利用中壢特高頻雷達觀測大氣閃電已有將近30年的歷史，但過去所發展的閃電回波訊號分析方法與閃電定位的雷達干涉法，並不完善，導致所獲得的閃電特性不是很精確。本研究利用2012年所增設的干涉天線陣列，推導出閃電發生次數的理論統計模型，可以精確地描述背景雜訊與階梯先導(Step Leader)雷達回波的功率分布。為了使雷達閃電資料有更快的自動化演算法，使用DBSCAN（Density-Based Spatial Clustering of Applications with Noise）依照階梯先導雷達回波將閃電事件分類，加速找尋主動式觀測閃電回波的方法。在定位上解決雷達回波的相位混淆(Phase Ambiguity)問題，並針對定位方法的誤區做進一步討論，且利用空拍機與民航機的廣播式自動相關監視（Automatic Dependent Surveillance-Broadcast, ADS-B）計算出系統相位偏差Δϕ_12=-3.9817°、Δϕ_13=-5.2656°、Δϕ_14=67.9898°、Δϕ_15=-54.019°，得以精確的定位出閃電在三維空間中的位置，並進而辨識出雲閃(Cloud-to-cloud Lightning)與地閃(Cloud-to-ground Lightning)的類別，最後與氣象局的閃電落雷系統即時觀測系統定位結果比較，探討兩系統之差異與過去研究之異同。;The Chung-Li VHF radar Station has been used to observe atmospheric lightning for nearly 30 years, but the lightning echo signal analysis method and the radar interferometry method for lightning location developed in the past are not perfect, resulting in inaccurate lightning characteristics. In this study, an interferometric antenna arrays was set up in the Republic of China in 2012 to derive a theoretical statistical model for the number of lightning occurrences, which can accurately describe the power distribution of background noise and step leader radar echoes. To enable faster automatic algorithms for radar lightning data, DBSCAN (Density-Based Spatial Clustering of Applications with Noise) was used to classify lightning events according to step leader radar echoes, accelerating the search for active lightning methods. In terms of positioning, the phase ambiguity problem of lightning echoes was solved, and further discussions were conducted on the error region of the positioning method, enabling the precise positioning of lightning in three-dimensional space and identifying the categories of intra-cloud lightning and cloud-to-ground lightning. Finally, compare the differences between the two systems and the similarities and differences of previous studies with the real-time observation system positioning results of the lightning strike system of the CWB.
