David V. Bekaert, Alizé Longeau, Camille Risi, Gabriel Fénisse, Charlotte Prud’Homme, Roi Ram, Jian Zhang, Dominique Genty, Alan M. Seltzer, Werner Aeschbach, Thomas Müller, Roberto Kirchheim, Sylvia Dee, Kanon Kino, Kei Yoshimura, Alexandre Cauquoin, Hayoung Bong, Allegra N. LeGrande, Pierre-Henri Blard
Advances in Geochemistry and Cosmochemistry 2(2): 1097
Voir en ligne : https://doi.org/10.33063/agc.v2i2.1097
Water stable isotopes (δ2H and δ18O) are widely used to reconstruct past hydrological changes and paleoclimate variability. A key limitation, however, is that fossil isotope records reflect not only local temperature but also changes in circulation, moisture sources, and hydrological processes along air-mass trajectories. Although modern monthly and interannual δ2H–δ18O–temperature relationships have been proposed to calibrate paleoclimate reconstructions, their applicability to longer timescales, such as glacial–interglacial transitions, remains uncertain. Here, we use fossil groundwaters worldwide as direct archives of past annual temperature and δ2H–δ18O co-variations over the last ∼35,000 years. Our global compilation of water stable isotope and noble gas temperature records (746 samples from 39 locations) shows that, while significant relationships occur in several regions, European fossil groundwaters exhibit the most consistent δ18O–temperature correlations. In Western Europe, δ18O variations during deglaciation were primarily temperature-driven, allowing δ18O to be used as a paleothermometer on glacial–interglacial timescales. In contrast, weaker correlations in North America, Asia, and Africa likely reflect stronger influences from atmospheric circulation, convection, and/or continental moisture recycling. For Western Europe, the observed relationships are consistent with a simple Rayleigh distillation model, although sites further inland seem to reflect more complex atmospheric processes. The mean European fossil groundwater-based δ18O–temperature slope, corrected for ice-volume effects since the LGM, is 0.48 ± 0.21 ‰ °C−1 (1 SD), comparable to modern monthly relationships for precipitation (0.29 ± 0.07 ‰ °C−1, 1 SD) but lower than modern interannual estimates ( 1.04 ± 0.30 ‰ °C−1, 1 SD). These results provide robust constraints for interpreting isotope-based climate archives and evaluating isotope-enabled climate models.



