Sprite Distribution of Different Polarities From ISUAL Observations With Machine Learning Method. Issue 19 (30th September 2022)

• Record Type: Journal Article
• Authors: Zhang, Mao; Lu, Gaopeng; Wang, Ziyi; Peng, Kang‐Ming; Huang, Hailiang; Ren, Huan; Liu, Feifan; Lei, Jiuhou
• Is Part Of: Journal of geophysical research. Volume 127:Issue 19(2022)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2022-09-30
• Subjects: machine learning -- red sprite -- sea‐land contrast -- lightning; Atmospheric physics -- Periodicals; Geophysics -- Periodicals; 551.5
• DOI: 10.1029/2022JD036968
• Languages: English
• ISSNs: 2169-897X

Abstract

Abstract: The morphological features of sprites are closely related to the polarity of their causative lightning strokes. Using the machine learning method, we develop a model with an accuracy of 93.8% to identify the polarity of sprite‐producing cloud‐to‐ground (CG) lightning strokes for events recorded during the Imager of Sprites and Upper Atmospheric Lightning (ISUAL) mission. Approximately 17% of the sprites are identified to be produced by negative CG lightning strokes. The global distribution of the polarity of sprite‐producing CG lightning strokes suggests that the ratio of sprites produced by negative CG lightning strokes relative to sprites produced by positive ones varies with latitude and sea‐land distribution. Sprites produced by negative CG lightning strokes appear to be generated in the tropical regions below 20° latitude and the oceanic area. Moreover, the proportion of sprites produced by negative CG lightning strokes over Africa and North America are much smaller than that over the rest of the continents and the sea. Plain Language Summary: Sprites are lightning‐induced optical transient emissions in the mesosphere. The morphological features of sprites are observed to be quite complex but closely related to the polarity of their causative cloud‐to‐ground (CG) lightning strokes. In our previous studies, the polarity of sprite‐producing CG lightning strokes has been determined for hundreds of events with the concurrent ground‐based measurement of associated lightning sferics. Based on the machine learning method, we develop a model to identify the polarity of sprite‐producing CG lightning strokes for events observed by the Imager of Sprites and Upper Atmospheric Lightning (ISUAL) mission. With such a model, the polarity of sprite‐producing CG lightning strokes without ground‐based observations can be recognized. In total, 258 sprites produced by negative CG lightning strokes, roughly 17% of the whole data set (1522) are identified. The proportion of sprites produced by negative CG lightning strokes greatly varies with latitude and sea‐land distribution, and it is greater in the tropical regions below 20° latitude and also larger over the sea. Moreover, the proportion of sprites produced by negative CG lightning strokes over Africa and North America is much smaller than that over the rest of the continents and the sea. Key Points: A model based on machine learning method is developed to determine the polarity of Imager of Sprites and Upper Atmospheric Lightning sprites, with an accuracy of 93.8% The ratio of negative sprites to positive sprites varies with latitude and sea‐land distribution The proportion of negative sprites over Africa and North America are much smaller than that over the rest of the continents and the sea