Estimating the accuracy of Russian paleotemperature reconstructions

A framework is developed for the estimation of errors in paleotemperature reconstructions, such as those compiled by Borzenkova (Borzenkova, 1992; Borzenkova et al., 1992). Three primary sources of error are considered: (1) methodological error—the difference between the temperature inferred (method dependent) from a sample and the actual temperature when the sample was formed, (2) time-dating error—the error in temperature caused by a sample formed at a time different than intended, and (3) spatial interpolation error—the error caused by interpolating from point data to create an average or map of paleotemperatures. Each of these error sources can be the primary source of uncertainty, depending on the epoch, location, and methods applied. Preliminary error estimates are given for Borzenkovaʹs Holocene Climatic Optimum reconstruction of the zonal-mean-annual paleotemperature distribution. For this reconstruction, methodological error is found to be the dominant source of errors at low latitudes, while spatial interpolation error is found to be the dominant source at high latitudes. This finding is due to higher expected spatial variability of temperature at high latitudes which results in an increase in spatial interpolation error with latitude relative to methodological error which is insensitive to latitude. At mid-latitudes, where the reconstruction contained a higher density of data, the temperature was found to be significantly warmer than present for the Holocene Climatic Optimum. For epochs earlier than the Quaternary, the epochs were defined by time slices that span as much as millions of years, which suggests that time-dating error could be the primary source of error for paleotemperature reconstructions which are intended to represent snapshots in time. Natural climate variability for the Cenozoic and Cretaceous over time scales of millions of years is, however, not well quantified; therefore, it remains ambiguous how to quantify errors in temperature caused by inaccurate time dating for pre-Quaternary epochs.