[2604.07928] Generative 3D Gaussian Splatting for Arbitrary-ResolutionAtmospheric Downscaling and Forecasting

Computer Science > Computer Vision and Pattern Recognition arXiv:2604.07928 (cs) [Submitted on 9 Apr 2026 (v1), last revised 10 Apr 2026 (this version, v2)] Title:Generative 3D Gaussian Splatting for Arbitrary-ResolutionAtmospheric Downscaling and Forecasting Authors:Tao Han, Zhibin Wen, Zhenghao Chen, Fenghua Lin, Junyu Gao, Song Guo, Lei Bai View a PDF of the paper titled Generative 3D Gaussian Splatting for Arbitrary-ResolutionAtmospheric Downscaling and Forecasting, by Tao Han and 6 other authors View PDF HTML (experimental) Abstract:While AI-based numerical weather prediction (NWP) enables rapid forecasting, generating high-resolution outputs remains computationally demanding due to limited multi-scale adaptability and inefficient data representations. We propose the 3D Gaussian splatting-based scale-aware vision transformer (GSSA-ViT), a novel framework for arbitrary-resolution forecasting and flexible downscaling of high-dimensional atmospheric fields. Specifically, latitude-longitude grid points are treated as centers of 3D Gaussians. A generative 3D Gaussian prediction scheme is introduced to estimate key parameters, including covariance, attributes, and opacity, for unseen samples, improving generalization and mitigating overfitting. In addition, a scale-aware attention module is designed to capture cross-scale dependencies, enabling the model to effectively integrate information across varying downscaling ratios and support continuous resolution adaptation. To o...