大气混沌特性和不完美模式使初始微小误差快速增长并最终限制可预报性，随系统时空尺度的减小，单一确定性预报的可预报性减小，因此将集合预报方法应用在风暴尺度系统上已成为国际研究热点。本项目针对风暴尺度集合预报扰动方法问题，采用数值模拟、诊断分析和理论研究相结合的方法，系统研究风暴尺度集合预报的初始条件、边界条件和物理过程扰动方法，针对风暴发展的敏感因子，研究自适应选择扰动变量的方法；发展基于集合资料同化和混合集合变换法的初始扰动方法，旨在构造在对流系统中能够快速增长且包含多尺度扰动信息的初始扰动结构，对比分析两种扰动方法的优劣，研究提出综合扰动方法；探讨对次网格独立物理过程进行倾向扰动的方法，构造新型随机物理扰动方案，使其能适用于边界层集合，并满足扰动前后模式物理方案能量守恒。研究结果不仅对风暴尺度集合预报扰动方法的认识有重要的科学意义，而且对风暴尺度集合预报系统的构建有实际应用价值。

It is well known that the chaotic nature in atmosphere dynamics and imperfect numerical models contribute intrinsically to the rapid growth of small errors in initial condition of any numerical weather prediction model as the forecast evolves.. As the scales of spatial and temporal become small, the predictability of a single deterministic system reduces. Therefore, ensemble forecast is considered as the effective tool to solve these problems by expressing uncertainties in initial condition, boundary condition and model physics. Recently, the application of ensemble forecast in storm-scale systems is increasingly focused on by more and more scientists in the world. In order to develop an initial perturbation scheme in the storm-scale ensemble system, the project will use the numerical simulation, analysis and diagnosis method to study the problem of boundary conditions, physical processes perturbation and sensitivity factor for storm-scale ensemble systems. In addition, the project will also develop the initial perturbation method based on data assimilation and mixed ensemble method of transformation, and generate the rapid-growth initial perturbations and physics perturbations in conservation of energy, depending upon the real-time developing storm systems. What’s more, the project will study the sub-grid independent perturbation method and build a new random physical perturbation scheme, which meets conservation of energy in model physics schemes. The above scheme can be used for the boundary ensemble system. In summary, the project has important scientific significance not only for initial perturbation scheme in the storm-scale ensemble system, but also for building the storm-scale ensemble systems.