改进气候系统预测模式的北极海冰季节预测

In the context of rapid decline and thinning of Arctic sea ice in recent years, there is an increasing demand for seasonal sea ice prediction, especially from summer to fall. Using a climate system forecast model as the platform, in this project, on the one hand, we will assimilate the newly developed satellite-derived Arctic sea ice thickness using a recently developed local singular evolutive interpolated Kalman filter. This provides more accurate initial conditions for the climate system forecast model, leading to improved seasonal sea ice prediction. On the other hand, we will incorporate a melt pond parameterization and modify the parameterization of penetration and absorption of solar radiation in sea ice in the climate system forecast model, leading to improved the simulation of atmosphere-sea ice-ocean radiative interactions. Finally, we will conduct multi-ensemble hindcast and forecast experiments using the improved climate system forecast model, and examine impacts of the assimilation of satellite sea ice thickness and improved atmosphere-sea ice-ocean radiative interactions on seasonal sea ice prediction. This project will advance our research on seasonal predictability of Arctic sea ice, serving as an important step towards predicting Arctic sea ice change and its consequence on climate and environment in China using our climate system models.

近年来北极海冰的快速减少和变薄迫切需要准确的北极海冰季节预测，特别是夏季到秋季的预测。本项目将以一个气候系统预测模式为平台，一方面，利用一个新近发展的局部奇异值演化集合卡尔曼滤波技术，同化新近发展的基于卫星遥感的北极海冰厚度数据，为模式提供更准确，更合理的海冰初始条件。另一方面，在气候系统预测模式中加入海冰表面融池参数化和改进太阳辐射在海冰中的传输参数化，使模式能更好模拟大气-海冰-海洋辐射相互作用。开展多组北极海冰后报和预测集合试验，研究同化海冰厚度观测和改进大气-海冰-海洋辐射相互作用对海冰季节预报能力的影响。推进我国北极海冰季节可预测性研究，为我国气候系统模式预测北极海冰变化及其对全球及我国气候和环境的影响奠定科学基础。

本项目完成了在气候系统预测模式中耦合了一个新近发展的局部奇异值演化集合卡尔曼滤波技术（LESTKF）和一个自适应权重局地化方法，实现了对卫星遥感的北极海冰主要变量的同化，特别是同化了新近发展的基于卫星遥感的北极海冰厚度数据，包括：SMOS和CryoSat2，为气候系统预测模式提供更准确、更合理的海冰初始条件。在此基础上，开展了一系列北极海冰数据同化敏感性试验，分析了采用不同海冰变量生成初始集合扰动，同化不同海冰变量，以及海冰厚度观测误差，对北极海冰模拟和预测的影响。在气候模式中加入一个采用海冰融池分布函数方程来计算融池体积的参数化方案，改进了太阳辐射在海冰中传输参数化方案，使模式更好的模拟海冰融池分布和大气-海冰-海洋辐射相互作用。在此基础上，完成了一系列北极海冰回报和预测集合试验。数值试验结果表明同化北极海冰厚度观测和改进大气-海冰-海洋辐射相互作用可以显著地提高对模式对北极海冰季节预报的能力。研究成果发表在业界国际主流期刊上。