| 摘要: |
| 【目的】贵州地形复杂,降水预报难度大,针对高分辨率区域模式预报产品的综合检验评估是提高模式预报性能的关键环节之一。【方法】基于2023年汛期(4月15 日-10月15 日)贵州省1560 个加密观测站的逐小时降水数据,从TS评分、,降水频率,强度、降水峰值时段等方面,对 CMA-MES0,CMA-CD,CMA-SH 等3 个高分辨率区域模式在贵州的预报能力进行详细评估。【结果】3 个模式对区域平均日降水量,降水频率的预报均偏多:3 个模式基本能预报出平均日降水量的大值中心,但位置相对观测偏东偏北。CMA-GD 中到暴雨 TS评分高:CMA-MESO 各降水量级下TS 评分均为最低。基于 60 个强降水日的逐6h累计降水预报评估表明各模式降水量和频率预报偏差出现时段有所不同。CMA-MESO 降水量负偏差多集中在下午时段,CMA-GD 和 CMA-SH 上午正偏差最为严重。观测中降水频率高值多出现在贵州东北部(白天)和西部(夜间)CMA-CD 对强降水日降水频率时空演变预报较为准确,其余模式偏差较大。模式对强降水日贵州东北部夜间降水峰值出现时间预报较为准确,而在贵州东南部3个模式预报峰值时间基本为夜间,与观测中白天峰值不同。结论】综合来看,CMA-GD对贵州汛期平均日降水量的TS评分和降水日内演变特征预报最优。此外,基于降水量、次、强度,目变化等特征的综合评估方法有助于正确认识模式在复杂地形区的降水预报偏差。 |
| 关键词: 高分辨率模式；汛期；检验评估；降水 |
| DOI： |
| 投稿时间：2024-04-03修订日期：2024-09-20 |
| 基金项目:中国气象局复盘总结专项（FPZJ2023-117）：高分辨率区域模式对贵州强降水的预报检验 |
|
| Evaluation of Three High-resolution Operational Weather Forecast Models for Precipitation in Guizhou Province During the Rainy Season 2023 |
| HeMu,YANG Fuyan,Tang Weiyao,Zhang Kun |
| (Guizhou Institute of Mountainous Environment and Climate;State Key Laboratory ofEnvironmental Geochemistry, Institute of Geochemistry, Chinese Academy of Sciences) |
| Abstract: |
| Due to the complex terrain, precipitation in Guizhou is diflicult to forecast accurately. Acomprehensive evaluation of the forecast products of high-resolution regional models is one of the keys to improvethe model forecast performance. Based on the hourly precipitation data from 1 $60 stations in Guizhou Provinceduring the rainy season ( from 15 April to 15 0ctober) in 2023, the forecast abilities of three high-resolutionregional modelS (CMA-MESO, CMA-GD, CMA-SH ) are evaluated in detail in terms of TS score, precipitationfrequency, intensity and peak period of precipitation in Guizhou. The evaluation results of the models on the dailyprecipitation forecasts show that the regional mean daily precipitation and frequency in the rainy season areoverestimated by all three models. The large-value center of the precipitation can be predicted by all modelsbasically, but the location of the precipitation center is by east and north relative to the observation. The CMA-GD shows a higher ""T""S in the forecasts of moderate rain to rainstorm, while CMA-MES0 is the worst for all precipitationmagnitudes in TS. The deviation time of precipitation and frequency forecast is different for each model based onthe forecast evaluation of 6 h accumulated precipitation in the 60 selected heavy precipitation days. "The negativeprecipitation deviations of CMA-MES0 forecast are mostly seen in the afternoon, while the forecasts by CMA-GDand CM A-Sll exhibit the largest positive deviations in the morning. Higher precipitation frequency is mostlyobserved in the northeastern part ( daytime) and western part ( nighttime) of Guizhou. CMA-GD performs better inforecasting the spalial and temporal variation of precipitation frequency, but the other two model forecasts havelarger deviations. In terms of the precipitation peak time on heavy precipitation days, the models can forecast moreaccurately for northeastern Guizhou, while the forecast time of peak precipitation by the three models in thesoutheast of Guizhou is at night rather than in the daytime from observation. ""To sum up, CMA-GD forecast showsthe best ""TS for daily precipitation on average and can correctly reproduce the diurnal variation of precipitation.Furthermore , the integrated evaluation method based on precipitation amount, frequency, intensity and diurnalvariation is helpful for correctly understanding the biases of model forecasts with complex terrains. |
| Key words: High-resolution Models;rainy season;evaluation;rainfall |
