The Science of Doom

In #6 we looked in a bit more detail at Imbers and co-authors from 2014. Natural variability is a big topic.

In this article we’ll look at papers that try to assess natural variability over long timescales – Peter Huybers & William Curry from 2006 who also cited an interesting paper from Jon Pelletier from 1998.

Here’s Jon Pelletier:

Understanding more about the natural variability of climate is essential for an accurate assessment of the human influence on climate. For example, an accurate model of natural variability would enable climatologists to make quantitative estimates of the likelihood that the observed warming trend is anthropogenically induced.

He notes another paper with this comment (explained in simpler terms below):

However, their stochastic model for the natural variability of climate was an autoregressive model which had an exponential autocorrelation dependence on time lag. We present evidence for a power-law autocorrelation function, implying larger low-frequency fluctuations than those produced by an autoregressive stochastic model. This evidence suggests that the statistical likelihood of the observed warming trend being larger than that expected from natural variations of the climate system must be reexamined.

In plain language, the paper he refers to used the simplest model of random noise with persistence, the AR(1) model we looked at in the last article.

He is saying that this simple model is “too kind” when trying to weigh up anthropogenic vs natural variations in temperature.

To see the whole article, visit the new Science of Doom on Substack page and please consider suscribing, for notifications on new articles.