MASNUM-WAM海浪模式集合Kalman滤波同化研究——Ⅱ.集合样本对同化效果的影响

ON EAKF DATA ASSIMILATION BASED ON MASNUM-WAM ——Ⅱ. ASSIMILATION EXPERIMENT AND RESULT

背景误差相关结构的确定是影响海浪同化效果的关键因素之一。集合Kalman滤波是一种较为成熟的同化方法,其可以对背景误差进行实时更新和动态估计,现已广泛应用于海洋和大气领域的研究。本文基于MASNUM-WAM海浪模式,分别采用静态样本集合Kalman滤波和EAKF方法,针对2014年全球海域开展海浪数据同化实验,同化资料为Jason-2卫星高度计数据,利用Saral卫星高度计资料对同化实验结果进行检验。结果表明,两组同化方案均有效提高了海浪模式的模拟水平,EAKF方案在风场变化较大的西风带区域表现显著优于静态样本集合Kalman滤波方案,但总体上两者相差不大。综合考虑计算成本和同化效果,静态样本集合Kalman滤波方案更适用于海浪业务化预报。

We applied two wave data assimilation schemes：static sample ensemble Kalman filter and EAKF （Ensemble Adjustment Kalman Filter）, to assimilate the altimeter data of the Jason-2 into a global wave model MASNUM-WAM （marine science and numerical modeling-wave modelling part） over the period of 2014. A practical assimilation design was proposed for numerical implement. Results were validated against the altimeter data of the Saral satellite. In the first scheme, static sample ensemble that consists of the difference in 24h-interval SWH from long-term history model results are superposed to SWH field at time window for assimilation to construct a model state variable ensemble that will be updated by two-part filter method. In the second scheme, wave model ensemble is driven by wind field with random field perturbation. The results show that these two assimilation schemes could improve the ability of numerical simulation significantly compared with the control run without assimilation. In addition, EnKF （Ensemble Kalman Filter） scheme has a remarkable advantage over static sample ensemble Kalman filter scheme in mid-high latitudes where wind field varies rapidly. Overall, the results of two assimilation schemes are similar. However, the static sample ensemble Kalman filter could be applied to operational wave forecast at a lower computation cost.