Inconsistent comparison of temperature reconstructions over the Common Era

Abstract

Büntgen et al. (2020) present a new reconstruction of extra-tropical summer temperatures based on updated versions of a large number of summer temperature sensitive tree-ring width chronologies from the Northern Hemisphere (NH), which cover the full Common Era (CE). This new dataset allows the authors to draw conclusions about NH temperature history and its relation to climate forcing, marking an important contribution to our understanding of past climate changes. While we have no issues with the main conclusions of B20, here we show that their comparison with PAGES 2k reconstructions is flawed: B20′s reconstruction focused on regional, summertime temperature, while the PAGES 2k reconstruction targeted global, annual mean temperature. For their reconstruction intercomparisons, B20 rescale all six tree-ring based reconstructions to their regional observational target but fail to do this same processing step with the PAGES 2k reconstructions. This inconsistent comparison leads B20 to incorrectly conclude that the PAGES 2k reconstructions severely lack variance and are therefore unreliable. In this contribution, we present a consistent comparison of the B20 and PAGES 2k reconstructions, and we highlight the importance of careful illustrations for interpreting scientific results both in the literature and in the public discussion. Our results show that, if more accurate methods for comparisons are applied, the temperature history and low-frequency amplitudes of the different proxy selection approaches and reconstruction products are not at odds, but actually consistent with the differences between their targets over the pre-industrial CE.