南海北部集合卡曼滤波同化SST试验

SST assimilation experiment in the northern South China Sea using ensemble Kalman filter

基于POM(Princeton Ocean Model)建立一个南海北部集合卡曼滤波的同化模式,主要用于卫星海表面温度的同化。模式的平均水平分辨率为5km,垂向分层为20层;侧边界条件嵌套到一个大范围的南海海洋模式,在同化方案上采用一个均方根集合卡曼滤波算法,避免观测的扰动;适当引入局地化算子,消除样本在空间上的虚假相关,同时增加集合样本的自由度。该同化试验同化了2008年夏季6月到7月的GHRSST(Global High-Resolution Sea Surface Temperature),然后采用2008年夏季SCOPE(Northern South China Sea Coastal Oceanographic Process Ex-periment)航次的温、盐数据对同化结果进行评估。结果表明,相对于未同化模式模拟结果,同化模式温度的改善比较明显,表现在加强了南海北部的上升流,校正了海表温度的偏差,改善了温度的垂向分布。由于集合卡曼滤波是一种多变量调整的同化方法,同化SST不仅能改善表层与次表层的温度分布,而且对流场和盐度的调整也比较明显。

An ensemble Kalman filter（EnKF） scheme is applied to assimilate sea surface temperature（SST） in the northern South China Sea（SCS） using the Princeton Ocean Model（POM）.The assimilation model has a horizontal resolution of 5km and a vertical resolution of 20 layers.Lateral boundary conditions are provided by a larger domain SCS model.A square root filter is applied to avoid perturbations induced by observations.Localization is used in the assimilation system to remove pseudo correlations and to add rank of ensemble.The Global High-Resolution Sea Surface Temperature（GHRSST） in June and July 2008 is assimilated in this study.To validate the assimilation results,hydrographic data from the Northern South China Sea Coastal Oceanographic Process Experiment（SCOPE） cruises are used.The results show that the assimilated SST can effectively improve the temperature distribution not only at surface but also in the subsurface.After the SST assimilation,upwelling in this region is strengthened and mixed layer is deepened.At the same time,because the EnKF is a multivariable assimilation scheme,salinity and currents are also corrected by assimilating SST.