Comparison of paleotemperature reconstructions as evidence for the paleo-analog hypothesis

There is an apparent inconsistency between the estimated low accuracy reported by Kheshgi and Lapenis (1996; hereinafter KL) for the reconstructed zonal-mean annual paleotemperatures for the mid-Holocene, and the high agreement reported by Shabalova and Können (1995; hereinafter SK) between the normalized temperature anomalies of these reconstructed paleotemperatures and those reconstructed for earlier epochs. The reasons for this inconsistency could be: (i) overestimation by KL of the reconstruction errors by more than a factor of two, (ii) significant smoothing of the paleodata resulting in a reduction in the number of independent pieces of information represented by the zonal-mean temperatures, or (iii) bias of the paleotemperature reconstruction by prior knowledge of the expected patterns of climate change. Because it is unlikely that the errors involved in producing the reconstructed mid-Holocene temperatures have been overestimated by more than a factor of two, one or both of the other reasons is the likely explanation for the inconsistency. If this holds true, then support for the paleo-analog hypothesis provided by the mid-Holocene paleotemperature reconstruction is severely weakened.