HURRI-GAN：基于生成对抗网络的飓风偏差校正方法——超越气象站
摘要
美国东部和南部沿海地区受到严重风暴事件的影响,导致人员伤亡和财产损失。准确预测飓风surge和风力影响对于减轻部分影响至关重要,例如及时准备疏散及其他措施。物理仿真模型如ADCIRC水动力学模型运行在高性能计算资源上,是一种精细的工具,随着计算网格分辨率的提高,其预测结果越来越准确。然而,这些模型的一个主要缺点是生成非常高分辨率结果所需的计算时间显著,这可能不符合应急响应人员的近实时需求。本文介绍的HURRI-GAN是一种 novel AI-driven approach that augments the results produced by physical simulation models using time series generative adversarial networks (TimeGAN) to compensate for systemic errors of the physical model, thus reducing the necessary mesh size and runtime without loss in forecasting accuracy. 我们present first results in extrapolating model bias corrections for the spatial regions beyond the positions of the water level gauge stations. 本文所呈现的结果表明,该方法可以准确地在水位测站位置之外的目标位置生成偏差校正。模型的性能,如由低于0.05的root mean squared error (RMSE)值所示,突显了其在生成准确外推数据方面的能力。将HURRI-GAN生成的校正应用于ADCIRC模型的水位,结果在大多数测试测站上均实现了overall prediction改进。
引用
@article{arxiv.2603.06649,
title = {HURRI-GAN: A Novel Approach for Hurricane Bias-Correction Beyond Gauge Stations using Generative Adversarial Networks},
author = {Noujoud Nadera and Hadi Majed and Stefanos Giaremis and Rola El Osta and Clint Dawson and Carola Kaiser and Hartmut Kaiser},
journal= {arXiv preprint arXiv:2603.06649},
year = {2026}
}
备注
18 pages, 6 figures