本文评估了地球系统模式FIO-ESM(First Institute of Oceanography-Earth System Model)基于集合调整Kalman滤波同化实验对1992-2013年北极海冰的模拟能力。结果显示:尽管同化资料只包括了全球海表温度和全球海面高度异常两类数据,而并没有对海冰进行同化,但实验结果能很好地模拟出与观测相符的北极海冰基本态和长期变化趋势,卫星观测和FIO-ESM同化实验所得的北极海冰覆盖范围在1992-2013年间的线性变化趋势分别为-7.06*10~5和-6.44*10~5 km~2/(10a),同化所得的逐月海冰覆盖范围异常和卫星观测之间的相关系数为0.78。与FIO-ESM 参加CMIP5(Coupled Model Intercomparison Project Phase 5)实验结果相比,该同化结果所模拟的北极海冰覆盖范围的长期变化趋势和海冰密集度的空间变化趋势均与卫星观测更加吻合,这说明该同化可为利用 FIO-ESM开展北极短期气候预测提供较好的预测初始场。
英文摘要:
In this study,Arctic sea ice during 1992-2013simulated by FIO-ESM(First Institute of Oceanography-Earth System Model)based on ensemble adjustment Kalman filter data assimilation experiment is assessed.Al- though only global sea surface temperature and global sea level anomaly are assimilated to FIO-ESM and there is no sea ice assimilation,our study shows that the climatology and long-term trend of Arctic sea ice can also be well reproduced with this kind of data assimilation.The linear trends of Arctic sea ice extent during 1992-2013from satellite observations and FIO-ESM simulations are-7.06*10~5 and-6.44*10~5 km~2/(10a),respectively.The correlation coefficient between modeled and observed Arctic sea ice extent anomalies is 0.78.Compared with the results from FIO-ESM in CMIP5(Coupled Model Intercomparison Project Phase 5)experiment,the long-term trends of Arctic sea ice extent and sea ice concentration from data assimilation experiment fit the observations much better,so these results from FIO-ESM data assimilation experiment can be used as initial condition for Arctic climate projection.