EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Multi-Scale Fourier Temporal Network for Multi-Source Precipitation Nowcasting

Jing Huang iD, Shanmin Yang iD, Xiaojie Li iD, Xi Wu iD

Sensors · 08.04.2026 · 26 (8) · 2303

DOI: 10.3390/s26082303 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Sensors, online erschienen am 08.04.2026. Verfasst von Jing Huang und 3 weiteren Autor:innen. Beteiligte Einrichtung: School of Computer Science, Chengdu University of Information Technology, Chengdu 610225, China. Die Arbeit nennt 30 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Sichuan Science and Technology program, National Natural Science Foundation of China, Sichuan Science and Technology.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Accurate precipitation nowcasting plays an important role in disaster prevention and hydrometeorological applications, yet it remains highly challenging due to the complex spatiotemporal variability and multi-scale structural characteristics of precipitation systems. Existing deep learning methods are largely data-driven and often struggle to effectively exploit multi-source observations or learn physically meaningful representations. To address these limitations, this study proposes a Multi-Scale Frequency–Temporal Network (MS-FTNet) for precipitation nowcasting. The framework leverages Fourier transform-based frequency-domain modeling to achieve an interpretable multi-scale decomposition of precipitation dynamics. Specifically, low-frequency components capture large-scale stratiform patterns and their temporal evolution, while high-frequency components represent localized convective structures and abrupt variations. Building on this, a Global Feature Collaboration (GFC) module integrates global frequency-domain representations with multi-scale convolutional features, and an Adaptive Temporal Fusion (ATF) module enhances temporal dependency modeling. Experiments on the SEVIR dataset demonstrate that MS-FTNet consistently outperforms representative baseline models in terms of MSE, CSI, and LPIPS, particularly for heavy precipitation events and longer forecast lead times.

Schlagwörter

Stichwörter der Autor:innen

frequency-domain representationmulti-scale modelingprecipitation nowcastingradar-satellite fusion

Autor:innen und Institutionen

  • School of Computer Science, Chengdu University of Information Technology, Chengdu 610225, China.

ORCID-Profile: Jing Huang, Shanmin Yang, Xiaojie Li, Xi Wu

Förderung

  • Sichuan Science and Technology program
  • National Natural Science Foundation of China
  • Sichuan Science and Technology

Zitieren

APA

Huang, J., Yang, S., Li, X., & Wu, X. (2026). Multi-Scale Fourier Temporal Network for Multi-Source Precipitation Nowcasting. Sensors, 26(8), 2303. https://doi.org/10.3390/s26082303

Vancouver

Huang J, Yang S, Li X, Wu X. Multi-Scale Fourier Temporal Network for Multi-Source Precipitation Nowcasting. Sensors. 2026;26(8):2303. doi:10.3390/s26082303

BibTeX
@article{Huang2026,
  author = {Huang, Jing and Yang, Shanmin and Li, Xiaojie and Wu, Xi},
  title = {Multi-Scale Fourier Temporal Network for Multi-Source Precipitation Nowcasting},
  journal = {Sensors},
  year = {2026},
  volume = {26},
  number = {8},
  pages = {2303},
  doi = {10.3390/s26082303}
}
Kostenlos starten →