Comparison of Holocene temperature reconstructions based on GISP2 multiple-gas-isotope measurements

Abstract

Nitrogen and argon stable isotope data obtained from ancient air in ice cores provide the opportunity to reconstruct past temperatures in Greenland. In this study, we use a recently developed fitting-algorithm based on a Monte Carlo inversion technique coupled with two firn densification and heat diffusion models to fit several Holocene gas-isotope data measured at the GISP2 ice core and infer temperature variations.We present for the first time the resulting temperature estimates when fitting δ15N, δ40Ar, and δ15Nexcess as individual targets. While the comparison between the reconstructions using δ15N and δ40Ar shows high agreement, the use of δ15Nexcess for temperature reconstruction is problematic because the statistical and systematic data uncertainty is higher and has a particular impact on multi-decadal to multi-centennial signals.Our analyses demonstrate that T(δ15N) provides the most robust estimate. The T(δ15N) estimate is in better agreement with Buizert et al. (2018) than with the temperature reconstruction of Kobashi et al. (2017). However, all three reconstruction strategies lead to different temperature realizations.