中文摘要：

利用WRFV2.2模式,对1977年5月20日发生在美国俄克拉荷马州的典型超级单体风暴进行风暴尺度集合预报试验。通过对比由Monte-Carlo法和增长模繁殖法构造的集合预报系统的性能,检验了增长模繁殖法应用于风暴尺度集合预报的合理性及其价值。研究表明,在预报误差的意义上,集合技术能够明显改善中尺度强对流天气的单一确定性预报;但纯统计意义的Monte-Carlo法动力意义不足,导致集合离散度偏低;作为改进方法,增长模繁殖法充分考虑了预报中误差增长的信息,改善了离散度增长偏慢的缺陷。研究结果还显示,增长模繁殖法构造的集合预报系统的各预报评分（RSS、ETS等）均高于Monte-Carlo法,体现了增长模繁殖法的显著优势,但不同的繁殖周期对非降水场和降水场的预报能力有所差异,合理的繁殖周期需要在两者之间寻求平衡点。

英文摘要：

It is an important question that how to get the growing errors which is comparable to the actual error growth.The breeding method,which has been used to generate perturbations for ensemble forecasting at NCEP,simulates the development of growing errors in the analysis cycle,so it is a reasonable choice through the breeding methods capturing the growing errors modes.An ideal supercell storm case is simulated by WRF model,occurred in central Oklahoma on 20 May 1977 to study the merit of the ensemble system by the breeding methods.Meanwhile,we compare the forecasting skills of the ensemble system by different pertubation methods like Monte-Carlo and breeding of growing modes.The results show that the ensemble average forecast based on Monte-Carlo is excel to the control forecast,but the advantage isn′t remarkable.The potential reason is that the method can′t include the dynamic process leading to the smaller spread.Breeding gets the growing errors comparable to the actual error growth and the more appropriate ensemble spread than the randow perturbation′s.Considering the different indexes,the forecasting skill of ensemble system by breeding of growing modes is more effective than randow method′s.