# Phys-Diff: A Physics-Inspired Latent Diffusion Model for Tropical Cyclone Forecasting

arXiv: https://arxiv.org/abs/2603.00521 · HTML: https://mlsanity.com/paper/2603.00521/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Flow matching, time series and science, one of 14 directions in Diffusion models.
- Builds on 6 of its 18 references that are in the map.
- First posted on arXiv 28 Feb 2026.

## Placed in

- Diffusion models / [Flow matching, time series and science](https://mlsanity.com/diffusion/flow-matching-time-series-and-science/)

## Builds on (most cited first, 6 of 6)

- [FengWu: Pushing the Skillful Global Medium-range Weather Forecast beyond 10 Days Lead](https://mlsanity.com/paper/2304.02948/) (2023)
- [Dynamical Tests of a Deep-Learning Weather Prediction Model](https://mlsanity.com/paper/2309.10867/) (2023)
- [Predicting Hurricane Trajectories using a Recurrent Neural Network](https://mlsanity.com/paper/1802.02548/) (2018)
- [Global Tropical Cyclone Intensity Forecasting with Multi-modal Multi-scale Causal Autoregressive Model](https://mlsanity.com/paper/2402.13270/) (2024)
- [VQLTI: Long-Term Tropical Cyclone Intensity Forecasting with Physical Constraints](https://mlsanity.com/paper/2501.18122/) (2025)
- [Improving Tropical Cyclone Forecasting With Video Diffusion Models](https://mlsanity.com/paper/2501.16003/) (2025)

## Abstract (arXiv)

Tropical cyclone (TC) forecasting is critical for disaster warning and emergency response. Deep learning methods address computational challenges but often neglect physical relationships between TC attributes, resulting in predictions lacking physical consistency. To address this, we propose Phys-Diff, a physics-inspired latent diffusion model that disentangles latent features into task-specific components (trajectory, pressure, wind speed) and employs cross-task attention to introduce prior physics-inspired inductive biases, thereby embedding physically consistent dependencies among TC attributes. Phys-Diff integrates multimodal data including historical cyclone attributes, ERA5 reanalysis data, and FengWu forecast fields via a Transformer encoder-decoder architecture, further enhancing forecasting performance. Experiments demonstrate state-of-the-art performance on global and regional datasets.