Using models with long-term persistence to interpret the rapid increase of Earth’s temperature

Statistical processes with long-term persistence (LTP) offer a promising approach to modeling the natural dynamics of Earth’s temperature. One such process, the family of 1/f-noises, is used here to assess the plausibility of a natural origin for recent global warming. Following earlier studies, a model of natural variability with LTP is parameterized via paleoclimate reconstructions. The method developed here resolves a number of limitations in existing studies, primarily the problem of inaccuracies in estimating the spectral exponent of LTP. The output of the model is compared with the observed rate of temperature rise (0.61 ∘C/century between 1850 and 2007 for Northern hemispheric land air temperatures). We find that rates comparable with the observed global warming are very rarely generated by the model of natural variability (the probability is less than 2.3×10−4). Thus, natural agencies are not a plausible explanation for the observed global warming unless all the paleoclimate reconstructions through which the model is parameterized are underestimating natural variance by a factor of at least four.