中文摘要：

为了研究GRAPES-Meso区域中尺度模式误差特点,评估模式动力框架和物理过程对预报误差的影响重要性,为GRAPES区域集合预报系统方案设计提供参考,基于GRAPES中尺度模式设计了4组对比试验,每组试验对2008年3个不同类型天气过程进行了数值模拟,获得如下结论：（1）GRAPES-Meso模式存在较为显著的系统性误差,系统性误差水平和垂直分布特征主要由GRAPES模式动力框架产生,物理过程对系统性误差影响相对较小;（2）在模式层底和模式层顶,GRAPES模式层与等压面层转换方案中,预报存在较为明显的垂直插值误差;（3）边界层方案对GRAPES模式低层动力场预报误差有重要影响,可以显著减少模式低层动力场预报误差。结果表明减少动力框架预报误差是改进GRAPES-Meso模式的重点,在GRAPES-Meso集合预报系统的设计中,需要重点考虑动力框架引起的模式不确定性。

英文摘要：

In order to study the characteristic of GRAPES-Mesa forecast error and assess the influence of dynamical core and physical parameterization on the GRAPES-Meso forecast error, also provide reference for designing GRAPES-Meso ensemble prediction system, four tests based on GRAPES-Meso model are implemented each with three different typical cases simulated. The result shows that： （1）there is significant systematic error in GRAPES- Mesa, with its horizontal and vertical distribution characteristic originated mainly from dynamical core and little from physical parameterization. （2）The vertical interpolation error could be introduced into forecast results, which is caused by the transformation scheme between model layer and pressure layer, and the top and bottom parts of model space are affected most; （3）Boundary parameterization can dramatically affect the forecasting result of dynamical quantity in lower level, as an significant error reduction of dynamical quantity in lower level. The conclusion is that much attention should be paid to the dynamical core for improving GRAPES-Mesa model. To construct the GRAPES-Meso ensemble prediction system, the model uncertainty caused by dynamical core should be considered carefully.