LangPrecip: Language-Aware Multimodal Precipitation Nowcasting

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Ling, Xudong, Li, Chaorong, Huang, Tianxi, Dong, Qian, Duan, Guiduo
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866916012355485696
author Ling, Xudong
Li, Chaorong
Huang, Tianxi
Dong, Qian
Duan, Guiduo
author_facet Ling, Xudong
Li, Chaorong
Huang, Tianxi
Dong, Qian
Duan, Guiduo
contents Short-term precipitation nowcasting is an inherently uncertain and under-constrained spatiotemporal forecasting problem, especially for rapidly evolving and extreme weather events. Existing generative approaches rely primarily on visual conditioning, leaving future motion weakly constrained and ambiguous. We propose a language-aware multimodal nowcasting framework(LangPrecip) that treats meteorological text as a semantic motion constraint on precipitation evolution. By formulating nowcasting as a semantically constrained trajectory generation problem under the Rectified Flow paradigm, our method enables efficient and physically consistent integration of textual and radar information in latent space.We further introduce LangPrecip-160k, a large-scale multimodal dataset with 160k paired radar sequences and motion descriptions. Experiments on Swedish and MRMS datasets show consistent improvements over state-of-the-art methods, achieving over 60 \% and 19\% gains in heavy-rainfall CSI at an 80-minute lead time.
format Preprint
id arxiv_https___arxiv_org_abs_2512_22317
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle LangPrecip: Language-Aware Multimodal Precipitation Nowcasting
Ling, Xudong
Li, Chaorong
Huang, Tianxi
Dong, Qian
Duan, Guiduo
Machine Learning
Artificial Intelligence
Computer Vision and Pattern Recognition
Short-term precipitation nowcasting is an inherently uncertain and under-constrained spatiotemporal forecasting problem, especially for rapidly evolving and extreme weather events. Existing generative approaches rely primarily on visual conditioning, leaving future motion weakly constrained and ambiguous. We propose a language-aware multimodal nowcasting framework(LangPrecip) that treats meteorological text as a semantic motion constraint on precipitation evolution. By formulating nowcasting as a semantically constrained trajectory generation problem under the Rectified Flow paradigm, our method enables efficient and physically consistent integration of textual and radar information in latent space.We further introduce LangPrecip-160k, a large-scale multimodal dataset with 160k paired radar sequences and motion descriptions. Experiments on Swedish and MRMS datasets show consistent improvements over state-of-the-art methods, achieving over 60 \% and 19\% gains in heavy-rainfall CSI at an 80-minute lead time.
title LangPrecip: Language-Aware Multimodal Precipitation Nowcasting
topic Machine Learning
Artificial Intelligence
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2512.22317