Oxygen isotopic signatures of major climate modes and implications for their detectability in speleothems

Abstract:

Major modes of climate variability, such as the El Niño/Southern Oscillation (ENSO), Pacific Decadal Oscillation (PDO) and Atlantic Multi-decadal Oscillation (AMO), strongly modulate terrestrial hydroclimate on interannual to multi-decadal time scales. Therefore it is important to understand the response of these modes to external forcing. Since general circulation models show large diversity in modal responses to climate change, paleoclimate reconstruction is highly useful for extending the temporal coverage of observations of climate modes, providing a longer baseline for evaluating model skill and thereby constraining model projections. In particular, the oxygen isotopic ratio in absolutely-dated speleothems (δ¹⁸O_spel) is an excellent tool for deciphering past climates due to its relatively long timespan, high temporal resolution, and availability across the world. Here we assess the usefulness of δ¹⁸O_spel for understanding past modal variance changes. We use the isotope-enabled Community Earth System Model’s Last Millennium Ensemble (iLME) combined with a forward proxy model to delineate the global expression of modal variability in “pseudo-stalagmite” (δ¹⁸O_spel) records worldwide. The modelled δ¹⁸O_spel shows coherent patterns, illustrating the spatial structure of modal signatures. Individual pseudo-stalagmites record changes in ENSO variance; however, substantial changes (>30 %) in modal variance are required for a modest response in δ¹⁸O_spel variance (<15%), indicating that any single speleothem has limited reconstruction skill. In contrast, using networks of δ¹⁸O_spel records from ENSO-teleconnected regions shows much greater skill in reconstructing ENSO variance, and the skill improves when the network is optimised with equal numbers of records from both sides of the Pacific. In the case of AMO and PDO variance reconstruction, the skill of modeled δ¹⁸O_spel is more limited. This can be either due to the weak relation between decadal modes and δ¹⁸O_spel in the real world, or to a potential underestimation of low frequency variability in iLME simulations.