| description abstract | The Melbourne University spectral atmospheric general circulation model is adapted to include prediction of stable water isotopes. The new scheme performs well when the modeled δ18O of precipitation is compared to both monthly observations from a global network and high-frequency measurements from two neighboring southern Australian sites. The associations between the modeled isotopic signal, temperature, and precipitation are examined on a variety of timescales by exploring the spatial distribution of temporal partial correlations. In contrast to the view commonly taken in paleoclimate studies, typically less than 20% of δ18O variance can be explained by temperature changes. The association with temperature is strongest when daily data are considered while the precipitation is more important on longer (interannual) timescales. This shows that as information about individual events is lost through the averaging process, simple distillation models, which have a strong theoretical temperature dependence, become less applicable. It is suggested that reconstruction of precipitation is more reliable on timescales longer than those considered, and the temperature dependence of precipitation facilitates an association between temperature and δ18O in proxy records. The small magnitudes of the correlation coefficients suggest that direct interpretation of proxy records such as temperature, or precipitation, should proceed under utmost scrutiny because reconstruction is far more complex than the simple problem of local regression. Specifically, should strong associations with temperature or precipitation exist, it is only partially due to the phenomenological covariance at the deposition site. As such, relationships used for paleoclimate reconstruction that incorporate information about the origin and condensation history of the moisture should be encouraged in place of overly simplistic relationships that involve just local conditions. | |
