Construction of deep-learning based WWBs parameterization for ENSO prediction

Westerly wind bursts (WWBs) significantly impact the occurrence and development of the El Niño-Southern Oscillation (ENSO). Current dynamical models, however, face significant challenges in representing WWBs. In this study, deep learning techniques were used to develop a new parameterization scheme...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Atmospheric research 2023-07, Vol.289 (C), p.106770, Article 106770
Hauptverfasser: You, Lirong, Tan, Xiaoxiao, Tang, Youmin
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Westerly wind bursts (WWBs) significantly impact the occurrence and development of the El Niño-Southern Oscillation (ENSO). Current dynamical models, however, face significant challenges in representing WWBs. In this study, deep learning techniques were used to develop a new parameterization scheme for WWBs and further compared against two widely used schemes. The results show that the scheme developed in this study has greater capability than previous schemes in reproducing WWBs characteristics, particularly in terms of occurrence probability, location, and duration. This improvement was mainly reflected in El Niño years, especially in strong events when the deep-learning-based scheme much realistically captures the location and strength of WWBs. It is expected that the new parameterization scheme will further improve ENSO prediction in dynamical models. •A deep-learning based WWBs parameterization scheme is designed to further improve ENSO prediction skill.•The deep-learning based scheme has a better capability than previous ones in reproducing WWBs characteristics.•The improvement of deep-learning based scheme is mainly reflected in El Niño years, especially in strong El Niño years.
ISSN:0169-8095
1873-2895
DOI:10.1016/j.atmosres.2023.106770