贵州铜仁地区汛期暴雨预报研究  

Research on Rainstorm Forecast in Flood Season in Tongren Area of Guizhou Province

作　　者：向楠 冉光镜 吕春艳 蔡成瑶 冉红玲 杨群 

机构地区：[1]铜仁市气象局,贵州 铜仁 [2]贵州省山地环境气候研究所,贵州 贵阳

出　　处：《气候变化研究快报》2024年第4期857-864,共8页Climate Change Research Letters

摘　　要：为逐步实现定量降水预报,提升本地暴雨预报准确率,利用2014~2022年铜仁市5~9月历史观测降水量数据、ERA5逐小时再分析数据,结合实况物理量场资料,从动力、水汽、热力条件中选取多个物理量,基于阈值条件和多元线性回归模型,分析了铜仁市汛期暴雨的变化特征以及建立汛期分月暴雨预报模型。结果表明:(1) 铜仁市汛期暴雨日数年变化呈现一个明显的起伏波动变化,暴雨多发生在刚入夏时期,集中在月中至下旬,在7月更容易出现极端性降水,且暴雨物理量阈值存在显著的季节变化特征。(2) 物理量模型表明5、7、8月选取的物理量因子均为水汽条件多于动力条件,且在5月选取的动力因子均在中低层,在7、8月高层水汽条件也较为重要。当形势多为稳定性降水时,准确率较高。(3) 对2023年汛期28次暴雨个例进行检验,结果表明:动力条件中的散度、垂直速度较为稳定,其中80%个例满足阈值条件,7月对其反应效果最好。模型对9月预报效果较好,75%个例预测雨量误差在20 mm内。该暴雨预报方法有效提升对汛期暴雨过程的最大降雨量进行预估,在本地化应用中有一定指示效果。In order to gradually realize quantitative precipitation forecast and improve the accuracy of local rainstorm forecast, the historical observed precipitation data of Tongren City from May to September from 2014 to 2022, the hourly re-analysis data of ERA5, and the actual physical quantity field data were combined to select a number of physical quantities from dynamic, water vapor and thermal conditions. The variation characteristics of heavy rain in Tongren during flood season and the model of monthly heavy rain forecast in flood season are analyzed. The results indicate that: (1) The annual variation of rainstorm days in flood season in Tongren City presents an obvious fluctuation change. Rainstorm mostly occurs in the early summer, concentrated in the middle to late ten days, and extreme precipitation is more likely to occur in July, and rainstorm physical quantity threshold has significant seasonal variation characteristics. (2) The physical quantity model shows that the physical quantity factors selected in May, July, and August are more water vapor conditions than dynamic conditions, and the dynamic factors selected in May are all in the middle and lower layers, and the water vapor conditions in the upper layers are also more important in July and August. When the situation is mostly stable precipitation, the accuracy is higher. (3) The test of 28 rainstorm cases in flood season 2023 shows that divergence and vertical velocity in dynamic conditions are relatively stable, 80% of which meet the threshold conditions, and the response effect is the best in July. The model performs well in September forecasting, with an error of within 20 mm in predicting rainfall for 75% of the cases. The rainstorm forecasting method effectively improves the prediction of the maximum rainfall in the flood season rainstorm process, and has certain indication effect in the local application.

关 键 词：暴雨模型 暴雨阈值 多元线性回归 预报指标 

分 类 号：P45[天文地球—大气科学及气象学]