Prudence in estimating coherence between planetary, solar and climate oscillations

TL;DR

This paper critically examines claims of correlation between planetary, solar, and climate oscillations, finding that high coherence estimates are not statistically significant and dismissing the planetary hypothesis due to lack of physical mechanism.

Contribution

It introduces a rigorous significance testing approach for coherence estimates and demonstrates that previous high coherence claims are not statistically valid.

Findings

High coherence estimates at certain periods are not statistically significant.

Monte Carlo simulations show no significant coherence between datasets.

The planetary hypothesis for climate influence is dismissed due to lack of evidence.

Abstract

There are claims that there is correlation between the speed of center of mass of the solar system and the global temperature anomaly. This is partly grounded in data analysis and partly in a priori expectations. The magnitude squared coherence function is the proper measure for testing such claims. It is not hard to produce high coherence estimates at periods around 15--22 and 50--60 years between these data sets. This is done in two independent ways, by wavelets and by a periodogram method. But does a coherence of high value mean that there is coherence of high significance? In order to investigate that, four different measures for significance are studied. Due to the periodic nature of the data, only Monte Carlo simulation based on a non-parametric random phase method is appropriate. None of the high values of coherence then turn out to be significant. Coupled with a lack of a physical mechanism that can connect these phenomena, the planetary hypothesis is therefore dismissed.