Application of Unsupervised Learning Techniques to Identify Atlantic Tropical Cyclone Rapid Intensification Environments

Abstract

Tropical cyclone (TC) track forecasts have improved in recent decades while intensity forecasts, particularly predictions of rapid intensification (RI), continue to show low skill. Many statistical methods have shown promise in predicting RI using environmental fields, although these methods rely heavily upon supervised learning techniques such as classification. Advances in unsupervised learning techniques, particularly those that integrate nonlinearity into the class separation problem, can improve discrimination ability for difficult tasks such as RI prediction. This study quantifies separability between RI and non-RI environments for 2004–16 Atlantic Ocean TCs using an unsupervised learning method that blends principal component analysis with k-means cluster analysis. Input fields consisted of TC-centered 1° Global Forecast System analysis (GFSA) grids (170 different variables and isobaric levels) for 3605 TC samples and five domain sizes. Results are directly compared with separability offered by operational RI forecast predictors for eight RI definitions. The unsupervised learning procedure produced improved separability over operational predictors for all eight RI definitions, five of which showed statistically significant improvement. Composites from these best-separating GFSA fields highlighted the importance of mid- and upper-level relative humidity in identifying the onset of short-term RI, whereas long-term, higher-magnitude RI was generally associated with weaker absolute vorticity. Other useful predictors included optimal thermodynamic RI ingredients along the mean trajectory of the TC. The results suggest that the orientation of a more favorable thermodynamic environment relative to the TC and midlevel vorticity magnitudes could be useful predictors for RI.

Evaluating Changes in the Moisture Sources for Tropical Cyclones Precipitation in the North Atlantic That Underwent Extratropical Transition

“…A similar procedure was applied for extracting the TC‐related precipitation from the MSWEP data set within 1,000 km of the TC center (Figure S1 in Supporting Information ). The TC‐centric methodology has been previously applied, that is, for investigating synoptic‐scale precursors for TC rapid intensification process (e.g., Grimes & Mercer, 2015; Mercer et al., 2021; Wang et al., 2022), and TC features from the ERA‐5 reanalysis (Slocum et al., 2022), for studying the TC precipitation (Naufal et al., 2022), and for estimating the radius of the outermost closed isobar (Weber et al., 2014).…”

An Advanced Artificial Intelligence System for Identifying the Near-Core Impact Features to Tropical Cyclone Rapid Intensification from the ERA-Interim Data

“…Every year, new instances and possibly new variables are added to the SHIPS Developmental Data while some old variables may be removed. SHIPS data are widely used for statistical or statistical-dynamical TC intensity prediction as well as in NHC's probabilistic RI guidance [3][4][5][6][7][8]18]. The 2018 SHIPS Developmental Data used in this study had TC instances from 1982 to 2017 in the Atlantic Basin.…”

“…The most recent version for statistical RI prediction was developed in 2015 by Kaplan et al [5]. In spite of the simple statistical models, more advanced machine learning models were employed to improve the RI prediction performance and identify extra important features [4,[7][8][9].…”