Semi-automated tree-fall pattern identification using image processing technique: Application to alonsa, MB tornado. Issue 208 (January 2021)

• Record Type: Journal Article
• Authors: Rhee, Daniel M.; Lombardo, Franklin T.; Kadowaki, Jason
• Is Part Of: Journal of wind engineering and industrial aerodynamics. Issue 208(2021)
• Publication Date: 2021-01
• Subjects: Automated -- Tree-fall identification -- Tree-fall direction -- Tornado -- Image processing; Wind-pressure -- Periodicals; Buildings -- Aerodynamics -- Periodicals; Pression du vent -- Périodiques; Constructions -- Aérodynamique -- Périodiques; Buildings -- Aerodynamics; Wind-pressure; Periodicals
• DOI: 10.1016/j.jweia.2020.104399
• Languages: English
• ISSNs: 0167-6105

Abstract

Abstract: A significant number of tornadoes in North America occur in rural forested regions and often cause a massive volume of trees to fall down. As a result, the demand for documenting the tree damage in forests and research in estimating near-surface wind fields of tornadoes using forest damage have increased in the past years. However, excessive manual labor and time in identifying the damaged trees has been a challenge. Thus, a new method of identifying tree-fall patterns using image processing was developed. The method identifies downed trees in aerial photographs and determines the direction of the tree-fall. The leaves and stems of the fallen trees were identified and separated using an RGB color filter, and the median fall direction of the downed trees was determined using the Hough Transformation and the relative position between the leaf and stem pixels. The proposed method was applied to Alonsa, MB tornado in Canada, and the tree-fall directions of the tornado were identified. The manually and automatically identified tree-fall directions were compared and the error was assessed, in which 59% of the samples had less than a 20-degree difference and 73% had less than a 45-degree difference between the two methods.