Preprint · 2026 · Open access

A geophysically constrained Transformer model for tropical cyclone track and intensity simulation over the western North Pacific and its contribution analysis

Ziqi Lin, Wenxuan Li, Shuo Wang, Jing Xu, Siqi Zhang, Heng Ma, Peijun Shi, Gangfeng Zhang

Research Square · Published 9 October 2026 · DOI 10.21203/rs.3.rs-11123951/v1

Open PDFView at the publisherCite this paper

In brief

“A geophysically constrained Transformer model for tropical cyclone track and intensity simulation over the western North Pacific and its contribution analysis” is a 2026 preprint by Ziqi Lin and 7 co-authors, published in Research Square. OpenAlex counts 0 citations as of 20 November 2026. It is open access under a CC BY licence: the full text is free to read at Research Square.

Citations
0
References
33
Authors
8

Abstract

Tropical cyclone represents one of the most devastating natural hazards in global coastal zones, inflicting massive economic damage and loss of life annually. Precise multi-step tropical cyclone forecasting is still confronted with great challenges, primarily attributed to intricate ocean-atmosphere interactions governing tropical cyclone movement. Here, we propose a geophysically constrained Transformer model to jointly predict tropical cyclone tracks and intensity using historical best-track records over the western North Pacific and reanalysis data, benchmarked against conventional XGBoost and LSTM models. Experimental results show that the geophysically constrained Transformer model realizes stable tropical cyclone forecasting, with track RMSEs of 200.72 km (6 h) and 357.02 km (36 h), and intensity RMSEs of 2.59 m/s (6 h) and 7.54 m/s (36 h). When compared with LSTM and XGBoost baselines, it reduces 6-hour track RMSE by 30% and 55%, and intensity RMSE by 23% and 36%, respectively, maintaining better accuracy over most forecast periods for both tasks. SHAP analysis further reveals that surface air pressure (contributing 32%) and wind speed (contributing 25%) are the dominant predictive variables for tropical cyclone track and intensity forecasting. By contrast, large-scale circulation and sea surface temperature function as secondary influencing factors, which further validates the physical interpretability and rationality of the developed model.. Overall, our findings advance the integration of deep learning techniques with meteorological and geophysical datasets for real-time operational tropical cyclone prediction, yielding critical implications for tropical cyclone monitoring and disaster risk governance.

Cite this paper

Formatted from the catalogue record. Check names, capitals and page numbers against the paper before you submit.

APA 7

Lin, Z., Li, W., Wang, S., Xu, J., Zhang, S., Ma, H., Shi, P., & Zhang, G. (2026). A geophysically constrained Transformer model for tropical cyclone track and intensity simulation over the western North Pacific and its contribution analysis. Research Square. https://doi.org/10.21203/rs.3.rs-11123951/v1

MLA 9

Lin, Ziqi, et al. “A geophysically constrained Transformer model for tropical cyclone track and intensity simulation over the western North Pacific and its contribution analysis.” Research Square, 2026. https://doi.org/10.21203/rs.3.rs-11123951/v1.

Chicago

Lin, Ziqi, Wenxuan Li, Shuo Wang, Jing Xu, Siqi Zhang, Heng Ma, Peijun Shi, and Gangfeng Zhang. 2026. “A geophysically constrained Transformer model for tropical cyclone track and intensity simulation over the western North Pacific and its contribution analysis.” Research Square. https://doi.org/10.21203/rs.3.rs-11123951/v1.

Harvard

Lin, Z. et al. (2026) ‘A geophysically constrained Transformer model for tropical cyclone track and intensity simulation over the western North Pacific and its contribution analysis’, Research Square. doi:10.21203/rs.3.rs-11123951/v1.

IEEE

Z. Lin et al., “A geophysically constrained Transformer model for tropical cyclone track and intensity simulation over the western North Pacific and its contribution analysis,” Research Square, 2026, doi: 10.21203/rs.3.rs-11123951/v1.

BibTeX

@misc{lin2026geophysically,
  title     = {{A geophysically constrained Transformer model for tropical cyclone track and intensity simulation over the western North Pacific and its contribution analysis}},
  author    = {Lin, Ziqi and Li, Wenxuan and Wang, Shuo and Xu, Jing and Zhang, Siqi and Ma, Heng and Shi, Peijun and Zhang, Gangfeng},
  howpublished = {Research Square},
  year      = {2026},
  doi       = {10.21203/rs.3.rs-11123951/v1},
  url       = {https://doi.org/10.21203/rs.3.rs-11123951/v1}
}

Download for a reference manager: BibTeX (.bib) · RIS for Zotero, Mendeley and EndNote (.ris)

Topic and keywords

Physical Sciences › Earth and Planetary Sciences › Atmospheric Science › Tropical and Extratropical Cyclones Research

  • tropical cyclone track forecasting
  • tropical cyclone intensity forecasting
  • transformer models
  • western North Pacific
  • ocean-atmosphere interaction
  • surface pressure
  • wind speed
  • sea surface temperature

Work with this paper

Most-cited papers in Earth and Planetary Sciences

All highly cited Earth and Planetary Sciences papers

Questions about this paper

How do I cite this paper in APA style?

Lin, Z., Li, W., Wang, S., Xu, J., Zhang, S., Ma, H., Shi, P., & Zhang, G. (2026). A geophysically constrained Transformer model for tropical cyclone track and intensity simulation over the western North Pacific and its contribution analysis. Research Square. https://doi.org/10.21203/rs.3.rs-11123951/v1

How many times has this paper been cited?

0 times, according to OpenAlex on 20 November 2026. Google Scholar and Scopus count different sets of sources, so their numbers differ.

Is this paper free to read?

Yes. It is open access at Research Square: sign in and open the PDF from this page.

Who wrote this paper?

Ziqi Lin, Wenxuan Li, Shuo Wang, Jing Xu, Siqi Zhang, Heng Ma, Peijun Shi, Gangfeng Zhang.

What is the DOI of this paper?

10.21203/rs.3.rs-11123951/v1, permanently at https://doi.org/10.21203/rs.3.rs-11123951/v1. Sign in and open it from this page.

Bibliographic data from OpenAlex (CC0), last checked 20 November 2026. ResearchGuru does not host this paper; once you are signed in, the links open it at the publisher or the repository that does.