Application of Artificial Intelligence in Lightning Detection and Nowcasting Using Polarimetric RADAR Data

Artificial Intelligence can be used to predict various weather phenomenon (McGovern et al. 2017). In this study, our focus is to use storm data in machine learning algorithm for lightning nowcasting. RADAR reflectivity has been associated with lightning. It has been found that the storm with reflectivity value of 30 dBZ or higher and at freezing temperature indicates the presence of lightning flashes (Petersen et al. 1996). The use of storm data in machine learning algorithm can be used for lightning nowcasting. On August 31, 2017, lightning occurred for several hours around Tate Carin’s Meteorological Station, Hongkong. Lightning data, both cloud to ground and intracloud, had been collected from Lightning Location Information System (LLIS) Network during three hours of storm. Different dual-pol variables from S-band RADAR (CINRAD) in nearby location has been associated with lightning. Storm Cells with higher reflectivity values and with high altitude have shown high level of correlation with number of flashes. Features like maximum reflectivity, average reflectivity, storm area, storm volume, maximum altitude can be used to indicate the number of flashes in the storm. Moreover, with the aid of storm tracking, features of storms from previous scans can be used to predict the number of flashes in the future storms. Storm parameters and number of flashes can be treated as input and output respectively to train a machine learning model using supervised learning approach. The model can be used for prediction with different data sets of storm and flashes including US National Lightning Detection Network with NEXRAD Reflectivity.

McGovern, A., and Coauthors, 2017: Using artificial intelligence to improve real-time decision-making for high-impact weather. Bulletin of the American Meteorological Society, 98, 2073-2090.

Petersen, W. A., S. A. Rutledge, and R. E. Orville, 1996: Cloud-to-ground lightning observations from TOGA COARE: Selected results and lightning location algorithms. Monthly weather review, 124, 602-620.