aiqbench analysis
This research demonstrates that standard Diffusion Transformers can effectively serve as a foundation for precipitation nowcasting without requiring overly specialized architectures. By incorporating a dynamics-aware noise prior and reinforcement learning, the proposed NowcastDiT model achieves state-of-the-art performance in both perceptual quality and meteorological skill on major benchmarks. The study highlights the scalability of general-purpose diffusion models when adapted with domain-specific constraints for spatiotemporal forecasting tasks.
NowcastDiT: Diffusion Transformers are Effective Precipitation Nowcasters
arXiv:2609.37038v1 Announce Type: new Abstract: Precipitation nowcasting demands accurate short-term forecasts under strong spatiotemporal variability. Diffusion models are well suited to modeling complex precipitation distributions, yet existing approaches often introduce increasingly specialized designs, leaving the capability of a standard diffusion architecture underexplored. We show that a standard Diffusion Transformer already provides a simple and scalable foundation for precipitation nowcasting, with domain-specific requirements accommodated naturally within its design space. Based on this principle, we develop NowcastDiT and instantiate this flexibility through two complementary adaptations: a dynamics-aware noise prior for temporally coherent forecasts, and end-to-end reinforcement learning with timestep-aware rewards for meteorological skill. Experiments on SEVIR and MRMS benchmarks show that NowcastDiT achieves state-of-the-art performance in both perceptual quality and meteorological skill. These results suggest that standard DiT can serve as an effective foundation for precipitation nowcasting.
Abstract:Precipitation nowcasting demands accurate short-term forecasts under strong spatiotemporal variability. Diffusion models are well suited to modeling complex precipitation distributions, yet existing approaches often introduce increasingly specialized designs, leaving the capability of a standard diffusion architecture underexplored. We show that a standard Diffusion Transformer already provides a simple and scalable foundation for precipitation nowcasting, with domain-specific requirements accommodated naturally within its design space. Based on this principle, we develop NowcastDiT and instantiate this flexibility through two complementary adaptations: a dynamics-aware noise prior for temporally coherent forecasts, and end-to-end reinforcement learning with timestep-aware rewards for meteorological skill. Experiments on SEVIR and MRMS benchmarks show that NowcastDiT achieves state-of-the-art performance in both perceptual quality and meteorological skill. These results suggest that standard DiT can serve as an effective foundation for precipitation nowcasting.
Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
arXiv:2609.37038 [cs.LG]
(or arXiv:2609.37038v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2609.37038 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 233 words
Related coverage
ORBIT-2 based Weather and Climate Downscaling and Downscaled Global Forecasts on AMD Instinct
AMD · Jun 8, 2026
Learning to Refine: Spectral-Decoupled Iterative Refinement Framework for Precipitation Nowcasting
arXiv · Jun 3, 2026
Diffusion-Based Refinement for Kilometer-Scale Probabilistic Precipitation Nowcasting
arXiv · Sep 1, 2026
Problems of Proliferation
LessWrong · Sep 29, 2026