Improving data-driven global weather prediction using deep convolutional neural networks on a cubed sphere

Abstract

We present a significantly improved data-driven global weather forecasting framework using a deep convolutional neural network (CNN) to forecast several basic atmospheric variables on a global grid. New developments in this framework include an off-line volume-conservative mapping to a cubed-sphere grid, improvements to the CNN architecture and the minimization of the loss function over multiple steps in a prediction sequence. The cubed-sphere remapping minimizes the distortion on the cube faces on which convolution operations are performed and provides natural boundary conditions for padding in the CNN. Our improved model produces weather forecasts that are indefinitely stable and produce realistic weather patterns at lead times of several weeks and longer. For short-to medium-range forecasting, our model significantly outperforms persistence, climatology, and a coarse-resolution dynamical numerical weather prediction (NWP) model. Unsurprisingly, our forecasts are worse than those from a high-resolution state-of-the-art operational NWP system. Our data-driven model is able to learn to forecast complex surface temperature patterns from few input atmospheric state variables. On annual time scales, our model produces a realistic seasonal cycle driven solely by the prescribed variation in top-of-atmosphere solar forcing. Although it currently does not compete with operational weather forecasting models, our data-driven CNN executes much faster than those models, suggesting that machine learning could prove to be a valuable tool for large-ensemble forecasting.Plain Language Summary Recent work has begun to explore building global weather prediction models using only machine learning techniques trained on large amounts of atmospheric data. We develop a vastly improved machine learning algorithm capable of operating like traditional weather models and predicting several fundamental atmospheric variables, including near-surface temperature. While our model does not yet compete with the state-of-the-art in numerical weather prediction, it computes realistic forecasts that perform well and execute extremely quickly, offering a potential avenue for future developments in probabilistic weather forecasting.

Data‐Driven Medium‐Range Weather Prediction With a Resnet Pretrained on Climate Simulations: A New Model for WeatherBench

“…Weyn et al. (2020) trained the model over two time steps. This, however, quickly becomes very computationally expensive for large model such as the ones in this study.…”

“…As an additional baseline, here we include the work by Weyn et al. (2020) who trained an neural network to predict Z500 and T850. Their model is iterative, that is, it consists of a sequence of 6 h forecasts.…”

Ensemble Methods for Neural Network‐Based Weather Forecasts

“…Work in this direction has been done by Scher (2018) and Scher and Messori (2019) on simplified systems, and Dueben and Bauer (2018); Weyn et al. (2019, 2020); Rasp and Thuerey (2020) on reanalysis data. However, machine‐learning forecast skill in the medium range (∼3–14 days) is typically much poorer than what operational NWP models achieve.…”

A Machine Learning‐Based Global Atmospheric Forecast Model

“…To the best of our knowledge, the only similar prior attempts were those by Scher (2018) and Scher and Messori (2019), but they trained their three‐dimensional multivariate ML model on data that were produced by low‐resolution numerical model simulations. In addition, Dueben and Bauer (2018) and Weyn et al, 2019 (2020) designed ML models to predict two‐dimensional, horizontal fields of select atmospheric state variables. Similar to our verification strategy, they also verified the ML forecasts against reanalysis data.…”