通过与观测资料和20世纪历史气候模拟试验结果(NoINT)的对比分析,评估了BCC_CSM1.1年代际预测试验(INT)中对全球海表温度(SST)年代际变化的回报能力。分析结果显示:①INT试验模拟的全球平均SST增暖趋势比NoINT的更加接近观测;②其在热带大西洋、热带西太平洋和热带印度洋有较高的预测技巧;③对于太平洋年代际振荡2个关键区——北太平洋和热带中东太平洋,模式的预测技巧较低,且海洋初始化的作用也很小;④在热带南印度洋,INT的预测技巧普遍高于NoINT,在提前3~6年和4~7年时技巧最高。这些结论与基于其他模式得到的已有研究结果类似,但是BCC模式对北大西洋,特别是其副极地区域的预测技巧明显低于其他模式。BCC模式无法合理模拟出北大西洋SST与热盐环流间的交替变化规律,可能是其预测技巧偏低的原因。

This study assesses retrospective decadal prediction skill of Sea Surface Temperature (SST) variability in initialized climate prediction experiments (INT) with the Beijing Climate Center Climate System Model (BCC_CSM1.1). Ensemble forecasts were evaluated using observations, and compared to an ensemble of uninitialized simulations (NoINT). The results show as follows: ①The warming trend of global mean SST simulated by the INT runs is closer to the observation than that in the NoINT runs.②The INT runs show high SST prediction skills over broad regions of tropical Atlantic, western tropical Pacific and tropical Indian Oceans. ③ In the North Pacific and the east-central tropical Pacific Ocean, the prediction skills are very weak, and there are few improvements coming from the initialization in the INT runs. ④ In the southern Indian Ocean, the prediction skills of the INT runs are significantly larger than that of the NoINT runs, with the maximum skill at the 3~6 and 4~7 years lead time. The above-mentioned conclusions are similar to the results of other climate models. However, the prediction skill in the North Atlantic Ocean is much lower than that of other models, especially in the subpolar region. The low skills in the Atlantic Ocean may be attributed to the misrepresentation of the lead-lag relationship between the Atlantic meridional heat transport and the SST in the BCC_CSM1.1.