Spatiotemporally Coherent Probabilistic Generation of Weather from Climate

2024-12-19

AI Summary

To address the challenge of generating high-resolution, long-term, multivariate weather sequences for climate model downscaling, this paper introduces, for the first time, score-matching diffusion models into meteorological generation tasks, proposing a conditional score distillation diffusion framework. The method conditions on coarse-resolution global climate fields and integrates uncertainty-aware multi-trajectory sampling with climate-field guidance, trained exclusively on high-resolution reanalysis data. Experiments demonstrate that the generated decadal-scale weather sequences significantly outperform existing methods in spatiotemporal continuity, physical consistency, and statistical fidelity, while strictly aligning with the coarse-resolution climate drivers. This enables robust regional climate impact assessment and risk-informed decision-making. The core contribution lies in achieving physically constrained, probabilistic, long-range coherent multivariate downscaling—marking a critical advance in interpretable, uncertainty-quantified weather sequence generation.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Generation

Application Category

• Web Mining and Content Analysis: Web data generation and simulation
• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Local climate information is crucial for impact assessment and decision-making, yet coarse global climate simulations cannot capture small-scale phenomena. Current statistical downscaling methods infer these phenomena as temporally decoupled spatial patches. However, to preserve physical properties, estimating spatio-temporally coherent high-resolution weather dynamics for multiple variables across long time horizons is crucial. We present a novel generative approach that uses a score-based diffusion model trained on high-resolution reanalysis data to capture the statistical properties of local weather dynamics. After training, we condition on coarse climate model data to generate weather patterns consistent with the aggregate information. As this inference task is inherently uncertain, we leverage the probabilistic nature of diffusion models and sample multiple trajectories. We evaluate our approach with high-resolution reanalysis information before applying it to the climate model downscaling task. We then demonstrate that the model generates spatially and temporally coherent weather dynamics that align with global climate output.

Problem

• Global Climate Information
• High-resolution Weather Prediction
• Local Climate Impact Assessment

Innovation

• Fractional Diffusion Model
• High-Resolution Weather Prediction
• Uncertainty Quantification