Saved in:
Bibliographic Details
Main Authors: Du, Haoxing, Kim, Lyna, Creus-Costa, Joan, Michaels, Jack, Shetty, Anuj, Hutchinson, Todd, Riedel, Christopher, Dean, John
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2503.22235
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909555985743872
author Du, Haoxing
Kim, Lyna
Creus-Costa, Joan
Michaels, Jack
Shetty, Anuj
Hutchinson, Todd
Riedel, Christopher
Dean, John
author_facet Du, Haoxing
Kim, Lyna
Creus-Costa, Joan
Michaels, Jack
Shetty, Anuj
Hutchinson, Todd
Riedel, Christopher
Dean, John
contents We present WeatherMesh-3 (WM-3), an operational transformer-based global weather forecasting system that improves the state of the art in both accuracy and computational efficiency. We introduce the following advances: 1) a latent rollout that enables arbitrary-length predictions in latent space without intermediate encoding or decoding; and 2) a modular architecture that flexibly utilizes mixed-horizon processors and encodes multiple real-time analyses to create blended initial conditions. WM-3 generates 14-day global forecasts at 0.25-degree resolution in 12 seconds on a single RTX 4090. This represents a >100,000-fold speedup over traditional NWP approaches while achieving superior accuracy with up to 37.7% improvement in RMSE over operational models, requiring only a single consumer-grade GPU for deployment. We aim for WM-3 to democratize weather forecasting by providing an accessible, lightweight model for operational use while pushing the performance boundaries of machine learning-based weather prediction.
format Preprint
id arxiv_https___arxiv_org_abs_2503_22235
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle WeatherMesh-3: Fast and accurate operational global weather forecasting
Du, Haoxing
Kim, Lyna
Creus-Costa, Joan
Michaels, Jack
Shetty, Anuj
Hutchinson, Todd
Riedel, Christopher
Dean, John
Machine Learning
Artificial Intelligence
We present WeatherMesh-3 (WM-3), an operational transformer-based global weather forecasting system that improves the state of the art in both accuracy and computational efficiency. We introduce the following advances: 1) a latent rollout that enables arbitrary-length predictions in latent space without intermediate encoding or decoding; and 2) a modular architecture that flexibly utilizes mixed-horizon processors and encodes multiple real-time analyses to create blended initial conditions. WM-3 generates 14-day global forecasts at 0.25-degree resolution in 12 seconds on a single RTX 4090. This represents a >100,000-fold speedup over traditional NWP approaches while achieving superior accuracy with up to 37.7% improvement in RMSE over operational models, requiring only a single consumer-grade GPU for deployment. We aim for WM-3 to democratize weather forecasting by providing an accessible, lightweight model for operational use while pushing the performance boundaries of machine learning-based weather prediction.
title WeatherMesh-3: Fast and accurate operational global weather forecasting
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2503.22235