CoDiCast: Conditional Diffusion Model for Weather Prediction with Uncertainty Quantification
Keywords: Diffusion Model, Weather Prediction, Uncertainty Quantification
TL;DR: This paper proposes a conditional diffusion model for global weather prediction (CoDiCast), which achieves a critical tradeoff between high accuracy, high efficiency and low uncertainty.
Abstract: Accurate weather forecasting is critical for science and society. Yet, existing methods have not demonstrated high accuracy, low uncertainty, and high computational efficiency simultaneously. On one hand, to quantify the uncertainty in weather predictions, the strategy of ensemble forecast (i.e., generating a set of diverse predictions) is often employed. However, traditional ensemble numerical weather prediction (NWP) is computationally intensive. On the other hand, even though most existing machine learning-based weather prediction (MLWP) approaches are efficient and accurate, they are deterministic and cannot capture the uncertainty of weather forecasting. To tackle these challenges, we propose $\texttt{CoDiCast}$, a conditional diffusion model to generate accurate global weather prediction, while achieving uncertainty quantification and modest computational cost. The key idea behind the prediction task is to generate realistic weather scenarios at a $\textit{future}$ time point, conditioned on observations from the $\textit{recent past}$. Due to the probabilistic nature of diffusion models, they can be properly applied to capture the uncertainty of weather predictions. Therefore, we accomplish uncertainty quantifications by repeatedly sampling from stochastic Gaussian noise for each initial weather state and running the denoising process multiple times. Experimental results demonstrate that $\texttt{CoDiCast}$ outperforms several existing MLWP methods in accuracy, and is faster than NWP models in the inference speed. $\texttt{CoDiCast}$ can generate 3-day global weather forecasts, at 6-hour steps and $5.625^\circ$ latitude-longitude resolutions, for over 5 variables, in about 12 minutes on a commodity A100 GPU machine with 80GB memory. The anonymous code is provided at \url{https://anonymous.4open.science/r/CoDiCast/}.
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: 7409
Loading