基于自组织映射神经网络的双流机场强对流特征分析

Synoptic Pattern Analysis of Severe Convective Weather at Shuangliu Airport Based on Self-Organizing Map (SOM) Neural Network

摘要: 本文利用2019~2023年双流机场的机场METAR报和SPECI报，统计影响飞机起降的重要强对流天气(雷暴、中降水、大降水)，基于ERA5 0.25 * 0.25再分析资料利用SOM方法对筛选出的强对流天气形势进行聚类分型。结果表明：影响双流机场的强对流天气形势主要有4类，各类在发生季节、强对流频率及日变化上存在明显差异，以强降水为主的华南前汛期形势，主要出现在5月；以中阵雨伴雷暴为主的长江中下游梅雨形势，主要出现在6~7月；第三、四类为盛夏副高不同位相形势，主要出现在7~8月，其中第三类副高偏弱偏东，第四类副高偏强偏西以强降水为主。

Abstract: Using routine and special meteorological reports from Shuangliu International Airport during 2019~2023, this study statistically analyzes major types of severe convective weather affecting aircraft takeoff and landing, including thunderstorms, moderate precipitation, and heavy precipitation. The Self-Organizing Map (SOM) method is applied to ERA5 reanalysis data (0.25 * 0.25) to classify the synoptic patterns associated with these severe convective events. The results show that severe convective weather affecting Shuangliu International Airport can be categorized into four main synoptic types, each exhibiting distinct seasonal occurrences, convective frequencies, and diurnal variations. The first type, dominated by heavy precipitation, corresponds to the pre-flood season pattern in South China and mainly occurs in May. The second type, characterized by thunderstorms with moderate rainfall, is associated with the Meiyu pattern over the middle and lower reaches of the Yangtze River and occurs primarily from June to July. The third and fourth types represent different phases of the western Pacific subtropical high during midsummer (July~August): the third type features a weaker and more eastward-positioned subtropical high, while the fourth type is marked by a stronger and more westward-positioned high accompanied by heavy precipitation.