Résumé
Combined measurements of H (super 2) O (sub 18) and HDO isotopic ratio in precipitation, leading to second-order parameter D-excess, have provided additional constraints on the water cycle and past climates compared to the H (super 2) O (sub 18) isotopic ratio alone. More recently, measurements of H (super 2) O (sub 17) have led to the definition of another second-order parameter: O (sub 17) -excess. Recent studies suggest that O (sub 17) -excess may provide information on evaporative conditions at the source of moisture in high latitudes, and on convective processes in the tropics. However, the processes controlling the spatio-temporal distribution of O (sub 17) -excess are still far from being fully understood. Here we use the isotopic general circulation model LMDZ to better understand what controls D-excess and O (sub 17) -excess in precipitation. The simulation of D-excess and O (sub 17) -excess is evaluated against a set of measurements in meteoric water and water vapor and polar ice cores. A set of sensitivity tests and diagnostics are then used to quantify the relative effects of evaporative conditions (sea surface temperature SST and relative humidity RH), of precipitation re-evaporation and of super-saturation during condensation at low temperature. In the tropics, simulations suggest that convective processes, in particular rainfall re-evaporation, are important controls on D-excess and O (sub 17) -excess. In the subtropics and mid-latitudes, simulated D-excess and O (sub 17) -excess decrease with latitude, consistent with observations. The simulated decrease in D-excess is due the decrease in SST, and to a lesser extent to the increase in RH, with latitude. In contrast, the simulated decrease of O (sub 17) -excess is mainly due to distillation processes. In high latitudes, LMDZ simulates the right sign of the D-excess and O (sub 17) -excess seasonality. In Antarctica, the higher d-excess in winter is due to stronger distillation at colder temperature, while the lower O (sub 17) -excess in winter is due to stronger super-saturation at colder temperature. At paleo time scales, LMDZ captures the lower D-excess and O (sub 17) -excess recorded in polar ice cores during the last glacial climate. Half of the simulated D-excess shift is due to lower SST and higher RH at evaporation during glacial climate, while the other half of the D-excess shift and all of the O (sub 17) -excess shift are due to stronger super-saturation at colder temperature. Results in high latitudes are however very sensitive to the super-saturation parameterization. Therefore, distillation effects, evaporative conditions and super-saturation control D-excess and O (sub 17) -excess in different proportions. This suggests that these two parameters could be jointly used with H (super 2) O (sub 18) isotopic ratio to better constrain past climatic conditions, provided that the super-saturation parameterization could be better calibrated.