Journal cover Journal topic
Hydrology and Earth System Sciences An interactive open-access journal of the European Geosciences Union
Journal topic

Journal metrics

Journal metrics

  • IF value: 4.936 IF 4.936
  • IF 5-year value: 5.615 IF 5-year
    5.615
  • CiteScore value: 4.94 CiteScore
    4.94
  • SNIP value: 1.612 SNIP 1.612
  • IPP value: 4.70 IPP 4.70
  • SJR value: 2.134 SJR 2.134
  • Scimago H <br class='hide-on-tablet hide-on-mobile'>index value: 107 Scimago H
    index 107
  • h5-index value: 63 h5-index 63
Discussion papers
https://doi.org/10.5194/hess-2019-368
© Author(s) 2019. This work is distributed under
the Creative Commons Attribution 4.0 License.
https://doi.org/10.5194/hess-2019-368
© Author(s) 2019. This work is distributed under
the Creative Commons Attribution 4.0 License.

Submitted as: research article 02 Aug 2019

Submitted as: research article | 02 Aug 2019

Review status
This discussion paper is a preprint. It is a manuscript under review for the journal Hydrology and Earth System Sciences (HESS).

Benchmarking a Catchment-Aware Long Short-Term Memory Network (LSTM) for Large-Scale Hydrological Modeling

Frederik Kratzert1, Daniel Klotz1, Guy Shalev2, Günter Klambauer1, Sepp Hochreiter1,*, and Grey Nearing3,* Frederik Kratzert et al.
  • 1LIT AI Lab & Institute for Machine Learning, Johannes Kepler University Linz, Austria
  • 2Google Research
  • 3Department of Geological Sciences, University of Alabama, Tuscaloosa, AL United States
  • *These authors contributed equally to this work.

Abstract. Regional rainfall-runoff modeling is an old but still mostly out-standing problem in Hydrological Sciences. The problem currently is that traditional hydrological models degrade significantly in performance when calibrated for multiple basins together instead of for a single basin alone. In this paper, we propose a novel, data-driven approach using Long Short-Term Memory networks (LSTMs), and demonstrate that under a big data paradigm, this is not necessarily the case. By training a single LSTM model on 531 basins from the CAMELS data set using meteorological time series data and static catchment attributes, we were able to significantly improve performance compared to a set of several different hydrological benchmark models. Our proposed approach not only significantly outperforms hydrological models that were calibrated regionally but also achieves better performance than hydrological models that were calibrated for each basin individually. Furthermore, we propose an adaption to the standard LSTM architecture, which we call an Entity-Aware-LSTM (EA-LSTM), that allows for learning, and embedding as a feature layer in a deep learning model, catchment similarities. We show that this learned catchment similarity corresponds well with what we would expect from prior hydrological understanding.

Frederik Kratzert et al.
Interactive discussion
Status: open (until 27 Sep 2019)
Status: open (until 27 Sep 2019)
AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment
[Subscribe to comment alert] Printer-friendly Version - Printer-friendly version Supplement - Supplement
Frederik Kratzert et al.
Frederik Kratzert et al.
Viewed  
Total article views: 329 (including HTML, PDF, and XML)
HTML PDF XML Total BibTeX EndNote
244 81 4 329 4 7
  • HTML: 244
  • PDF: 81
  • XML: 4
  • Total: 329
  • BibTeX: 4
  • EndNote: 7
Views and downloads (calculated since 02 Aug 2019)
Cumulative views and downloads (calculated since 02 Aug 2019)
Viewed (geographical distribution)  
Total article views: 88 (including HTML, PDF, and XML) Thereof 87 with geography defined and 1 with unknown origin.
Country # Views %
  • 1
1
 
 
 
 
Cited  
Saved  
No saved metrics found.
Discussed  
No discussed metrics found.
Latest update: 24 Aug 2019
Publications Copernicus
Download
Short summary
A new approach for regional rainfall-runoff modeling using LSTM based models is presented and benchmarked against a range of well-known hydrological models. The approach significantly outperforms regionally calibrated hydrological models but also basin-wise calibrated models. Further more we propose an adaption of the LSTM that allows to extract the learned catchment understanding of the model and show that it matches our hydrology expert knowledge.
A new approach for regional rainfall-runoff modeling using LSTM based models is presented and...
Citation