Kilometer-Scale AI Downscaling of Atlantic Hurricanes with Generative Ensembles
Abstract
This study presents an AI-based dynamical downscaling system for Tropical Cyclones (TCs). The system incorporates an AI-based limited-area model that downscales 3-hourly low-resolution boundary forcings into hourly high-resolution fields autoregressively, and a diffusion model that converts the outputs into ensembles of hazard-relevant variables. The system is trained on the regridded CONUS404 data with ERA5 forcings, and is evaluated on 20 TCs in 2020--2024. Verification shows stable downscaling performance across Atlantic hurricane seasons, with energy spectra closely matching the CONUS404 reference. The system is also verified to produce skillful TC-relevant weather extremes, largely improved over a deterministic AI baseline. The system performs well with forcing data from other models (GDAS/FNL) and can produce detailed eyewall, rainband, and landfall structures in TC case studies. The study provides a good example of how AI-based dynamical downscaling systems can be designed to resolve small-scale extreme weather events.
Related
- Less is more: error-distance scaling relation for data-efficient kilometer-scale downscaling of extreme heat
- Efficient Analog-Initialized Latent Transport for Spatially Coherent Probabilistic Downscaling from Global to Kilometer Scales
- SwAIther-Precip: Lead-Time-Aware Bias Correction Enables Kilometer-Scale Downscaling of Global AI Precipitation Forecasts over Switzerland
- Stress-Testing Dynamical and Generative Downscaling Using Subseasonal Extreme Precipitation Forecasts
- Generative modelling for multivariate spatio-temporal downscaling of monthly climate fields
BibTeX
@misc{sha2026kilometerscale,
title = {{Kilometer-Scale AI Downscaling of Atlantic Hurricanes with Generative Ensembles}},
author = {Yingkai Sha and Talea L. Mayo and Ethan D. Gutmann and Lulin Xue and Andrew Newman},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.02034},
eprint = {2609.02034},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.02034},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Kilometer-Scale AI Downscaling of Atlantic Hurricanes with Generative Ensembles
AU - Yingkai Sha
AU - Talea L. Mayo
AU - Ethan D. Gutmann
AU - Lulin Xue
AU - Andrew Newman
PY - 2026
DA - 2026/09/02
PB - arXiv
DO - 10.48550/arxiv.2609.02034
UR - https://arxiv.org/abs/2609.02034
AB - This study presents an AI-based dynamical downscaling system for Tropical Cyclones (TCs). The system incorporates an AI-based limited-area model that downscales 3-hourly low-resolution boundary forcings into hourly high-resolution fields autoregressively, and a diffusion model that converts the outputs into ensembles of hazard-relevant variables. The system is trained on the regridded CONUS404 data with ERA5 forcings, and is evaluated on 20 TCs in 2020--2024. Verification shows stable downscaling performance across Atlantic hurricane seasons, with energy spectra closely matching the CONUS404 reference. The system is also verified to produce skillful TC-relevant weather extremes, largely improved over a deterministic AI baseline. The system performs well with forcing data from other models (GDAS/FNL) and can produce detailed eyewall, rainband, and landfall structures in TC case studies. The study provides a good example of how AI-based dynamical downscaling systems can be designed to resolve small-scale extreme weather events.
ER -