The Antarctic sea ice reconstruction (CMST-South) based on the optimized Data Assimilation System for the Southern Ocean

The wealth of historical sea ice concentration (SIC) observations, coupled with their extensive spatial coverage, renders them indispensable for the reconstruction of long-term Antarctic sea ice variability. However, recent studies have pointed out the presence of significant uncertainties in certain aspects of Antarctic sea ice reanalyses obtained from assimilating SIC. Notably, while previous studies on ocean data assimilation have already demonstrated the significance of optimizing model-dependent parameters for assimilating oceanic observations, this aspect has received limited attention in current sea ice data assimilation studies. As a result, whether optimizing model-dependent parameters can enhance the effectiveness of assimilating SIC remains an open question. Thus, we address this gap by refining the model-dependent parameters of Data Assimilation System for the Southern Ocean (DASSO), including the development of a latitude-dependent localization scheme and the objective estimation of observation error variance of SIC which takes into account both measurement errors and representation errors. Here, the monthly anomalies in Antarctic sea ice extent and volume (1980 -2018) are uploaded which is produced by the optimized Data Assimilation System for the Southern Ocean (DASSO) with assimilating SIC. Besides, a 13-month moving mean is applied to monthly anomalies to focus on the low-frequency variability of Antarctic sea ice.海量历史海冰密集度（sea ice concentration, SIC）观测资料凭借其广泛的空间覆盖范围，成为重建长期南极海冰变率的不可或缺的核心数据支撑。然而，近期研究指出，通过同化海冰密集度（SIC）得到的南极海冰再分析产品在若干维度存在显著不确定性。值得注意的是，尽管此前海洋数据同化领域的研究已证实，优化模式依赖参数对提升海洋观测资料同化效果的重要性，但当前海冰数据同化研究对该方向的关注仍较为不足。因此，优化模式依赖参数能否提升海冰密集度（SIC）同化效果，仍是一个尚未解决的科学问题。为此，本研究通过优化南大洋数据同化系统（Data Assimilation System for the Southern Ocean, DASSO）的模式依赖参数填补这一研究空白，具体工作包括构建纬度依赖局域化方案，以及同时考虑测量误差与表征误差的海冰密集度（SIC）观测误差方差客观估计方法。本数据集包含经优化后的南大洋数据同化系统（DASSO）同化海冰密集度（SIC）所生成的1980至2018年南极海冰范围与体积月距平序列；此外，为聚焦南极海冰的低频变率，我们对月距平序列进行了13个月滑动平均处理。