Preprints
https://doi.org/10.5194/hess-2021-423
https://doi.org/10.5194/hess-2021-423

  18 Aug 2021

18 Aug 2021

Review status: this preprint is currently under review for the journal HESS.

Deep learning rainfall-runoff predictions of extreme events

Jonathan Frame1,2, Frederik Kratzert3, Daniel Klotz3, Martin Gauch3, Guy Shelev4, Oren Gilon4, Logan M. Qualls2, Hoshin V. Gupta5, and Grey S. Nearing6,7 Jonathan Frame et al.
  • 1National Water Center, National Oceanic and Atmospheric Administration, Tuscaloosa, AL, United States
  • 2University of Alabama, Tuscaloosa, AL, United States
  • 3LIT AI Lab & Institute for Machine Learning, Johannes Kepler University, Linz, Austria
  • 4Google Research, Tel Aviv, Israel
  • 5The University of Arizona, Tucson, AZ, United States
  • 6Google Research, Mountain View, CA, United States
  • 7University of California Davis, Department of Land, Air & Water Resources, Davis, CA, United States

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.

Jonathan Frame et al.

Status: open (until 16 Oct 2021)

Comment types: AC – author | RC – referee | CC – community | EC – editor | CEC – chief editor | : Report abuse
  • CC1: 'Comment on hess-2021-423', John Ding, 20 Aug 2021 reply
  • RC1: 'Comment on hess-2021-423', Anonymous Referee #1, 05 Sep 2021 reply

Jonathan Frame et al.

Data sets

CAMELS return period analysis Jonathan M. Frame https://doi.org/10.4211/hs.c7739f47e2ca4a92989ec34b7a2e78dd

Model code and software

Model results analysis Jonathan M. Frame https://github.com/jmframe/mclstm_2021_extrapolate/tree/main/results

NeuralHydrology Frederick Kratzert https://github.com/neuralhydrology/neuralhydrology

Code for calibrating SAC-SMA Grey S. Nearing https://github.com/Upstream-Tech/SACSMA-SNOW17

Jonathan Frame et al.

Viewed

Total article views: 985 (including HTML, PDF, and XML)
HTML PDF XML Total BibTeX EndNote
711 265 9 985 4 4
  • HTML: 711
  • PDF: 265
  • XML: 9
  • Total: 985
  • BibTeX: 4
  • EndNote: 4
Views and downloads (calculated since 18 Aug 2021)
Cumulative views and downloads (calculated since 18 Aug 2021)

Viewed (geographical distribution)

Total article views: 908 (including HTML, PDF, and XML) Thereof 908 with geography defined and 0 with unknown origin.
Country # Views %
  • 1
1
 
 
 
 
Latest update: 19 Sep 2021
Download
Short summary
The most accurate rainfall-runoff predictions are currently based on deep learning. There is a concern among hydrologists that deep learning models may not be reliable in extrapolation or for predicting extreme events. This study tests that hypothesis. The deep learning models remained relatively accurate in predicting extreme events compared traditional models, even when extreme events are not included in the training set.