Data-driven medium-range weather prediction with a Resnet pretrained on climate simulations: A new model for WeatherBench

TL;DR

This paper introduces a deep residual neural network pretrained on climate simulations for medium-range weather prediction, outperforming previous models and showing promise for higher-resolution forecasts.

Contribution

It presents a novel Resnet model pretrained on climate data, improving weather forecast accuracy and interpretability compared to existing data-driven approaches.

Findings

Outperforms previous WeatherBench submissions

Comparable in skill to physical models at similar resolution

Shows potential for higher-resolution forecasts

Abstract

Numerical weather prediction has traditionally been based on physical models of the atmosphere. Recently, however, the rise of deep learning has created increased interest in purely data-driven medium-range weather forecasting with first studies exploring the feasibility of such an approach. To accelerate progress in this area, the WeatherBench benchmark challenge was defined. Here, we train a deep residual convolutional neural network (Resnet) to predict geopotential, temperature and precipitation at 5.625 degree resolution up to 5 days ahead. To avoid overfitting and improve forecast skill, we pretrain the model using historical climate model output before fine-tuning on reanalysis data. The resulting forecasts outperform previous submissions to WeatherBench and are comparable in skill to a physical baseline at similar resolution. We also analyze how the neural network creates its predictions and find that, with some exceptions, it is compatible with physical reasoning. Finally, we perform scaling experiments to estimate the potential skill of data-driven approaches at higher resolutions.