IntegratingWeather Foundation Model and Satellite to Enable Fine-Grained Solar Irradiance Forecasting

2026-03-16

AI Summary

Existing approaches struggle to simultaneously achieve high spatiotemporal resolution and long-range forecasting capability, limiting the accuracy of day-ahead solar irradiance prediction. This work proposes Baguan-solar, a two-stage multimodal fusion framework: in the first stage, it leverages the global weather foundation model Baguan together with high-resolution geostationary satellite imagery to predict intermediate meteorological variables—such as cloud cover—continuously across day and night; in the second stage, it derives kilometer-scale 24-hour solar irradiance forecasts from these predictions. By effectively integrating fine-grained satellite cloud structures with large-scale meteorological constraints, the method significantly enhances the capture of cloud-induced irradiance transients. Evaluated over East Asia, Baguan-solar reduces RMSE by 16.08% compared to strong baselines including ECMWF IFS, the original Baguan, and SolarSeer, and has been operationally deployed in an eastern province of China.

Technology Category

• Computer Vision: Multi-modal Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Machine Learning: Multimodal Learning

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Web Mining and Content Analysis: Large pretrained models with web data
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Accurate day-ahead solar irradiance forecasting is essential for integrating solar energy into the power grid. However, it remains challenging due to the pronounced diurnal cycle and inherently complex cloud dynamics. Current methods either lack fine-scale resolution (e.g., numerical weather prediction, weather foundation models) or degrade at longer lead times (e.g., satellite extrapolation). We propose Baguan-solar, a two-stage multimodal framework that fuses forecasts from Baguan, a global weather foundation model, with high-resolution geostationary satellite imagery to produce 24- hour irradiance forecasts at kilometer scale. Its decoupled two-stage design first forecasts day-night continuous intermediates (e.g., cloud cover) and then infers irradiance, while its modality fusion jointly preserves fine-scale cloud structures from satellite and large-scale constraints from Baguan forecasts. Evaluated over East Asia using CLDAS as ground truth, Baguan-solar outperforms strong baselines (including ECMWF IFS, vanilla Baguan, and SolarSeer), reducing RMSE by 16.08% and better resolving cloud-induced transients. An operational deployment of Baguan-solar has supported solar power forecasting in an eastern province in China, since July 2025. Our code is accessible at https://github.com/DAMO-DI-ML/Baguansolar. git.

Problem

• solar irradiance forecasting
• fine-grained prediction
• cloud dynamics
• numerical weather prediction
• satellite imagery

Innovation

• weather foundation model
• satellite imagery fusion
• solar irradiance forecasting
• multimodal framework
• fine-grained prediction