By utilizing 8 CMIP5 model outputs in historical experiment that simulated daily mean sea surface temperature (SST) and NCEP reanalysis data over 12 ocean basins around the world from 1960 to 2005, this paper evaluates the performance of CMIP5 models based on the detrended fluctuation analysis (DFA) method. The results of National Centers for Environmental Prediction (NCEP) data showed that the SST in most ocean basins of the world had long-range correlation (LRC) characteristics. The DFA values of the SST over ocean basins are large in the tropics and small in high latitudes. In spring and autumn, the zonal average DFA of SST are basically distributed symmetrically in the Northern and Southern Hemispheres. In summer, the zonal average values of DFA in the Northern Hemisphere are larger than those in the southern hemisphere, and vice versa in winter. The performance of HadGEM2-AO, CNRM-CM5 and IPSL-CM5A-MR are all relative well among the 8 models in simulating SST over most regions of the global ocean.

中文翻译：

---

CMIP5模型和NCEP再分析之间海洋全球地表气温的比较

通过在历史实验中利用8个CMIP5模型输出，模拟了1960年至2005年全球12个海盆的日平均海表温度（SST）和NCEP再分析数据，本文基于去趋势波动分析评估了CMIP5模型的性能（ DFA）方法。国家环境预测中心（NCEP）数据的结果表明，世界上大多数海盆的SST具有远距离相关（LRC）特征。海洋盆地海表温度的DFA值在热带地区较大，而在高纬度地区较小。在春季和秋季，南半球海温的纬向平均DFA基本对称地分布在北半球和南半球。夏季，北半球的DFA纬向平均值大于南半球，在冬天，反之亦然。在模拟全球海洋大部分区域的海表温度的8个模型中，HadGEM2-AO，CNRM-CM5和IPSL-CM5A-MR的性能都相对较好。