A Pseudoproxy Evaluation of Bayesian Hierarchical Modeling and Canonical Correlation Analysis for Climate Field Reconstructions over Europe*

Abstract

Abstract A pseudoproxy comparison is presented for two statistical methods used to derive annual climate field reconstructions (CFRs) for Europe. The employed methods use the canonical correlation analysis (CCA) procedure presented by Smerdon et al. and the Bayesian hierarchical model (BHM) method adopted from Tingley and Huybers. Pseudoproxy experiments (PPEs) are constructed from modeled temperature data sampled from the 1250-yr paleo-run of the NCAR Community Climate System Model (CCSM) version 1.4 model by Ammann et al. Pseudoproxies approximate the distribution of the multiproxy network used by Mann et al. over the European region of interest. Gaussian white noise is added to the temperature data to mimic the combined signal and noise properties of real-world proxies. Results indicate that, while both methods perform well in areas with good proxy coverage, the BHM method outperforms the CCA method across the entire field and additionally returns objective error estimates.