Machine learning based automated identification of thunderstorms from anemometric records using shapelet transform. Issue 220 (January 2022)

• Record Type: Journal Article
• Authors: Arul, Monica; Kareem, Ahsan; Burlando, Massimiliano; Solari, Giovanni
• Is Part Of: Journal of wind engineering and industrial aerodynamics. Issue 220(2021)
• Publication Date: 2022-01
• Subjects: Thunderstorm detection -- Time series shapelets -- Shapelet transform -- Machine learning -- Wind monitoring network; Wind-pressure -- Periodicals; Buildings -- Aerodynamics -- Periodicals; Pression du vent -- Périodiques; Constructions -- Aérodynamique -- Périodiques; Buildings -- Aerodynamics; Wind-pressure; Periodicals
• DOI: 10.1016/j.jweia.2021.104856
• Languages: English
• ISSNs: 0167-6105

Abstract

Abstract: Detection of thunderstorms is important to the wind hazard community to better understand extreme wind field characteristics and associated wind-induced load effects on structures. This paper contributes to this effort by proposing an innovative course of research that uses machine learning techniques, independent of wind statistics-based parameters, to autonomously identify thunderstorms from large databases containing high-frequency sampled continuous wind speed data. In this context, the use of Shapelet transform is proposed to identify key individual attributes distinctive to extreme wind events based on similarity of the shape of their time series signature. This shape-based representation, when combined with machine learning algorithms, yields a practical event detection procedure with minimal domain expertise. In this paper, the shapelet transform along with Random Forest classifier is employed for the identification of thunderstorms from 1-year of data from 14 ultrasonic anemometers that are a part of an extensive in-situ wind monitoring network in the Northern Mediterranean ports. A collective total of 240 non-stationary records associated with thunderstorms were identified using this method. The results lead to enhancing the pool of thunderstorm data for a more comprehensive understanding of a wide variety of thunderstorms that have not been previously detected using conventional gust factor-based methods. Highlights: The use of Shapelet Transform (ST) based machine learning is proposed to detect thunderstorms from wind velocity measurements. This method is independent of wind statistics-based parameters so that a wide variety of thunderstorms can be identified. Data from 14 ultrasonic anemometers in the Northern Mediterranean ports are used A collective total of 240 non-stationary records associated with thunderstorms are autonomously identified using this method.