Physics-Assisted and Topology-Informed Deep Learning for Weather Prediction
Keywords: global weather prediction, physics-assisted deep learning, topology-informed deep learning
TL;DR: We develop a novel physics-assisted and topology-informed deep learning model for weather prediction, attributing the weather evolution to advection process and the Earth-atmosphere interaction.
Abstract: Weather prediction is crucial for decision-making in various social and economic sectors. The classical numerical weather prediction methods cannot incorporate the historical observations to enhance the underlying physical models, whereas the existing data-driven, deep learning-based weather prediction methods disregard either the $\textbf{physics}$ of the weather evolution or the $\textbf{topology}$ of the Earth's surface. In light of these disadvantages, we develop PASSAT, a novel Physics-ASSisted And Topology-informed deep learning model for weather prediction. PASSAT attributes the weather evolution to two key factors: (i) the advection process that can be characterized by the advection equation and the Navier-Stokes equation; (ii) the Earth-atmosphere interaction that is difficult to both model and calculate. PASSAT also takes the topology of the Earth's surface into consideration, other than simply treating it as a plane. Therefore, PASSAT numerically solves the advection equation and the Navier-Stokes equation on the spherical manifold, utilizes a spherical graph neural network to capture the Earth-atmosphere interaction, and generates the initial velocity fields that are critical to solving the advection equation from the same spherical graph neural network. These building blocks constitute a deep learning-based, $\textbf{physics-assisted}$ and $\textbf{topology-informed}$ weather prediction model. In the $5.625^\circ$-resolution ERA5 data set, PASSAT outperforms both the state-of-the-art deep learning-based weather prediction models and the operational numerical weather prediction model IFS T42.
Primary Area: applications to physical sciences (physics, chemistry, biology, etc.)
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 8824
Loading