Deep learning approach for detecting tropical cyclones and their precursors in the simulation by a cloud-resolving global nonhydrostatic atmospheric model

Daisuke Matsuoka, Masuo Nakano, Daisuke Sugiyama, Seiichi Uchida

Research output: Contribution to journalArticle

3 Citations (Scopus)

Abstract

We propose a deep learning approach for identifying tropical cyclones (TCs) and their precursors. Twenty year simulated outgoing longwave radiation (OLR) calculated using a cloud-resolving global atmospheric simulation is used for training two-dimensional deep convolutional neural networks (CNNs). The CNNs are trained with 50,000 TCs and their precursors and 500,000 non-TC data for binary classification. Ensemble CNN classifiers are applied to 10 year independent global OLR data for detecting precursors and TCs. The performance of the CNNs is investigated for various basins, seasons, and lead times. The CNN model successfully detects TCs and their precursors in the western North Pacific in the period from July to November with a probability of detection (POD) of 79.9–89.1% and a false alarm ratio (FAR) of 32.8–53.4%. Detection results include 91.2%, 77.8%, and 74.8% of precursors 2, 5, and 7 days before their formation, respectively, in the western North Pacific. Furthermore, although the detection performance is correlated with the amount of training data and TC lifetimes, it is possible to achieve high detectability with a POD exceeding 70% and a FAR below 50% during TC season for several ocean basins, such as the North Atlantic, with a limited sample size and short lifetime. [Figure not available: see fulltext.].

Original languageEnglish
Article number80
JournalProgress in Earth and Planetary Science
Volume5
Issue number1
DOIs
Publication statusPublished - Dec 1 2018

Fingerprint

tropical cyclone
learning
simulation
longwave radiation
ocean basin
cyclone
atmospheric model
detection
basin

All Science Journal Classification (ASJC) codes

  • Earth and Planetary Sciences(all)

Cite this

Deep learning approach for detecting tropical cyclones and their precursors in the simulation by a cloud-resolving global nonhydrostatic atmospheric model. / Matsuoka, Daisuke; Nakano, Masuo; Sugiyama, Daisuke; Uchida, Seiichi.

In: Progress in Earth and Planetary Science, Vol. 5, No. 1, 80, 01.12.2018.

Research output: Contribution to journalArticle

@article{e5b45fc565d4423f98f805b62826625c,
title = "Deep learning approach for detecting tropical cyclones and their precursors in the simulation by a cloud-resolving global nonhydrostatic atmospheric model",
abstract = "We propose a deep learning approach for identifying tropical cyclones (TCs) and their precursors. Twenty year simulated outgoing longwave radiation (OLR) calculated using a cloud-resolving global atmospheric simulation is used for training two-dimensional deep convolutional neural networks (CNNs). The CNNs are trained with 50,000 TCs and their precursors and 500,000 non-TC data for binary classification. Ensemble CNN classifiers are applied to 10 year independent global OLR data for detecting precursors and TCs. The performance of the CNNs is investigated for various basins, seasons, and lead times. The CNN model successfully detects TCs and their precursors in the western North Pacific in the period from July to November with a probability of detection (POD) of 79.9–89.1{\%} and a false alarm ratio (FAR) of 32.8–53.4{\%}. Detection results include 91.2{\%}, 77.8{\%}, and 74.8{\%} of precursors 2, 5, and 7 days before their formation, respectively, in the western North Pacific. Furthermore, although the detection performance is correlated with the amount of training data and TC lifetimes, it is possible to achieve high detectability with a POD exceeding 70{\%} and a FAR below 50{\%} during TC season for several ocean basins, such as the North Atlantic, with a limited sample size and short lifetime. [Figure not available: see fulltext.].",
author = "Daisuke Matsuoka and Masuo Nakano and Daisuke Sugiyama and Seiichi Uchida",
year = "2018",
month = "12",
day = "1",
doi = "10.1186/s40645-018-0245-y",
language = "English",
volume = "5",
journal = "Progress in Earth and Planetary Science",
issn = "2197-4284",
publisher = "Springer Open",
number = "1",

}

TY - JOUR

T1 - Deep learning approach for detecting tropical cyclones and their precursors in the simulation by a cloud-resolving global nonhydrostatic atmospheric model

AU - Matsuoka, Daisuke

AU - Nakano, Masuo

AU - Sugiyama, Daisuke

AU - Uchida, Seiichi

PY - 2018/12/1

Y1 - 2018/12/1

N2 - We propose a deep learning approach for identifying tropical cyclones (TCs) and their precursors. Twenty year simulated outgoing longwave radiation (OLR) calculated using a cloud-resolving global atmospheric simulation is used for training two-dimensional deep convolutional neural networks (CNNs). The CNNs are trained with 50,000 TCs and their precursors and 500,000 non-TC data for binary classification. Ensemble CNN classifiers are applied to 10 year independent global OLR data for detecting precursors and TCs. The performance of the CNNs is investigated for various basins, seasons, and lead times. The CNN model successfully detects TCs and their precursors in the western North Pacific in the period from July to November with a probability of detection (POD) of 79.9–89.1% and a false alarm ratio (FAR) of 32.8–53.4%. Detection results include 91.2%, 77.8%, and 74.8% of precursors 2, 5, and 7 days before their formation, respectively, in the western North Pacific. Furthermore, although the detection performance is correlated with the amount of training data and TC lifetimes, it is possible to achieve high detectability with a POD exceeding 70% and a FAR below 50% during TC season for several ocean basins, such as the North Atlantic, with a limited sample size and short lifetime. [Figure not available: see fulltext.].

AB - We propose a deep learning approach for identifying tropical cyclones (TCs) and their precursors. Twenty year simulated outgoing longwave radiation (OLR) calculated using a cloud-resolving global atmospheric simulation is used for training two-dimensional deep convolutional neural networks (CNNs). The CNNs are trained with 50,000 TCs and their precursors and 500,000 non-TC data for binary classification. Ensemble CNN classifiers are applied to 10 year independent global OLR data for detecting precursors and TCs. The performance of the CNNs is investigated for various basins, seasons, and lead times. The CNN model successfully detects TCs and their precursors in the western North Pacific in the period from July to November with a probability of detection (POD) of 79.9–89.1% and a false alarm ratio (FAR) of 32.8–53.4%. Detection results include 91.2%, 77.8%, and 74.8% of precursors 2, 5, and 7 days before their formation, respectively, in the western North Pacific. Furthermore, although the detection performance is correlated with the amount of training data and TC lifetimes, it is possible to achieve high detectability with a POD exceeding 70% and a FAR below 50% during TC season for several ocean basins, such as the North Atlantic, with a limited sample size and short lifetime. [Figure not available: see fulltext.].

UR - http://www.scopus.com/inward/record.url?scp=85059023556&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85059023556&partnerID=8YFLogxK

U2 - 10.1186/s40645-018-0245-y

DO - 10.1186/s40645-018-0245-y

M3 - Article

AN - SCOPUS:85059023556

VL - 5

JO - Progress in Earth and Planetary Science

JF - Progress in Earth and Planetary Science

SN - 2197-4284

IS - 1

M1 - 80

ER -