长江中下游地区暴雨数值预报试验

NUMERICAL PREDICTION EXPERIMENTS OF HEAVY RAIN IN THE MIDDLE AND LOWER OF CHANGJIANG RIVER VALLEY

本文利用 ＮＣＡＲ／ＰＳＵ联合研制的中尺度天气数值模式 ＭＭＳ，对长 江中下游地区在１９９６年６月下旬至７月中旬发生的暴雨过程，以国家气象中心业 务运行的全球谱模式Ｔ６３的分析场作为初值和侧边界条件，进行了一系列的数值 预报试验。结果表明，该模式较好地预报了与实况相近的降雨区的位置，但对暴雨 的中心落点和强度，其预报结果与实测有一定程度的偏差。如果模式提前１２小时 开始积分，则可在一定程度上克服模式中凝结降水的旋转加强问题，使预报的降水 强度等有明显的改善。

In this paper a series of experiments of the heavy rain processes occurring in the period from the third dekad in June to the second dekad in July of 1996 are made by using the NCAR/PSU mesoscale numerical weather model MM5. The analysis data of the global spectral model T63 at National Meteorological Center in China are used as initial and lateral boundary conditions necessary to drive the model runs. The results show that the model can predict the location of rainfall area, which is much close to the observation. However, there are some departures for location and density of heavy rain. If the model begins 12 hours earli- er,the spin-up problem of condensation rainfall in the model would been got over and pre- dicted rainfall density would become more reasonable.