Technical note: Deep learning for creating surrogate models of precipitation in Earth system models

Theodore Weber, Austin Corotan, Brian Hutchinson, Ben Kravitz, Robert Link

Atmospheric chemistry and physics · 26 February 2020

In one sentence. A CNN surrogate trained on CanESM2 monthly precipitation outperforms persistence and stays skillful over 120-month rollouts, with scheduled sampling and preforecasting documented.

Abstract

Abstract. We investigate techniques for using deep neural networks to produce surrogate models for short term climate forecasts. A convolutional neural network is trained on 97 years of monthly precipitation output from the 1pctCO2 run (the CO2 concentration increases by 1 % per year) simulated by the CanESM2 Earth System Model. The neural network clearly outperforms a persistence forecast and does not show substantially degraded performance even when the forecast length is extended to 120 months. The model is prone to underpredicting precipitation in areas characterized by intense precipitation events. Scheduled sampling (forcing the model to gradually use its own past predictions rather than ground truth) is essential for avoiding amplification of early forecasting errors. However, the use of scheduled sampling also necessitates preforecasting (generating forecasts prior to the first forecast date) to obtain adequate performance for the first few prediction time steps. We document the training procedures and hyperparameter optimization process for researchers who wish to extend the use of neural networks in developing surrogate models.

Labels

Labels and summary are generated by Claude and reviewed weekly. Wrong? Edit this entry.