Earth observation embeddings enhance weather downscaling accuracy

Researchers have demonstrated that Earth observation embeddings can serve as effective descriptors for probabilistic weather downscaling. By integrating these embeddings, derived from TESSERA data at 10m resolution, into a convolutional conditional neural process, they improved the accuracy of predicting 2m temperature and 10m wind speed. This method showed a 11.5% improvement in CRPS skill for temperature and 6.2% for wind speed across diverse climatic regions, even when using different input data like the Aurora AI forecasting model. AI

IMPACT This research suggests a new method for improving weather forecasting accuracy by leveraging Earth observation data with AI models.

RANK_REASON The cluster contains a research paper detailing a novel application of machine learning for weather prediction. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• arXiv
• Aurora AI forecasting model
• convolutional conditional neural process
• Earth observation embeddings
• ERA5
• tessera

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →