中国CMIP5模式对未来北极海冰的模拟偏差

本文对中国参加CMIP5的6个气候模式对未来北极海冰的模拟情况进行了评估。通过与1979-2005年海冰的观测值以及2050年代的多模式集合平均值对比发现，中国的气候模式对海冰范围的模拟结果与CMIP5模式的平均水平存在一定差距，具体表现为：BNU-ESM和FGOALS-s2对当前海冰范围估计很好，但对温度敏感性略偏高；FIO-ESM对当前海冰范围估计很好，但由于海冰对温度的敏感性偏低，导致其模拟的未来海冰在各种RCP情景中都融化缓慢；FGOALS-g2（BCC-CSM1-1和BCC-CSM1-1-m）对当前海冰范围的模拟存在显著偏多（显著偏少）的问题，这导致其对未来海冰融化的估计也持续偏多（偏少）。中国模式对北极海冰的模拟偏差导致它们对极区地表大气温度和湿度的模拟出现偏差，并且这些极区气象要素的偏差会进一步通过动力过程传导到对秋、冬季西风带、极涡的模拟中去。研究表明：从对海冰本身的模拟以及海冰偏差带来的气候影响这两个角度看，BNU-ESM在中国模式中水平较高，但总体上中国6个气候模式在海冰分量的模拟上仍与世界平均水平存在差距，这需要中国各模式中心的持续改进。

Performance of Chinese Climate Models in Simulating Arctic Sea-Ice in CMIP5 Experiments

The performance of six Chinese climate models in simulating Arctic sea-ice in CMIP5 are revisited to investigate their uncertainty in future climate projections. We re-estimate all CMIP5 models using two indicators and suggest six “good” (G6) out of 40 models in reproducing Arctic sea-ice: CESM1-BGC, HadGEM2-CC, IPSL-CM5A-LR, MPI-ESM-LR, MPI-ESM-MR, MRI-CGCM3. However, it is found that Chinese models cannot well reproduce observed sea-ice extent in the past decades (BCC-CSM1-1, BCC-CSM1-1-m, FGOALS-g2) and also fail to predict Arctic climate compared to the multi-model ensemble (MME) mean around 2050s. FGOALS-g2 (BCC-CSM1-1, BCC-CSM1-1-m) significantly overestimated (underestimated) the current extent of Arctic sea-ice and present a larger (smaller) sea-ice extent than MME mean in the future. FIO-ESM well estimated the current Arctic sea-ice extent but fail in predicting it in future, with a huge bias of the sensitivity to the warming surface. Relatively, BNU-ESM and FGOALS-s2 successfully reproduce the current Arctic sea-ice extent and predict an Arctic climate close to CMIP5 MME mean in 2050s. BNU-ESM, with the smallest bias in simulating Arctic sea-ice, is considered as the best one among the six Chinese models. Furthermore, we would expect a significant improvement of Chinese climate models in simulating Arctic sea-ice in the next phase of CMIP experiments, so that smaller biases in projecting not only local climate but also remote responses that close to CMIP MME mean projections could be expected.