Yunlong Zhou

I am a PhD candidate at the School of Intelligence Science and Technology, Nanjing University, advised by Prof. Xiao-Tong Yuan. Before that, I received my Master’s degree in Computer Technology from Nanjing University of Information Science & Technology (NUIST) in 2024, and my Bachelor’s degree in Computer Science and Technology from Shandong Technology and Business University in 2021.

My research focuses on spatiotemporal prediction and generative models, with precipitation nowcasting as the main application. Specifically, I work on:

Generative modeling: diffusion models, flow matching / rectified flow, and making them fast and physically grounded.
Spatiotemporal sequence prediction: Transformer and recurrent architectures, frequency-domain modeling and Fourier neural operators.
AI for weather: radar-based nowcasting, spectral fidelity of forecasts, and bringing research models into operational use.

I am currently looking for research internship opportunities in generative modeling and spatiotemporal prediction. Feel free to reach out by email.

selected publications

ICML
Learning to Refine: Spectral-Decoupled Iterative Refinement Framework for Precipitation Nowcasting

Yunlong Zhou, Chen Zhao, Danyang Peng, Fanfan Ji, and Xiao-Tong Yuan

In Proceedings of the 43rd International Conference on Machine Learning (ICML), 2026

Abs arXiv Bib HTML Code Slides

Accurate precipitation nowcasting is vital for disaster mitigation, but deep learning methods face a key trade-off: regression models produce over-smoothed, spectrally decaying predictions that blur convective details and violate turbulence power laws; diffusion models generate realistic yet unanchored hallucinations lacking physical grounding. We propose Spectral-Decoupled Iterative Refinement (SDIR), a deterministic framework that reformulates nowcasting as progressive frequency-decoupled refinement. SDIR first extracts a stable low-frequency synoptic skeleton, then iteratively refines high-frequency textures under physical constraints, eliminating both blurring and hallucinations. It features a dual-path design: the Synoptic Frequency-Guided Former (SFG-Former) with Scale-Adaptive Transformers for global structure, and the Fourier Residual Refiner (FR-Refiner) with Scale-Conditioned Fourier Neural Operators for fine residuals. A Physically Consistent Power Spectral Density (PCPSD) loss with dynamic masking enforces a turbulence-consistent spectral distribution. Experiments on three benchmarks show SDIR significantly outperforms SOTA methods in spatial accuracy while achieving spectral fidelity competitive with diffusion-based methods, enabling reliable high-resolution operational nowcasting.
@inproceedings{zhou2026sdir, title = {Learning to Refine: Spectral-Decoupled Iterative Refinement Framework for Precipitation Nowcasting}, author = {Zhou, Yunlong and Zhao, Chen and Peng, Danyang and Ji, Fanfan and Yuan, Xiao-Tong}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning (ICML)}, year = {2026}, }

ICML

LoPhyDA: Low-Rank Tensor and Physics Gradient Guided Diffusion for Atmospheric Data Assimilation

Danyang Peng, Yang Chen, Yunlong Zhou, and Xiao-Tong Yuan

In Proceedings of the 43rd International Conference on Machine Learning (ICML), 2026

Bib HTML Slides

@inproceedings{peng2026lophyda,
  title = {LoPhyDA: Low-Rank Tensor and Physics Gradient Guided Diffusion for Atmospheric Data Assimilation},
  author = {Peng, Danyang and Chen, Yang and Zhou, Yunlong and Yuan, Xiao-Tong},
  booktitle = {Proceedings of the 43rd International Conference on Machine Learning (ICML)},
  year = {2026},
}

EAAI

More realistic and accurate precipitation nowcasting with Conditional Rectified Flow Transformers

Yunlong Zhou, Chen Zhao, Fanfan Ji, Renlong Hang, Qingshan Liu, and 1 more author

Engineering Applications of Artificial Intelligence, 2026

DOI Bib Code

@article{zhou2026crft,
  title = {More realistic and accurate precipitation nowcasting with Conditional Rectified Flow Transformers},
  author = {Zhou, Yunlong and Zhao, Chen and Ji, Fanfan and Hang, Renlong and Liu, Qingshan and Yuan, Xiao-Tong},
  journal = {Engineering Applications of Artificial Intelligence},
  volume = {165},
  pages = {113402},
  year = {2026},
  doi = {10.1016/j.engappai.2025.113402},
}

JSTARS

Spatiotemporal Enhanced Adversarial Network for Precipitation Nowcasting

Yunlong Zhou, Renlong Hang, Fanfan Ji, Zefeng Pan, Qingshan Liu, and 1 more author

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2024

DOI Bib

@article{zhou2024stean,
  title = {Spatiotemporal Enhanced Adversarial Network for Precipitation Nowcasting},
  author = {Zhou, Yunlong and Hang, Renlong and Ji, Fanfan and Pan, Zefeng and Liu, Qingshan and Yuan, Xiao-Tong},
  journal = {IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing},
  volume = {17},
  pages = {7608--7620},
  year = {2024},
  doi = {10.1109/JSTARS.2024.3381822},
}