WeatherBench: A Benchmark Data Set for Data‐Driven Weather Forecasting

• Published in: Journal of advances in modeling earth systems Vol. 12; no. 11
• Main Authors: Rasp, Stephan, Dueben, Peter D., Scher, Sebastian, Weyn, Jonathan A., Mouatadid, Soukayna, Thuerey, Nils
• Format: Journal Article
• Language: English
• Published: Washington John Wiley & Sons, Inc 01.11.2020; American Geophysical Union (AGU)
• Subjects: Algorithms; artificial intelligence; benchmark; Data; Datasets; Deep learning; Evaluation; Forecasting; Forecasting data; Global weather; Intercomparison; Machine learning; Neural networks; NWP; Regularization methods; Statistical methods; Variables; Weather forecasting; Weather patterns
• ISSN: 1942-2466; 1942-2466
• DOI: 10.1029/2020MS002203

Data‐driven approaches, most prominently deep learning, have become powerful tools for prediction in many domains. A natural question to ask is whether data‐driven methods could also be used to predict global weather patterns days in advance. First studies show promise but the lack of a common data set and evaluation metrics make intercomparison between studies difficult. Here we present a benchmark data set for data‐driven medium‐range weather forecasting (specifically 3–5 days), a topic of high scientific interest for atmospheric and computer scientists alike. We provide data derived from the ERA5 archive that has been processed to facilitate the use in machine learning models. We propose simple and clear evaluation metrics which will enable a direct comparison between different methods. Further, we provide baseline scores from simple linear regression techniques, deep learning models, as well as purely physical forecasting models. The data set is publicly available at https://github.com/pangeo‐data/WeatherBench and the companion code is reproducible with tutorials for getting started. We hope that this data set will accelerate research in data‐driven weather forecasting. Plain Language Summary WeatherBench provides a new benchmark to test data‐driven approaches to weather forecasting. Traditional weather models are based on the discretized equations governing the atmosphere. They perform very well for many tasks but are still found lacking for some others. Data‐driven approaches, such as deep learning, directly learn from the best available observations and could potentially produce better forecasts. In this paper, we define a benchmark task—predicting pressure and temperature across the globe 3 and 5 days ahead—which will hopefully lead to progress in data‐driven weather prediction and foster collaboration across disciplines. Key Points Benchmarks with strong baselines are a key ingredient for rapid progress on a problem Here, we define a benchmark for data‐driven global, medium‐range weather prediction The data are processed for convenient use in machine learning models, and a quickstart guide is provided