GCM simulations of stable isotopes in the water cycle in comparison with GNIP observations over East Asia

In this paper, we examine the performance of four isotope incorporated GCMs, i.e., ECHAM4 (Univer- sity of Hamburg), HadCM3 (Hadley Centre), GISS E (Goddard Institute of Space Sciences), and MUGCM (Melbourne University), by comparing the model results with GNIP (Global Network of Isotopes in Precip- itation) observations. The spatial distributions of mean annual δD and mean annual deuterium excess d in precipitation, and the relationship between δ ¹⁸ O and δD in precipitation, are compared between GCMs and GNIP data over East Asia. Overall, the four GCMs reproduce major characteristics of δD in precipitation as observed by GNIP. Among the four models, the results of ECHAM4 and GISS E are more consistent with GNIP observed precipitation δD distribution. The simulated d distributions are less consistent with the GNIP results. This may indicate that kinetic fractionation processes are not appropriately represented in the isotopic schemes of GCMs. The GCM modeled MWL (meteoric water line) slopes are close to the GNIP derived MWL, but the simulated MWL intercepts are significantly overestimated. This supports that the four isotope incorporated GCMs may not represent the kinetic fractionation processes well. In term of LMWLs (local meteoric water lines), the simulated LMWL slopes are similar to those from GNIP observa- tions, but slightly overestimated for most locations. Overall, ECHAM4 has better capability in simulating MWL and LMWLs, followed by GISS E. Some isotopic functions (especially those related to kinetic frac- tionation) and their parameterizations in GCMs may have caused the discrepancy between the simulated and GNIP observed results. Future work is recommended to improve isotopic function parameterization on the basis of the high-resolution isotope observations.