EnKF集合同化下黄海海雾数值确定性预报初始场构造方法的探究

在黄海海雾的数值模拟中，EnKF（ensemble Kalman filter）是一种优于3DVAR（three-dimensional variational）的数据同化方法。研究发现，对EnKF初始场集合体采取常用的集合平均所产生的确定性预报初始场，会出现初始场中海雾在预报开始后就迅速消失以及接下来海雾难以生成的异常现象。通过详细的海雾个例研究，清晰地揭示并解释了此现象，指出这是集合平均造成初始场中云水与温度湿度之间存在不协调关系所导致的后果，并提出了一种择优加权平均方法来取代常用的集合平均。研究结果表明，海雾确定性预报采用择优加权平均所构建的初始场，可以消除上述异常现象，显著改进海雾模拟效果。

CONSTRUCTION OF INITIAL FIELD FOR NUMERICAL FORECAST OF THE YELLOW SEA FOG BASED ON ENKF DATA ASSIMILATION

In the numerical simulation of sea fog over the Yellow Sea, the EnKF (ensemble Kalman filter) is a data assimilation method superior to 3DVAR (three-dimensional variational). However, an abnormal phenomenon is that sea fog in the initial field disappears quickly after forecasting and it is difficult to generate subsequently when using common ensemble average method with which the initial field for deterministic forecast with EnKF data assimilation can be constructed. By a case study of sea fog, the phenomenon was clearly explained to be resulted from the inconsistent relationship among cloud water, temperature, and humidity in the initial field constructed by ensemble average, to which a new method was proposed using preferred-weighted-average to replace the ensemble average. It is shown that the deterministic forecast of sea fog base on the new method could eliminate the abnormal phenomena, and consequently improve the sea fog forecasting considerably.