Testing the hockey-stick hypothesis by statistical analyses of a large dataset of proxy records.

This paper is a statistical time-series investigation addressed at testing the anthropogenic climate change hypothesis known as the “hockey-stick”. The time-series components of a select batch of 258 long-term yearly Climate Change Proxies (CCP) included in 19 paleoclimate datasets, all of which running back as far as the year 2192 B.C., are reconstructed by means of univariate Bayesian Calibration. The instrumental temperature record utilized is the Global Best Estimated Anomaly (BEA) of the HADCRUT4 time series readings available yearly for the period 1850-2010. After performing appropriate data transformations, Ordinary Least Squares parameter estimates are obtained, and subsequently simulated by means of multi-draw Gibbs sampling for each year of the pre-1850 period. The ensuing Time-Varying Parameter sequence is utilized to produce high-resolution calibrated estimates of the CCP series, merged with BEA to yield Millennial-scale Time Series (MTS). Finally, the MTS are individually tested for temperature single break date and multiple peak dates. As a result, the estimated temperature breaks and peaks suggest widespread rejection of the hockey-stick hypothesis since they are mostly centered in the Medieval Warm Period.