海冰资料同化中集合卡尔曼滤波的非高斯性问题研究

极地海冰是地球系统的重要组成部分，但现有海冰数据的可靠性以及对海冰的预测能力，尚不能完全满足科学研究和社会服务的需求。资料同化能够基于观测和模式信息给出更准确的状态估计，但目前应用于海冰资料同化的方法主要是基于高斯假设，而主要的海冰状态参数具有明显的非高斯特征，这使得海冰同化无法得到最优的海冰估计，甚至会产生没有物理意义的分析值。因此，本项目针对海冰密集度观测误差的非高斯性研究较为缺乏、海冰厚度同化的非高斯性研究尚属空白的现状，以南极海冰同化为切入点，围绕海冰资料同化中集合卡尔曼滤波的非高斯性问题开展研究。主要分为两部分：1、针对海冰密集度和海冰厚度，发展适用于南极海冰同化的高斯变换方法；2、基于上述高斯变换方法，探究南极海冰同化对南极海冰-海洋模拟的影响及其物理机制。本项目成果将被推广至北极海冰研究，期望能为极地海冰预测和资料同化研究提供有效经验和基础性工具，从而提升极地环境保障能力。

Polar sea ice is an important component of the earth system. However, the reliability of sea ice data and the sea ice prediction cannot fully meet the requirements of scientific research and social service. Taking information from observation and numerical model into consideration, data assimilation can provide more accurate state estimation. Unfortunately, most of data assimilation techniques for sea ice applications are based on Gaussian hypotheses so far, while main sea ice state parameters are of strong non-Gaussian characteristics, which hinders the sea ice data assimilation from the optimal estimation of sea ice, and even leads it to unphysical values. Owing to lack of studies on the non-Gaussianity of sea ice concentration (SIC) observation error as well as research gaps in the non-Gaussianity of sea ice thickness (SIT) data assimilation, this project starts from Antarctic sea ice data assimilation, and aims at the non-Gaussianity of ensemble Kalman filter in sea ice data assimilation. This project will develop the Gaussian anamorphosis for assimilating SIC and SIT in the Antarctic. Then based on the above Gaussian anamorphosis, the impact of sea ice data assimilation on Antarctic sea ice-ocean coupled modelling will be investigated. The expected outcome of the project can be extended to the Arctic sea ice research. These results will help to carry out the sea ice prediction, and to develop the method for sea ice data assimilation, which can improve the support capacity for polar environment security.