大气环流模式（SAMIL）海气耦合前后性能的比较

基于耦合器框架，中国科学院大气物理研究所大气科学和地球流体力学数值模拟国家重点实验室大气环流谱模式 （SAMIL）最近成功地实现了与海洋、海冰等气候分量模式的耦合，形成了“非通量调整”的海－陆－气－冰直接耦合的气候模式系统（FGOALS-s）。在耦合系统中，由于海温、海冰等的分布由预报模式驱动，大气与海洋、海冰之间引入了相互作用过程，这样大气环流的模拟特征与耦合前会有不同。为分析耦合系统的性能，作者对耦合前后的模拟结果进行了分析比较，重点是大气模拟特征的差异。结果表明，耦合前、后大气环流的基本特征相似，都能成功地模拟出主要的环流系统分布及季节变化，但是由于海温和海冰的模拟存在系统性的偏差，使得耦合后的大气环流受到明显影响。例如耦合后热带海温偏冷，南大洋、北太平洋和北大西洋等中纬度地区的海温偏高，导致海温等值线向高纬海域的伸展较弱，海温经向梯度减小。耦合后海冰在北极区域范围偏大，在南极周边地区则偏小。海温、海冰分布模拟的偏差影响到中、高纬低层大气的温度。热带海温偏低，使得赤道地区降水偏弱，凝结潜热减少，热带对流层中高层温度比耦合前要低，大气温度的经向梯度减小。经向温度梯度的改变，直接影响到对平均经圈环流及西风急流强度的模拟。尽管耦合系统中海温、海冰的模拟存在偏差，但在亚洲季风区，耦合后季风环流及降水等的分布都比耦合前单独大气模式的结果合理，表明通过海CD*2]气相互作用可减少耦合前季风区的模拟误差，改善季风模拟效果。比较发现，海温、海冰模拟的偏差，除与海洋模式中经向热输送偏弱、海冰模式中海冰处理等有关外，也与大气模式中总云量模拟偏低有关。大气模式本身的误差，特别是云、辐射过程带来的误差，对耦合结果具有极为重要的影响。完全耦合后，这些误差通过与海洋、海冰的反馈作用而放大。因此，对于FGOALS-s而言，要提高耦合系统的整体性能，除改进各气候分量模式的模拟性能外，需要重点改进大气模式中的云、辐射过程。

A Comparison of the Atmospheric Circulations Simulated by the FGOALS-s and SAMIL

The spectral atmosphere model (SAMIL) developed at the State Key Laboratory of Numerical Modeling for Atmospheric Sciences and Geophysical Fluid Dynamics, Institute of Atmospheric Physics, Chinese Academy of Sciences (LASG/IAP/CAS), has been coupled successfully with other climate components such as ocean and sea ice models recently through the coupler, upon which the Flexible non-flux-correction Global Ocean-Atmosphere-Land-Ice climate model system (FGOALS-s) has been built. Since the sea surface temperature (SST) and sea ice distribution are predicted by the models, and the interactions between the atmosphere with the ocean and sea ice are introduced in this coupled system, the simulated atmosphere circulation features may be different from those in the uncoupled system. To understand the performance of the coupled system, the simulated results, especially for the atmosphere circulation differences between the coupled and uncoupled systems, are compared in this paper. The results reveal that the mean atmospheric circulation features as well as the seasonal variations are very similar, which imply that the simulated SST and sea ice distributions are in agreement with the climatic ones. However there exist some biases in the SST and sea ice simulation, which influence the atmosphere circulations obviously. For example the tropical SST is colder after coupling, and SST in the middle latitudes in the Southern Ocean, the North Pacific Ocean and the North Atlantic Ocean are warmer, which cause the weaker SST extension to the high latitudes and weaker SST meridional gradient. The sea ice coverage around the north pole is wider while that around the South Pole is narrower after coupling. The biases of SST and sea ice influence the lower atmosphere temperature in the middle and high latitudes. Due to the colder SST in the tropics, the tropical precipitation along with the condensation heat in FGOALS-s is reduced significantly, which causes the colder atmosphere temperature in the middle and upper troposphere and reduces the meridional temperature gradient compared with SAMIL. The change of meridional temperature gradient influences the strength of the mean meridional circulation and the westerly jets directly. Though there exist the aforementioned biases in SST and sea ice simulation, the circulation and precipitation are more reasonable than those before coupling in the Asian monsoon area, which indicates that the air-sea interactions may reduce the simulation errors and play important roles in the mean monsoon circulation simulation. It is also found from the comparison that the biases of SST and sea ice are not only related to the weaker meridional heat flux transport in the ocean model and special processes in the sea ice model, but also related to the underestimated total cloud amount simulation in the atmosphere model. The biases induced by the atmosphere model, especially related to the cloud and radiation processes, have great influence on the coupling performance. These biases can be amplified through the interactions with the ocean and sea ice after coupling. From this point, more attentions should be paid to the cloud and radiation processes in the atmosphere model, while every component is to be updated to improve the whole capabilities of the coupling system in the future.