Score-based diffusion nowcasting of GOES imagery

2025-05-15

AI Summary

To address subgrid parameterization errors in 0–3-hour cloud and precipitation nowcasting and the ambiguity inherent in early machine learning approaches, this work introduces, for the first time in meteorological forecasting, a residual-corrected diffusion model (CorrDiff) integrated with a latent diffusion model (LDM). Leveraging GOES infrared satellite image sequences, the method explicitly models physical cloud evolution—including cloud generation, dissipation, and convective initiation—beyond the restrictive pure-advection assumption. It performs spatiotemporally consistent score matching and residual correction in the latent space, enabling well-calibrated ensemble forecasting. Experiments demonstrate that CorrDiff reduces RMSE by 1–2 K relative to U-Net and persistence baselines, markedly improving fine-scale structural fidelity. Furthermore, ensemble spread exhibits strong correlation with forecast error, empirically validating its capability for reliable uncertainty quantification.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Computer Vision: Diffusion Models for Vision
• Reasoning under Uncertainty: Relational Probabilistic Models

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Clouds and precipitation are important for understanding weather and climate. Simulating clouds and precipitation with traditional numerical weather prediction is challenging because of the sub-grid parameterizations required. Machine learning has been explored for forecasting clouds and precipitation, but early machine learning methods often created blurry forecasts. In this paper we explore a newer method, named score-based diffusion, to nowcast (zero to three hour forecast) clouds and precipitation. We discuss the background and intuition of score-based diffusion models - thus providing a starting point for the community - while exploring the methodology's use for nowcasting geostationary infrared imagery. We experiment with three main types of diffusion models: a standard score-based diffusion model (Diff); a residual correction diffusion model (CorrDiff); and a latent diffusion model (LDM). Our results show that the diffusion models are able to not only advect existing clouds, but also generate and decay clouds, including convective initiation. These results are surprising because the forecasts are initiated with only the past 20 mins of infrared satellite imagery. A case study qualitatively shows the preservation of high resolution features longer into the forecast than a conventional mean-squared error trained U-Net. The best of the three diffusion models tested was the CorrDiff approach, outperforming all other diffusion models, the traditional U-Net, and a persistence forecast by one to two kelvin on root mean squared error. The diffusion models also enable out-of-the-box ensemble generation, which shows skillful calibration, with the spread of the ensemble correlating well to the error.

Problem

• Nowcasting clouds and precipitation using score-based diffusion models
• Overcoming blurry forecasts in machine learning weather prediction
• Generating high-resolution cloud features with minimal input data

Innovation

• Score-based diffusion models for nowcasting
• Residual correction diffusion (CorrDiff) outperforms others
• Generates and decays clouds with high resolution