不同模式模拟的降水稳定同位素效应的比较和评估

COMPARISONS AND ASSESSMENT ON STABLE ISOTOPIC EFFECTS IN PRECIPITATION SIMULATED BY DIFFERENT MODELS

利用引入水稳定同位素循环的ECHAM4、GISS E、HadCM3、MUGCM以及iAWBM的模拟数据，分析了全球降水中稳定同位素效应的空间分布特征，对不同模式的模拟结果之间以及模拟结果与GNIP（Global Network of Isotopes in Precipitation）的实际监测结果之间进行了比较，旨在对稳定同位素大气环流模式的模拟有效性进行评价，改善对水循环中水稳定同位素效应的理解和认识。结果显示，5个模式均很好地再现了全球降水中平均δ18O和平均δ18O季节差的空间分布特征，降水中稳定同位素的温度效应、降水量效应的分布特点以及全球大气水线GMWL（Global Meteoric Water Line）均被很好地模拟出。比较而言，ECHAM4模拟的降水中的平均δ18O以及δ18O平均季节差的空间分布与GNIP的实际分布最接近，拟合水平也最高；ECHAM4、GISS E、MUGCM和iAWBM再现全球温度效应空间分布的能力较强，拟合水平大致相当；由iAWBM模拟的降水量效应空间分布与实际分布之间的相关性最强，5个模式模拟的与实测的δ18O/P相关系数符号相同的站点数大致位于同一水平；GISS E和iAWBM模拟的全球大气水线与实测的GMWL最接近。

The best way to reconstruct the spatial and temporal variations of stable water isotopes in the water cycle is to incorporate stable water isotope cycles into general circulation models.Only after modes are validated with global measurements of water isotopes could they be deployed with confidence to reproduce real physical phenomena,track physical processes and reveal physical mechanism in nature.Using simulation data from the isotope enabled ECHAM4,GISS E,HadCM3,MUGCM and iAWBM,the spatial variations of stable isotopic effects in precipitation are analyzed in order to assess the validity of stable isotopes in precipitation simulated by different atmospheric circulation models and improve the understanding of the stable water isotope cycle in the globe,through the comparison between simulations and GNIP observations and intercomparisons among simulations.Results show that the latitude effect,continent effect,altitude effect and seasonality of δ18O in global precipitation simulated by five models are in good agreement with GNIP observations,with their correlation coefficients between simulated and observed spatial series all exceeding the confidence level of 0.001.The distributional characteristics of temperature effect and amount effect,as well as global meteoric water line （GMWL） is well simulated.Comparatively,the spatial distribution of mean annual δ18O and δ18O seasonality in precipitation simulated by ECHAM4 are closer to GNIP observations,with two root mean square errors （RMSE） of 2.53‰ and 1.90‰,respectively,at the best fitting level.The ability of reproducing global spatial distribution of δ18O/T correlations is stronger in ECHAM4,GISS E,MUGCM and iAWBM,with the station numbers with same signs in simulated versus observed δ18O/T correlated coefficients achieving 87.94%,86.43%,86.18% and 83.67%,respectively,at roughly equivalent fitting level.There is the strongest correlation between iAWBM modeling and GNIP observed global δ18O/P distribution;The station numbers with same sign in simulated versus observed δ18O/P correlated coefficients are at roughly equivalent level for 5 models.The global meteoric water lines simulated by GISS E （δ2H=8.07δ18O＋11.33） and iAWBM （δ2H=8.07δ18O＋13.14） are closer to the observed one （δ2H=8.14δ18O＋10.96）.