A Machine-Learning Based Tool for Diagnosing Inland Tropical Cyclone Maintenance or Intensification Events
Tropical Cyclone Maintenance and Intensification (TCMI) is a generalized definition of tropical cyclones that strengthen or maintain intensity inland while maintaining tropical characteristics. Herein, a novel methodology, using a machine learning method was created to examine the tropical cyclone r...
Gespeichert in:
Veröffentlicht in: | Frontiers in earth science (Lausanne) 2022-03, Vol.10 |
---|---|
Hauptverfasser: | , |
Format: | Artikel |
Sprache: | eng |
Schlagworte: | |
Online-Zugang: | Volltext |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
Zusammenfassung: | Tropical Cyclone Maintenance and Intensification (TCMI) is a generalized definition of tropical cyclones that strengthen or maintain intensity inland while maintaining tropical characteristics. Herein, a novel methodology, using a machine learning method was created to examine the tropical cyclone record to improve climatological representation of such cases. Using the International Best Track Archive for Climate Stewardship (IBTrACS) dataset, individual times of inland tropical cyclones were classified into TCMI and non-TCMI (weakening) events. The MERRA-2 dataset was applied to develop a prototypical machine-learning model to help diagnose future TCMI events. A list of possible TCMI storms for case studies in future analyses is provided. Two of these storms were examined for attributes characteristic of the Brown Ocean Effect, a hypothesized mechanism for TCMI centered on warm, moist soils. It was revealed that variables that were important at the time of storm arrival were important the prior day, which indicates that a TCMI event is a reaction to the environment. Moreover, the variables that were finally selected show a heavy emphasis on land-surface processes. This supports the idea that the accurate representation of the land surface state is critical to the accurate diagnosis of TCMI. |
---|---|
ISSN: | 2296-6463 2296-6463 |
DOI: | 10.3389/feart.2022.818671 |