifically rejected.