Deep learning for weather prediction: Forecasting globally-gridded 500-hPa geopotential heights on short- to medium-range time scales

We develop elementary weather prediction models using deep convolutional neural networks (CNNs) trained on past weather data to forecast fundamental meteorological fields on a global grid with no explicit knowledge about physical processes. At forecast lead times up to 3 days, CNNs trained to predict only 500‐hPa geopotential height easily outperform persistence, climatology, and the dynamics‐based barotropic vorticity model, but do not beat an operational full‐physics weather prediction model. These CNNs are capable of forecasting significant changes in the intensity of weather systems, which is notable because this is beyond the capability of the fundamental dynamical equation that relies solely on 500‐hPa data, the barotropic‐vorticity equation. Our best‐performing CNNs do a good job of capturing the climatology and annual variability of 500‐hPa heights and are capable of forecasting realistic atmospheric states at lead times of 14 days. With the application of suitable boundary conditions, the potential use of CNNs for medium-range to subseasonal forecasting is explored. Additionally, the potential efficiency of CNNs might make them attractive for ensemble seasonal forecasting.