Preprints
https://doi.org/10.5194/egusphere-2024-1449
https://doi.org/10.5194/egusphere-2024-1449
30 May 2024
 | 30 May 2024

Robustness of the long short-term memory network in rainfall-runoff prediction improved by the water balance constraint

Qiang Li and Tongtiegang Zhao

Abstract. While the water balance constraint is fundamental to catchment hydrological models, there is yet no consensus on its role in the long short-term memory (LSTM) network. This paper is concentrated on the part that this constraint plays in the robustness of the LSTM network for rainfall-runoff prediction. Specifically, numerical experiments are devised to examine the robustness of the LSTM and its architecturally mass-conserving variant (MC-LSTM); and the Explainable Artificial Intelligence (XAI) is employed to interrogate how this constraint affects the robustness of the LSTM in learning rainfall-runoff relationships. Based on the Catchment Attributes and Meteorology for Large-sample Studies (CAMELS) dataset, the LSTM, MC-LSTM and EXP-HYDRO models are trained under various amounts of training data and different seeds of parameter initialization over 531 catchments, leading to 95,580 (3×6×10×531) tests. Through large-sample tests, the results show that incorporating the water balance constraint into the LSTM improves the robustness, while the improvement tends to decrease as the amount of training data increases. Under 9 years’ training data, this constraint significantly enhances the robustness against data sparsity in 37 % (196 in 531) of the catchments and improves the robustness against parameter initialization in 73 % (386 in 531) of the catchments. In addition, it improves the robustness in learning rainfall-runoff relationships by increasing the median contribution of precipitation from 45.8 % to 47.3 %. These results point to the compensation effects between training data and process knowledge on the LSTM’s performance. Overall, the in-depth investigations facilitate insights into the use of the LSTM for rainfall-runoff prediction.

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this preprint. The responsibility to include appropriate place names lies with the authors.
Qiang Li and Tongtiegang Zhao

Status: final response (author comments only)

Comment types: AC – author | RC – referee | CC – community | EC – editor | CEC – chief editor | : Report abuse
  • RC1: 'Comment on egusphere-2024-1449', Anonymous Referee #1, 10 Jun 2024
    • AC1: 'Reply on RC1', Tongtiegang Zhao, 30 Jun 2024
  • RC2: 'Comment on egusphere-2024-1449', Anonymous Referee #2, 24 Jun 2024
    • AC2: 'Reply on RC2', Tongtiegang Zhao, 30 Jun 2024
    • AC3: 'Reply on RC2', Tongtiegang Zhao, 18 Jul 2024
Qiang Li and Tongtiegang Zhao
Qiang Li and Tongtiegang Zhao

Viewed

Total article views: 444 (including HTML, PDF, and XML)
HTML PDF XML Total Supplement BibTeX EndNote
319 97 28 444 39 15 13
  • HTML: 319
  • PDF: 97
  • XML: 28
  • Total: 444
  • Supplement: 39
  • BibTeX: 15
  • EndNote: 13
Views and downloads (calculated since 30 May 2024)
Cumulative views and downloads (calculated since 30 May 2024)

Viewed (geographical distribution)

Total article views: 482 (including HTML, PDF, and XML) Thereof 482 with geography defined and 0 with unknown origin.
Country # Views %
  • 1
1
 
 
 
 
Latest update: 14 Oct 2024
Download
Short summary
This paper focuses on the effect of the water balance constraint on the robustness of the long short-term memory (LSTM) network in learning rainfall-runoff relationships. Through large-sample tests, it is found that incorporating this constraint into the LSTM improves the robustness, while the improvement tends to decrease as the amount of training data increases. The results point to the compensation effects between training data and process knowledge on the LSTM’s performance.