Journal of Arid Meteorology

Previous Articles     Next Articles

Simulation of Different Ensemble Forecast Schemes on a Large Area Heavy Rainfall by WRF Model


  1. 1. Unit of 63610 of the Chinese People’s Liberation Army, Korla 841001, China;
    2. Unit of 61243 of the Chinese People’s Liberation Army, Urumqi 830000, China
  • Online:2016-12-30 Published:2016-12-30



  1. 1.中国人民解放军63610部队,新疆 库尔勒 841001;
    2.中国人民解放军61243部队,新疆 乌鲁木齐 830000
  • 作者简介:袁有林(1987- ),男,甘肃平凉人,硕士,主要从事天气预报工作. E-mail:


In order to examine the effects of uncertainty in ensemble forecast, a typical large range rainstorm process occurred in Shaanxi, Shanxi, Hebei and Shandong Provinces from 12 to 13 July 2013 was simulated by WRF V3.6 meso-scale model from the disturbance of initial value, lateral boundary and physical process. The results are as follows: (1) The effects of uncertainty in WRF model on the simulation of the rainstorm were great. The perturbation of physical process had the greatest influence on the simulation of the rainstorm. The effect of initial value perturbation on the simulation result was obvious in the beginning of simulation, but gradually weakened in the later stage. However, the effect of lateral boundary uncertainty on the simulation result was small in the beginning of integral, subsequently became larger and larger with the transportation of perturbation to the simulation centre. (2) The ensemble forecasts of physical process scheme and initial value in WRF model were optimal to light rain and heavy rain and above, while that of lateral boundary scheme were optimal to moderate rain and rainstorm and above. (3) Comparing the dispersion of three kinds of ensemble forecast, we found that the ensemble forecast of physical process perturbations was the best, while that of initial value uncertainty was the worst. (4) The ensemble forecast considering three kinds of uncertainty was better than that of simple uncertainty, which significantly improved the forecast of the rainfall.

Key words: ensemble forecast, initial value, physical process, lateral boundary, WRF, rainstorm


应用WRF V3.6模式,对陕、晋、冀、鲁4省2013年7月12—13日的一次大范围暴雨过程,从初值、侧边界和物理过程扰动出发进行了集合预报研究。结果表明:(1)物理过程扰动对此次降水的影响最大,初值扰动在积分初期影响较大,而后逐渐减弱,而侧边界扰动随着时间积分向模拟区域中心传播并逐步增大;(2)物理过程扰动、初值扰动的集合预报分别对小雨和大雨及以上量级降水预报最优,而侧边界扰动的集合预报对中雨和暴雨及以上量级的降水预报最优;(3)从集合预报的离散度分析得出,物理过程扰动的集合预报最优,其次是侧边界扰动,初值扰动最差;(4)同时考虑3种不确定性的集合预报,总体上好于单个因子扰动的集合预报,使模式的降水预报效果得到显著改善。

关键词: 集合预报, 初值, 物理过程, 侧边界, WRF, 暴雨

CLC Number: