基于伴随方法的2007-2011年北极海冰面积变化原因研究

Recent Arctic sea ice coverage is the lowest in the satellite record (since 1979) and is occurring at least 30 years earlier than was anticipated in the recent assessment report issued by the IPCC (Intergovernmental Panel on Climate Change).Focusing on the aim of finding reasons for the accelerated decline in the Arctic sea ice cover in recent years, the atmospheric forcing fields from three widely used reanalysis datasets (ERA-Interim,NCEP2 and JRA25) will be investigated and numerical sensitivity experiments will be performed with global sea ice-ocean coupled model and its adjoint code (the adjoint is an ideal tool to quantify the.sensitivity of a small set of target quantities with respect to a change in a larger number of independent variables). Runs of adjoint model for years from 2007 to 2011 respectively will compute the sensitivities of total Arctic sea ice area with respect to the following input quantities on the full model grid:the surface atmospheric forcing fields (wind in two directions at 10m, temperature and specific humidity at 2m, precipitation rate and downward radiation fluxes) as well as the initial values of the ocean state (ocean currents,ocean temperatures and ocean salinities) and the sea ice state (ice thickness, ice concentration and ice currents). Influences of bottom boundary condition (slip or non-slip) will also be studied. The features of sensitivities from results for years from 2007 to 2011(2007 is the year with the minimum Arctic sea ice cover and 2011 is the year with the second miminum sea ice cover) will be analyzed comparatively to study the variation of effectiveness of factors affecting sea ice variation and to deepen the understanding of reasons for the "accelerated decline in the Arctic sea ice cover" in the century. In addition, the differences between influences of atmospheric forcings and boundary conditions will be analyzed to deepen the understanding of uncertainties in numerical simulations of sea ice.

围绕寻找本世纪北极海冰面积加速减小原因目标，在对三套大气强迫场（ERA-Interim、NCEP2和JRA25）进行深入对比分析的基础上，利用其驱动全球海冰-海洋耦合模式及其伴随模式，开展北极海冰总面积对大气强迫（10米风、2米气温和比湿、降水率及向下辐射通量）、海洋（温度、盐度及流速）和海冰状态（厚度、密集度及流速）初值及底边界条件（有滑动或无滑动）敏感性数值试验，对比分析模拟结果中2007-2011年（2007年为近几十年北极海冰面积最小年,2011年为次最小年,2008年为第三最小年）敏感性差异特点，研究影响北极海冰变化因子作用程度的变化特点,为本世纪"北极海冰融化加速"问题增加认识。此外，分析大气强迫及边界条件影响的差异，加深对海冰数值模拟不确定性问题的认识。

海冰既是气候系统的组成成分，又可看作是气候变化的一个指示因子。在涉及海冰的气候研究中，单纯采用资料分析方法是不够的。利用数值试验手段，通过分别考虑大气热力和动力强迫变化对海冰变化的影响，可以加深对此问题的理解。但由于计算条件限制，考察的影响因子个数不能很多。而且，由于数值模式中包含复杂的非线性作用机制，在讨论模拟结果时往往也存在困难。与之相对比，伴随方法可以在一次积分中给出对上百甚至更多因子的一阶响应结果，且在线性近似框架下物理意义更容易解释。此外，采用伴随方法还可开展对海冰密集度等海冰变量进行优化工作。项目首先对造成海冰变化的大气强迫资料(三套再分析数据集)进行分析，然后开展控制试验及伴随模式发展工作，在此基础上开展理论研究。结果表明：NCEP2与JRA25地表气温相似性更好，不同数据集间辐射通量可有明显差异，大气辐射计算问题可使海冰模拟的不确定性增大；大西洋热输送敏感性试验结果与前人工作一致，验证了伴随模式的可靠性；利用伴随模式得到了北极不同海区海冰与初值、大气强迫及海洋模式参数间的关系，对北极海冰变化原因加深了认识；代价函数权重系数取法对海冰密集度分布形势优化结果影响不大；模式其它变量初值误差可使海冰密集度初值优化结果明显变差。这意味着同时对多个重要模式变量进行优化，优化效果可能会更好；缩小优化空间范围并没有改善海冰密集度初值优化结果。