• Home
  • Search
  • Ensemble random forest for tropical cyclone tracking
  • https://doi.org/10.5194/nhess-25-4655-2025Copy DOI Icon

Ensemble random forest for tropical cyclone tracking

Show More
  • Abstract
  • PDF
  • Literature Map
  • References
  • Similar Papers
Abstract

Abstract. Even though tropical cyclones (TCs) are well documented during the intense part of their lifecycle until they weaken, many physical and statistical properties governing them are not well captured by gridded reanalysis or simulated by Earth System Models. Thus, tracking TCs remains a matter of interest for investigating observed and simulated tropical cyclones. Two types of cyclone tracking schemes are available. On the one hand, some trackers rely on physical and dynamical properties of the TCs and user-prescribed thresholds, which make them rigid. They need numerous variables that are not always available in the models. On the other hand, trackers leaning on deep learning need, by nature, large amounts of data and computing power. Besides, given the number of physical variables required for the tracking, they can be prone to overfitting, which hinders their transferability to climate models. This study explores the ability of a Random Forest (RF) approach to track TCs with a limited number of aggregated variables. Our analysis focuses on the Eastern North Pacific and North Atlantic basins, for which 514 and 431 observed tropical cyclone track records are available from the IBTrACS database during the 1980–2021 period. For each 6-hourly time step, RF associates TC occurrence or absence (1 or 0) to atmospheric situations described by predictors extracted from the ERA5 reanalysis. Hence, the tracking is considered a binary supervised classification problem of TC-free (zero) and TC (one) situations. Then, situations with TC occurrences are stitched to reconstruct TC trajectories. Results show the good ability and performance of this method for tracking tropical cyclones over both basins and good temporal and spatial generalisation. RF has a similar TC detection rate as trackers based on TCs' properties and a significantly lower false alarm rate. RF allows us to detect TC situations for diverse predictor combinations, which brings more flexibility than threshold-based trackers. Last but not least, this study sheds light on the most relevant variables for tropical cyclone detection.

Loading PDF

Similar Papers
  • Research Article
  • Citations205

Assessing Impacts of Global Warming on Tropical Cyclone Tracks*

  • Apr 01, 2004
  • Journal of Climate
  • Liguang Wu +1
  • Research Article
  • Citations16

Eastern North Pacific tropical cyclone activity in historical and future CMIP5 experiments: assessment with a model-independent tracking scheme

  • May 28, 2019
  • Climate Dynamics
  • Samuel S Bell +4
  • Research Article
  • Citations1

Mechanism for initial brows-like meso-scale vortex effects on tropical cyclone track

  • Aug 05, 2011
  • Science China Earth Sciences
  • Zhexian Luo +1
  • Research Article
  • Citations20

Using the Orthogonal Conditional Nonlinear Optimal Perturbations Approach to Address the Uncertainties of Tropical Cyclone Track Forecasts Generated by the WRF Model

  • Oct 01, 2023
  • Weather and Forecasting
  • Han Zhang +2
  • Research Article
  • Citations27

Idealized simulations of the effect of local and remote topographies on tropical cyclone tracks

  • Dec 30, 2014
  • Quarterly Journal of the Royal Meteorological Society
  • Chi Kit Tang +1
  • Research Article
  • Citations68

Forced and Internal Variability of Tropical Cyclone Track Density in the Western North Pacific*

  • Dec 31, 2014
  • Journal of Climate
  • Wei Mei +3
  • Research Article
  • Citations53

Lightning in Eastern North Pacific Tropical Cyclones: A Comparison to the North Atlantic

  • Dec 29, 2015
  • Monthly Weather Review
  • Stephanie N Stevenson +2
  • Book Chapter
  • Citations3

Tropical Cyclone Rainfall in the Observations, Reanalysis and ARPEGE Simulations in the North Atlantic Basin

  • Jan 01, 2010
  • Anne S Daloz +2
  • PDF
  • Research Article
  • Citations10

Tropical cyclone formation regions in CMIP5 models: a global performance assessment and projected changes

  • Aug 29, 2020
  • Climate Dynamics
  • K J Tory +2
  • Research Article
  • Citations40

Impacts of Climate Change on Tropical Cyclones in the Western North Pacific Basin. Part II: Late Twenty-First Century Projections

  • Jan 04, 2019
  • Tropical Cyclone Research and Review
  • Ming Ying +3
  • Research Article
  • Citations45

Statistical Assessment of the OWZ Tropical Cyclone Tracking Scheme in ERA-Interim

  • Dec 19, 2017
  • Journal of Climate
  • Samuel S Bell +3
  • Research Article
  • Citations22

Sensitivity of Tropical Cyclone Track to the Vertical Structure of a Nearby Monsoon Gyre

  • Jun 01, 2018
  • Journal of the Atmospheric Sciences
  • Xuyang Ge +4
  • Research Article
  • Citations33

Climatological features of WRF-simulated tropical cyclones over the western North Pacific

  • Nov 21, 2014
  • Climate Dynamics
  • Dasol Kim +4
  • Research Article
  • Citations4

An Example of the Value of Strong Climatological Signals in Tropical Cyclone Track Forecasting: Hurricane Ivan (2004)

  • May 01, 2006
  • Monthly Weather Review
  • Bradford S Barrett +2
  • Research Article
  • Citations24

A Statistical Analysis of the Relationship between Upper-Tropospheric Cold Low and Tropical Cyclone Track and Intensity Change over the Western North Pacific

  • May 01, 2016
  • Monthly Weather Review
  • Na Wei +4
Cactus Communications logo

Copyright 2026 Cactus Communications. All rights reserved.