Calibrating climate-δ18O regression models for the interpretation of high-resolution speleothem δ18O time series

Providing estimates of past climate changes on interannual–millenial timescales requires suitable regression models between climate and climate proxies. Many proxies appear to show relationships with climate that are timescale dependent. Any proxy‐climate model should be able to replicate the major patterns that are observed at multiple timescales. Here we develop a new climate‐isotope regression model for speleothems from a middle latitude site. In the low to middle latitudes, daily variation in precipitation isotopes (within individual months) is largely negatively correlated with daily rainfall amount. On interdecadal timescales, though, this relationship appears to be nonstationary. These two points provide a theoretical basis for a new climate‐isotope regression model in which the slope and the intercept of a δ18Oday–Pday line for a given month are modulated by organized patterns of climate variability, such as the extratropical zonal waves (including the annular modes). In constructing this new regression model, we show how daily precipitation–δ18O relationships can be estimated using only monthly δ18O data and daily rainfall amounts. The new regression model provides a consistent picture of 18O variability over a range of timescales, and this has not been the case with any previous climate‐isotope regression model.