融合物理理解与模糊逻辑的分类强对流客观短期预报系统:(2)表现评估  

Forecasting System for Short-Term Multi-Category Convective Phenomena Combining Physical Understanding and Fuzzy Logic PartⅡ:Performance Evaluation

作　　者：田付友 郑永光 坚参扎西 吕新民[3] 孙建华[4] 黄玥 赤曲 TIAN Fuyou;ZHENG Yongguang;JIANCAN Zhaxi;LYU Xinmin;SUN Jianhua;HUANG Yue;CHI Qu(National Meteorological Centre,Beijing 100081;Meteorological Observatory of Tibet Autonomous Region,Lhasa 850000;395820 Troops of People’s Liberation Army,Beijing 102206;Key Laboratory of Cloud-Precipitation Physics and Severe Storms,Institute of Atmospheric Physics,Chinese Academy of Sciences,Beijing 100029)

机构地区：[1]国家气象中心,北京100081 [2]西藏自治区气象台,拉萨850000 [3]中国人民解放军95820部队,北京102206 [4]中国科学院大气物理研究所云降水物理与强风暴重点实验室,北京100029

出　　处：《气象》2024年第6期649-660,共12页Meteorological Monthly

基　　金：西藏自治区科技计划项目(XZ202101ZY0004G);国家自然科学基金联合基金项目(U2142202、U2342204);国家重点研发计划(2022YFC3004104);中国气象局重点创新团队(CMA2022ZD07)共同资助。

摘　　要：本文对分类强对流客观短期概率预报系统2022年6月13日强对流过程预报产品的表现进行分析,基于2022年的雷暴、短时强降水、雷暴大风及冰雹客观概率预报产品和可用的分类强对流监测实况资料,结合强对流预报业务中使用的空间检验方法和常用的确定性及概率性检验指标,对该短期预报系统提供的四类强对流天气客观概率预报产品进行了详细的性能评估。用于评估的预报资料是时段为2022年4月1日至9月30日每天08时(北京时)起报,96 h内逐12 h间隔的预报产品。预报个例分析显示,四类产品均可提前24 h指示需要关注的强对流天气区域。统计检验结果表明,短时强降水各方面性能最好,其次是雷暴,雷暴大风也有一定的可参考性。四类强对流天气预报产品均存在预报概率与实况频率相比偏高的过度预报问题。雷暴、短时强降水和雷暴大风预报产品均存在与预报覆盖时效有关的日变化。评估结果为预报模型和系统后续改进发展奠定了基础,为应用基于融合物理理解与模糊逻辑人工智能方法的分类强对流预报产品提供了有益参考。In this paper,the performance of the forecasting system of short-term multi-category convective phenomena in the event that happened on 13 June 2022 is analyzed first.Then,based on the objective probability forecasts of thunderstorm,short-time severe rainfall,thunderstorm gale and hail events in 2022 as well as available multi-category severe convective monitoring data,the performance of objective probability forecast products of the four types of severe convective weather provided by the short-term forecasting system is evaluated in detail by adopting the spatial test methods used in the severe convection forecast operations and the indices that indicate deterministic and probabilistic properties.The evaluated forecast period of the forecast products initiated at 08:00 BT from 1 April to 30 September 2022 is 96 h with interval of 12 h.Case studies show that the potential area of the four different convective phenomena could be well forecasted 24 h in advance.Statistical verification results show that the short-time severe rainfall forecast has the best performance among the four convective weather phenomena,followed by the forecast of thunderstorm.The forecast of the thunderstorm gale has certain applicability as well.There are obvious problems of overestimation in all the four convective weather phenomena compared to the observations.The diurnal variations of thunderstorm,short-time severe rainfall and thunderstorm gale forecasts are related to the forecast coverage time.These evaluation results are beneficial to subsequent improvement and development of forecast model and system,and could provide a useful reference for the operational application of multi-category severe convection forecast results based on the fusion of physical understanding and fuzzy logic artificial intelligence.

关 键 词：物理理解 模糊逻辑人工智能 分类强对流 短期预报系统 确定性属性 概率性属性 

分 类 号：P456[天文地球—大气科学及气象学] P445