Deep learning rainfall-runoff predictions of extreme events

This paper investigates the hypothesis that deep learning models may not be reliable in extrapolating extreme events.

Abstract

The most accurate rainfall-runoff predictions are currently based on deep learning. There is a concern among hydrologists that data-driven models based on deep learning may not be reliable in extrapolation or for predicting extreme events. This study tests that hypothesis using Long Short-Term Memory networks (LSTMs) and an LSTM variant that is architecturally constrained to conserve mass. The LSTM (and the mass-conserving LSTM variant) remained relatively accurate in predicting extreme (high return-period) events compared to both a conceptual model (the Sacramento Model) and a process-based model (US National Water Model), even when extreme events were not included in the training period. Adding mass balance constraints to the data-driven model (LSTM) reduced model skill during extreme events.

Paper

Frame, J. M., Kratzert, F., Klotz, D., Gauch, M., Shalev, G., Gilon, O., Qualls, L. M., Gupta, H. V., and Nearing, G. S.: Deep learning rainfall–runoff predictions of extreme events, Hydrol. Earth Syst. Sci., 26, 3377–3392, https://doi.org/10.5194/hess-26-3377-2022, 2022.

Code

All experiments were made with the NeuralHydrology Python library. The exact snapshot for reproducing the results can be found in this GitHub repository.

Citation

@Article{frame2021extreme,
author = {Frame, J. and Kratzert, F. and Klotz, D. and Gauch, M. and Shelev, G. and Gilon, O. and Qualls, L. M. and Gupta, H. V. and Nearing, G. S.},
title = {Deep learning rainfall-runoff predictions of extreme events},
journal = {Hydrology and Earth System Sciences Discussions},
volume = {2021},
year = {2021},
pages = {1--20},
doi = {10.5194/hess-2021-423}
}