CoDiCast: Conditional Diffusion Model for Global Weather Prediction with Uncertainty Quantification

TL;DR

CoDiCast introduces a conditional diffusion framework for global weather forecasting that conditions on recent observations to generate probabilistic future states. By combining a pre-trained encoder with cross-attention-augmented denoising and a diffusion-based generative process, it delivers competitive accuracy while providing ensemble-based uncertainty quantification at a fraction of the computational cost of traditional NWP. Extensive experiments on ERA5/WeatherBench demonstrate improved RMSE and ACC compared with ML-based baselines, with uncertainty bands that grow logically with lead time. The approach offers a practical balance between predictive performance, quantified uncertainty, and runtime efficiency, and is released with open-source code for broader adoption.

Abstract

Accurate weather forecasting is critical for science and society. Yet, existing methods have not managed to simultaneously have the properties of high accuracy, low uncertainty, and high computational efficiency. 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, most existing machine learning-based weather prediction (MLWP) approaches are efficient and accurate. Nevertheless, they are deterministic and cannot capture the uncertainty of weather forecasting. In this work, we propose CoDiCast, a conditional diffusion model to generate accurate global weather prediction, while achieving uncertainty quantification with ensemble forecasts and modest computational cost. The key idea is to simulate a conditional version of the reverse denoising process in diffusion models, which starts from pure Gaussian noise to generate realistic weather scenarios for a future time point. Each denoising step is conditioned on observations from the recent past. Ensemble forecasts are achieved by repeatedly sampling from stochastic Gaussian noise to represent uncertainty quantification. CoDiCast is trained on a decade of ERA5 reanalysis data from the European Centre for Medium-Range Weather Forecasts (ECMWF). Experimental results demonstrate that our approach outperforms several existing data-driven methods in accuracy. Our conditional diffusion model, CoDiCast, can generate 6-day global weather forecasts, at 6-hour steps and $5.625^\circ$ latitude-longitude resolution, for over 5 variables, in about 12 minutes on a commodity A100 GPU machine with 80GB memory. The open-souced code is provided at https://github.com/JimengShi/CoDiCast.

Figures (15)

• Figure 1: Deterministic vs Probabilistic.
• Figure 2: Framework of CoDiCast for global weather forecast. The superscript $T$ and the subscript $N$ denote the time point and iteration step of adding/denoising noise. $H$ and $W$ represent the height (#latitude) and width (#longitude) of grid data. $C$ is the number of variables of interest.
• Figure 3: Autoencoder structure.
• Figure 4: Attention-based denoiser structure.
• Figure 5: Forecast with confidence intervals.