基于2024年成都七月上中旬雷暴初生预报研究

Research on Thunderstorm Initiation Forecasting in Early to Mid-July 2024 in Chengdu

摘要: 为了研究雷暴初生阶段的对流云团发生发展机理和云团相关特征，以有效提高预警效率，文章使用中国气象局常规气象观测资料与日本气象厅(JMA)的葵花9号卫星的云图资料，利用卫星的单通道阈值法的云检测、多通道组合检测初生云系以及采用其他通道的方法，对发生在2024年成都地区七月上中旬的2次雷暴初生云团特征及其产生的暴雨过程进行分析，结果表明：1) 雷暴天气从初生至成熟阶段历时短，天气尺度小，伴随的每小时降水大多在20 mm以上，极端情况下每小时可达到55 mm以上。2) 单通道阈值法识别的初生云团的云顶亮温值位于200 K至240 K，随时间推移呈下降趋势，当云顶亮温低于235 K的阈值时，雷暴天气发生。3) 多通道协同方法下的雷暴云团在红外–分裂窗和红外–水汽通道的亮温差，在对流中心附近15~30 km的占比随时间显著增加，当红外–分裂窗通道亮温差位于0.5~5 K之间时，降水量增大且雷暴天气发生。4) 云有效半径的增大和云光学厚度的减小，表明雷暴云团从初生往成熟期过渡，是提前预警的关键时刻，云有效半径超过30 μm，以及云光学厚度低于25时，雷暴天气发生概率增加，初生云团云顶高度持续抬升至9~13 km，同时云顶温度下降，皆与对流发展密切相干。5) 经时效性检验，基于葵花9号光谱通道特征识别法相较于地面观测能够提前预警10~90 min，表现出较高的准确性和可靠性。通过研究雷暴云团初生阶段对流云的形成机制及其演变特征，有助于提升成都地区乃至全国的气象预警的精准性，从而降低其对公众安全和国民生产的潜在危害，对提高防灾减灾能力有一定的科学意义。

Abstract: To investigate the mechanisms of convective cloud cluster development during thunderstorm initiation and analyze their associated characteristics for improving early warning efficiency, this study utilizes conventional meteorological observation data from the China Meteorological Administration (CMA) and cloud imagery from the Japan Meteorological Agency’s (JMA) Himawari-9 satellite. By employing single-channel threshold-based cloud detection, multi-channel composite detection for nascent cloud systems, and other channel-based methods, we analyze the characteristics of thunderstorm-initiated cloud clusters and their associated heavy rainfall events in Chengdu during early to mid-July 2024. The results indicate that: 1) Thunderstorm weather progresses rapidly from the initial to the mature stage, with small synoptic-scale features and hourly precipitation mostly exceeding 20 mm, reaching over 55 mm per hour under extreme conditions. 2) The brightness temperature of nascent cloud clusters identified by the single-channel threshold method ranges between 200 K and 240 K, showing a decreasing trend over time. Thunderstorms are likely to occur when the cloud-top brightness temperature falls below the threshold of 233 K. 3) For thunderstorm cloud clusters detected via multi-channel synergy, the brightness temperature differences (BTDs) between infrared (IR)-split window and IR-water vapor channels increase significantly within 15-30 km around the convective center. Precipitation intensifies, and thunderstorms are more likely to occur when the BTD in the IR-split window channel ranges between 0.5-5 K. 4) An increase in cloud effective radius and a decrease in cloud optical thickness indicate the transition of thunderstorm cloud clusters from the initial to the mature stage, which is a critical period for early warning. When the cloud effective radius exceeds 30 μm and the cloud optical thickness drops below 25, the probability of thunderstorm occurrence increases. The continuous rise of cloud-top height to 9-13 km, accompanied by a decrease in cloud-top temperature, is closely related to convective development. 5) According to timeliness verification, the spectral channel feature identification method based on Himawari-9 provides warnings 10-90 minutes earlier than ground-based observations, demonstrating high accuracy and reliability. By studying the formation mechanisms and evolutionary characteristics of convective clouds during the initial stage of thunderstorm cloud clusters, this research contributes to enhancing the accuracy of meteorological warnings in the Chengdu region and nationwide. This advancement helps mitigate potential risks to public safety and national production, offering scientific significance for improving disaster prevention and mitigation capabilities.