The application of decision tree to intensity change classification of tropical cyclones in western North Pacific

Abstract

This study applies the C4.5 algorithm to classify tropical cyclone (TC) intensity change in the western North Pacific. The 24 h change in TC intensity (i.e., intensifying and weakening) is regarded as a binary classification problem. A decision tree, with three variables and five leaf nodes, is built by the C4.5 algorithm. The variables include intensification potential (maximum potential intensity minus current intensity), previous 12 h intensity change, and zonal wind shear. All five rules, discovered from the tree by forming a path from the root node to each leaf node, can be interpreted by theories on TC intensification. Data mining results identify a predictor set (i.e., the mined rules) with high classification accuracy. The present study suggests that this data mining approach can shed some light on investigating TC intensity change processes and therefore has the potential to improve the forecasting of TC intensity.

Statistical‐dynamical seasonal forecast of western North Pacific and East Asia landfalling tropical cyclones using the high‐resolution GFDL FLOR coupled model

“…Statistical models have been used to forecast TC occurrence frequency/genesis [e.g., Chan , ; Chan et al ., ; Klotzbach , ; Wang et al ., ; Zhang et al ., ], track [e.g., J.‐H. Kim et al ., ; Zhang et al ., ], landfall [e.g., Goh and Chan , ; Zhang et al ., ,], and intensity [e.g., Knaff et al ., ; Zhang et al ., ] in the western North Pacific (WNP). Scientists have widely used Poisson regression [e.g., Liu and Chan , ; Chu and Zhao , ], Bayesian models [e.g., Chu et al ., ; Lu et al ., ], and multiple linear regression models [ Fan , ; Fan and Wang , ] to predict the occurrence and frequency of WNP TCs.…”

“…Over the recent decades, considerable attention has been paid to improve TC predictions based on statistical methods and dynamic models [e.g., Nicholls, 1979;Gray et al, 1993;Marks and Shay, 1998;Chan et al, 2001;Fan and Wang, 2009;Vecchi et al, , 2013Vecchi et al, , 2014Villarini and Vecchi, 2013]. Statistical models have been used to forecast TC occurrence frequency/genesis [e.g., Chan, 1995;Chan et al, 1998Chan et al, , 2001Klotzbach, 2007;Wang et al, 2013;Zhang et al, 2015], track [e.g., J.-H. Zhang et al, 2013a], landfall [e.g., Goh and Chan, 2010;Zhang et al, 2013aZhang et al, ,2013b, and intensity [e.g., Knaff et al, 2005;Zhang et al, 2013d] in the western North Pacific (WNP). Scientists have widely used Poisson regression [e.g., Liu and Chan, 2003;Chu and Zhao, 2007], Bayesian models [e.g., Chu et al, 2010;Lu et al, 2010], and multiple linear regression models [Fan, 2007;Fan and Wang, 2009] to predict the occurrence and frequency of WNP TCs.…”

A prediction scheme for the frequency of summer tropical cyclone landfalling over China based on data mining methods

“…Zheng et al () also classified the TC tracks over the same ocean using the k ‐mean clustering algorithm, which is not adaptive to track length (Camargo et al , ). Zhang et al () classified and predicted the changes of TCs (strengthened and weakened) by using a decision tree algorithm. This algorithm also performed well when applied to TC landfalls and recurvature (Zhang et al , , ).…”

Application of Machine Learning to Classification of Volcanic Deformation in Routinely Generated InSAR Data

“…In the Earth Sciences, machine learning has been employed in several applications (Lary et al, 2016), such as predicting earthquake magnitudes (Adeli & Panakkat, 2009), land surface classification (C. Li et al, 2014), vegetation indices (Brown et al, 2008), landslide susceptibility mapping (Yilmaz, 2010), and so forth. The techniques used previously include tree-based methods (Wei et al, 2013), artificial neural networks (Conforti et al, 2014), support vector machines (SVMs; Tien Bui et al, 2017), and Bayesian methods (Totaro et al, 2016).…”