用数字滤波方法进行数值模式的初始化

INITIALIZATION OF A NWP MODEL USING DIGITAL FILTERING METHOD

将数字滤波原理用于Ｔ４２Ｌ９全球谱模式的初始化。选取两个初值进行了实验。通过对模式短时间积分得到的时间序列进行滤波处理，有效地滤去了初始场中的高频重力波振荡，保留了其中有天气意义的扰动，同时对分析场的改变很小。５ｄ预报的结果说明，经过初始化后的预报较为平稳，场较为光滑。对比实验表明，绝热和非绝热初始化的效果很接近。这种方法简便有效。

A digital filtering method is used in the initialization of T42L9 global spectral model. The experiments have been done on two initial conditions. Through filtering processes on the time series taken from short time integrations of the model, the high frequency gravity wave oscillations have been effectively removed from the initial fields and at the same time, the disturbances which are synoptically significant are reserved. Moreover, the analysed initial fields are little changed by the initialization. The results of five day forecasts show that, after this initialization, the forecasts are more stable and the forecast fields are smoother. Comparative experiments show that the effectiveness of adiabatic and diabatic initializations are quite similar. This initialization method is not only efficient but also very simple and convenient to implement, so it is a very practical method.