Composite Classifier-Free Guidance for Multi-Modal Conditioning in Wind Dynamics Super-Resolution

2025-12-13

AI Summary

Wind field super-resolution demands both high fidelity and computational efficiency, yet existing diffusion models struggle to effectively leverage multi-modal meteorological conditioning inputs with ≥10 channels, leading to guidance failure and reconstruction artifacts. To address this, we propose Composite Classifier-Free Guidance (CCFG), the first extension of classifier-free guidance to multi-condition input settings—compatible with standard CFG training pipelines without architectural modification. Building upon CCFG, we introduce WindDM, a diffusion model integrating multi-modal meteorological encoding, the CCFG mechanism, and industrial-grade training strategies. On wind dynamics super-resolution, WindDM achieves state-of-the-art performance: it significantly improves fidelity over conventional CFG, reduces computational cost by 1000× compared to numerical simulation methods, and simultaneously delivers high accuracy, efficiency, and engineering practicality.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Multimodal Learning
• Cognitive Modeling & Cognitive Systems: Computational Creativity

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Various weather modelling problems (e.g., weather forecasting, optimizing turbine placements, etc.) require ample access to high-resolution, highly accurate wind data. Acquiring such high-resolution wind data, however, remains a challenging and expensive endeavour. Traditional reconstruction approaches are typically either cost-effective or accurate, but not both. Deep learning methods, including diffusion models, have been proposed to resolve this trade-off by leveraging advances in natural image super-resolution. Wind data, however, is distinct from natural images, and wind super-resolvers often use upwards of 10 input channels, significantly more than the usual 3-channel RGB inputs in natural images. To better leverage a large number of conditioning variables in diffusion models, we present a generalization of classifier-free guidance (CFG) to multiple conditioning inputs. Our novel composite classifier-free guidance (CCFG) can be dropped into any pre-trained diffusion model trained with standard CFG dropout. We demonstrate that CCFG outputs are higher-fidelity than those from CFG on wind super-resolution tasks. We present WindDM, a diffusion model trained for industrial-scale wind dynamics reconstruction and leveraging CCFG. WindDM achieves state-of-the-art reconstruction quality among deep learning models and costs up to $1000 imes$ less than classical methods.

Problem

• Develops composite classifier-free guidance for multi-modal conditioning
• Addresses high-resolution wind data acquisition cost-accuracy trade-off
• Enhances diffusion models for wind dynamics super-resolution tasks

Innovation

• Composite classifier-free guidance for multi-modal conditioning
• WindDM diffusion model for industrial-scale wind reconstruction
• Cost reduction up to 1000x compared to classical methods