青狮潭水库降水集合预报模式构建及应用

数值降水集合预报能够降低因单值预报带来的不确定性,是国内外关注的热点。但当前研究多将集合样本预报技巧等同对待,采用集合算数平均得到最终预报结论,这种集合预报方式较为粗放,难以有效区分不同集合样本的差异。针对该问题,本文选择位于广西暴雨中心的青狮潭水库为研究区,构建了基于ETS评分的综合定量评价模型,通过WRF模式21组参数化方案敏感性分析,以及样本数量与集合预报技巧的定量分析,依据ETS评价结果实现了青狮潭水库降水集合预报方案的构建。经验证,本文构建的集合预报方案较集合算数平均预报技巧更高,表现更加稳定。

Construction and application of precipitation ensemble forecast model for Qingshitan Reservoir

The numerical ensemble forecast of precipitation can reduce the uncertainty caused by single value forecasting, and it is a hot topic both at home and abroad. But most of the researches at present treat the forecast skills of set members equally, and obtain the final forecast conclusion with the average value of the collective arithmetic extensively, so this kind of method can hardly effectively show the differences among the forecast skills of different parameterization schemes. Aimed at this problem,the Qingshitan Reservoir located at the rainstorm center of Guangxi Zhuang Autonomous Region was selected in this paper as the research area and a comprehensive quantitative evaluation model based on the ETS grading was built, through sensitivity analysis on the WRF model and the 21 groups of parameterization schemes and the quantitative analysis for the number of samples and ensemble forecast skills,the establishment of ensemble forecast scheme of rainfall in Qingshitan Reservoir was realized based on the ETS evaluation results. Through verification, compared with the collective arithmetic average forecast, the ensemble forecast scheme constructed in this paper has higher skills and a more stable performance.