大气环流模式（GCMs）是定量评估气候变化及其影响效应的有效工具，GCMs的优选是使用该工具时不可缺少的重要环节。将TOPSIS综合评价法引入到GCMs的优选中，并结合熵权的概念对各项评价指标进行赋权，采用加权的TOPSIS综合评价法对CMIP5中18个GCMs对黑河流域上游降水模拟方面的适用性进行评价，用传统的Rank Score（RS）方法进行验证；同时，基于GCMs对未来的降水预估进一步验证TOPSIS方法的可靠性；结果表明，CSIRO-Mk3.6.0模式对研究区降水的模拟效果最好，与两种检验方法的结果保持一致；由于TOPSIS综合评价法具有较严密的理论基础、计算简单、对样本大小没有要求、在计算过程中不存在信息损失等优势，因此在多方案、多目标的GCMs优选中具有广阔的应用前景。

The Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS) method is widely used in resolving the multi-objective and multi-indicator decision making problems in previous literatures. In this study, it was applied in evaluating the performance of 18 General Circulation Models (GCMs) in precipitation simulation over the upper reaches of Heihe River Basin. Six indicators, namely, mean value (Mean), normalized root mean square error (NRMSE), coefficient of variation (CV), ZC statistic value in Mann-Kendall trend analysis (ZC), coefficient of inhomogeneity (CL) and Pearson correlation coefficient (r), were constructed firstly in the evaluation, and then the weight of each indicator was determined by using entropy approach, and finally the 18 GCMs was ranked by the entropy weighted TOPSIS method. The result showed that CSIRO-Mk3.6.0 ranked the first and therefore was selected as the best alternative selection in precipitation simulation over the study area. The result was quite comparable with that from the traditional method of Rank Score. Compared with the observations during the period of 2006 to 2015, CSIRO-Mk3.6.0 model also performed the best in predicting the precipitation time series for the same period. Thus, it is concluded that the TOPSIS method can be applied in the similar studies widely since it has a rigorous theoretical basis, no requirement for data sample size, no loss of information in the calculation process, and is also easy to calculate.

中央高校基本科研业务费专项资金项目(No. 35832015028)