Advancing Parsimonious Deep Learning Weather Prediction Using the HEALPix Mesh

Author:

Karlbauer Matthias1ORCID,Cresswell‐Clay Nathaniel2ORCID,Durran Dale R.23ORCID,Moreno Raul A.2ORCID,Kurth Thorsten4ORCID,Bonev Boris4ORCID,Brenowitz Noah3ORCID,Butz Martin V.1ORCID

Affiliation:

1. Department of Computer Science Neuro‐Cognitive Modeling Group University of Tübingen Tübingen Germany

2. Department of Atmospheric Sciences University of Washington Seattle WA USA

3. NVIDIA Corporation Seattle WA USA

4. NVIDIA Switzerland AG Zürich Switzerland

Abstract

AbstractWe present a parsimonious deep learning weather prediction model to forecast seven atmospheric variables with 3‐hr time resolution for up to 1‐year lead times on a 110‐km global mesh using the Hierarchical Equal Area isoLatitude Pixelization (HEALPix). In comparison to state‐of‐the‐art (SOTA) machine learning (ML) weather forecast models, such as Pangu‐Weather and GraphCast, our DLWP‐HPX model uses coarser resolution and far fewer prognostic variables. Yet, at 1‐week lead times, its skill is only about 1 day behind both SOTA ML forecast models and the SOTA numerical weather prediction model from the European Center for Medium‐Range Weather Forecasts. We report several improvements in model design, including switching from the cubed sphere to the HEALPix mesh, inverting the channel depth of the U‐Net, and introducing gated recurrent units (GRU) on each level of the U‐Net hierarchy. The consistent east‐west orientation of all cells on the HEALPix mesh facilitates the development of location‐invariant convolution kernels that successfully propagate weather patterns across the globe without requiring separate kernels for the polar and equatorial faces of the cube sphere. Without any loss of spectral power after the first 2 days, the model can be unrolled autoregressively for hundreds of steps into the future to generate realistic states of the atmosphere that respect seasonal trends, as showcased in 1‐year simulations.

Funder

Deutsche Forschungsgemeinschaft

Office of Naval Research Global

Publisher

American Geophysical Union (AGU)

Reference67 articles.

1. Delving deeper into convolutional networks for learning video representations;Ballas N.;arXiv preprint arXiv:1511.06432,2015

2. Relational inductive biases, deep learning, and graph networks;Battaglia P. W.;arXiv preprint arXiv:1806.01261,2018

3. The quiet revolution of numerical weather prediction

同舟云学术

1.学者识别学者识别

2.学术分析学术分析

3.人才评估人才评估

"同舟云学术"是以全球学者为主线,采集、加工和组织学术论文而形成的新型学术文献查询和分析系统,可以对全球学者进行文献检索和人才价值评估。用户可以通过关注某些学科领域的顶尖人物而持续追踪该领域的学科进展和研究前沿。经过近期的数据扩容,当前同舟云学术共收录了国内外主流学术期刊6万余种,收集的期刊论文及会议论文总量共计约1.5亿篇,并以每天添加12000余篇中外论文的速度递增。我们也可以为用户提供个性化、定制化的学者数据。欢迎来电咨询!咨询电话:010-8811{复制后删除}0370

www.globalauthorid.com

TOP

Copyright © 2019-2024 北京同舟云网络信息技术有限公司
京公网安备11010802033243号  京ICP备18003416号-3