随机物理集合预报研究进展

由于暴雨对流尺度集合预报模式中随机参数扰动能量偏小，本文利用WRFv3.9模式对我国长江流域一次特大暴雨天气过程开展对流尺度集合预报试验，对比分析随机参数扰动(SPP)方法和随机物理倾向扰动(SPPT)方法、随机动能补偿(SKEB)方法混合的扰动特征、评估多种混合扰动方案预报效果，主要结论如下：暴雨对流尺度集合预报中多随机物理扰动气象要素的离散度在单SPP方案基础上均增大，三种方案混合SPP+SPPT+SKEB的离散度最大，总体评分最优；其中SPP+SPPT试验较SPP+SKEB地面气象要素离散度更大，而SPP+SKEB试验较SPP+SPPT气象要素在高空(尤其是风场)离散度更大；SPP+SPPT试验在积分初期各气象要素离散度增加明显，SPP+SKEB试验随着积分时间的延长各气象要素离散度在各高度层增加更为凸显；多种随机扰动混合对降水的预报在集合平均、离散度分布和概率预报技巧等方面均优于单SPP方案，综合来看SPP+SPPT+SKEB方案总体评分最优。SPP+SPPT+SKEB综合了SPP、SPPT和SKEB的优势，在各高度层整个积分时间段各气象要素离散度达到最佳，多随机扰动方案的混合达到了有效互补的效果。

Review of the Ensemble Prediction Using Stochastic Physics

Because of the small disturbance energy of random parameter in the rainstorm convective-allowing ensemble forecast model, a convective-allowing ensemble forecasting test based on WRFv3.9 model was carried out for a heavy rainstorm case in the Yangtze River Basin in China. The disturbance characteristics of various hybrid perturbation schemes of stochastically perturbed parameterization (SPP), stochastically perturbed parameterization tendencies (SPPT), and stochastic kinetic-energy backscatter (SKEB) were compared and analyzed, and the prediction effects of these schemes were evaluated. The main conclusions are as follows: the spread of meteorological elements of multi-stochastic physical disturbance in rainstorm convective-allowing ensemble forecast increases comparing to the single SPP scheme. The spread of SPP+SPPT+SKEB is the largest and the overall score is the best. The spread of surface meteorological elements of SPP+SPPT is larger than that of SPP+SKEB, while the spread of meteorological elements in the upper air (especially in the wind field) of SPP+SKEB is larger than that of SPP+SPPT. In the SPP+SPPT test, the spread of meteorological elements increased significantly at the initial stage of integration time, while the spread of meteorological elements at each altitude increased more prominently with the extension of integration time in the SPP+SKEB test. In terms of ensemble average, spread distribution and probability prediction skills, the hybrid perturbation scheme of multi-stochastic disturbance are better than those of the single SPP scheme. Overall, the SPP+SPPT+SKEB scheme performs the best. The SPP+SPPT+SKEB scheme integrate the advantages of SPP, SPPT and SKEB, and achieve the best spread of meteorological elements in the whole integration period of each altitude layer, because of the complementary effects of the hybrid perturbation of multi-stochastic disturbance schemes.