A Variational U-Net for Weather Forecasting
In one sentence. Variational U-Net combining VAE probabilistic modelling with U-Net detail recovery placed third in the Weather4cast satellite forecasting challenge.
Abstract
Not only can discovering patterns and insights from atmospheric data enable more accurate weather predictions, but it may also provide valuable information to help tackle climate change. Weather4cast is an open competition that aims to evaluate machine learning algorithms' capability to predict future atmospheric states. Here, we describe our third-place solution to Weather4cast. We present a novel Variational U-Net that combines a Variational Autoencoder's ability to consider the probabilistic nature of data with a U-Net's ability to recover fine-grained details. This solution is an evolution from our fourth-place solution to Traffic4cast 2020 with many commonalities, suggesting its applicability to vastly different domains, such as weather and traffic.
Labels
- Topic
- Nowcasting
- Type
- New model or method
- Method
- GAN or VAE, CNN / U-Net
- Time scale
- 0–6 h
- Domain
- Regional
- Data
- Satellite
- Status
- Research
- Identifiers
- arXiv:2111.03476 OpenAlex:W3213235731
- Added
- 2026-10-22
Labels and summary are generated by Claude and reviewed weekly. Wrong? Edit this entry.