多普勒雷达基数据在短时对流天气数值预报中的应用

Application of Doppler Raw Radar Data on Meso-scale Model for Convection Nowcasting and Short-Range Forecast

ARPS模式中原本有处理多普勒雷达Lev-Ⅲ资料的模块，但在一个雷达测站只处理4个仰角的径向风和反射率的资料。在此基础上，进行修改扩充将我国20世纪末、21世纪初全国布网的新一代多普勒雷达基数据Lev-Ⅱ全部14个仰角的原始径向风和反射率资料应用于ARPS模式。对我国2003年7月一次特大暴雨过程的个例研究表明，初始时刻加入多普勒雷达基数据的数值模拟，能改善短时0～12h的降水范围、强度和降水中心的预报结果。即使在目前雷达初始化数值模式预报低谷区的3～4h处，仍表现出较好的预报水平。因此，本文认为使用多普勒雷达资料初始化中尺度数值模式，能够弥补“spin-up（起转）”带来的副作用，表现在开始预报的第1h，就预报出雷达估计回波的强中心和大致分布，并且预报出与实况接近的降水；改变了由“spin-up”引起的回波估计预报延迟和预报前期没有降水的现象。且试验表明提高数值模式分辨率，可以使多普勒雷达资料弥补“spin-up”副作用的能力更明显。多普勒雷达资料应用于数值模式之所以有这种能力，是由于从多普勒雷达资料可以得到云和降水场的信息，一方面为模式微物理变量提供初值，填补了初始时刻云微物理量的空白，使随后的微物理量演变进程更加合理化，从而对降水预报产生正面影响；另一方面，为中尺度模式的非绝热初始化提供对流尺度数据源，对潜热、动力场和湿度进行调整，有效地实现模式的“hot—start（热启动）”，缩短了“预热”时间。

After some modification, the authors can apply China New Generation Doppler raw radar data which contains all 14 levels reflectivity and radial velocities to ARPS model. Using this method, the authors can study the efforts on weather prediction when applying raw radar data in ARPS model. The authors chose a case study, which is a China mesc-scale characteristics of the heavy rainfall happened on 4-5 July 2003 in the Huaihe Rivor valley. The results show that the 0-12 h precipitation forecast is greatly improved in precipitation range, intensity, the location of precipitation center and radar echo estimate; even the 3-4 h forecast which is the most difficult part in meso-scale NWP model nowadays also shows better. The forecast of radar reflectivity and rainfall in the beginning hour is similar to the observation, which proved that assimilating Doppler raw radar data in this case have alleviated the effort of ＂spin-up＂. The reasons are that the model can obtain the information of cloud and precipitation when applying raw radar data to the model. On one hand, this information can provide the initials to microphysical variable in the model, which can speed up microphysical processes and shorten the adjustment time of the model. On the other hand, the cloud and precipitation analysis can provide data source for moisture and diabatic initialization of convective scale numerical model, and then the model has the conditions for ＂hot start＂, which also can alleviate the effort of ＂spin-up＂.