Correction to “Optimal surface temperature reconstructions using terrestrial borehole data”

Abstract

[1] In the paper ‘‘Optimal surface temperature reconstructions using terrestrial borehole data’’ by Michael E. Mann, Scott Rutherford, Raymond S. Bradley, Malcolm K. Hughes, and Frank T. Keimig (Journal of Geophysical Research, 108(D7), 4203, doi:10.1029/2002JD002532, 2003), an error was made with respect to the normalization of the grid box weights used to form an areal hemispheric mean series (as pointed out by Pollack and Smerdon [2004]). The error led to an underestimate of the amplitudes of hemispheric mean trends by a factor of 1.47. The correction leads to a minor change in the ‘‘optimal’’ hemispheric mean trend and a more substantial change in the larger trend in the simple areally weighted grid box mean series. The grid box temperature series, principal components (PCs) series, and empirical orthogonal functions (EOFs) are correct as shown previously. [2] The corrected versions of Figures 6 and 7 are shown here. The primary conclusions stated by Mann et al. [2003] (hereinafter referred to as M03) stand with some minor revision: [3] 1. The ‘‘optimal’’ borehole hemispheric surface air temperature (SAT) trend is slightly greater in amplitude than shown by M03, with about 0.1 C greater cooling back in time. The optimal estimate is nonetheless only modestly colder than various other proxy-based SAT reconstructions and instrumental estimates in past centuries and is consistent with them well within the mutual uncertainty estimates (Figures 6 and 7). [4] 2. The areally weighted, gridded borehole data display a smaller trend (Figure 6) than the simple composite of Huang et al. [2000]. The trend is larger, however, than that shown originally by M03, with about 0.2 C greater cooling back in time. The trend is thus closer in amplitude to the Huang et al. [2000] estimate than shown previously. Gridding and areally weighting thus appear to be less of a factor in resolving differences between borehole and other proxy-based estimates than might have been inferred from M03.