Preprints
https://doi.org/10.5194/hess-2022-245
https://doi.org/10.5194/hess-2022-245
 
08 Aug 2022
08 Aug 2022
Status: this preprint is currently under review for the journal HESS.

The suitability of differentiable, learnable hydrologic models for ungauged regions and climate change impact assessment

Dapeng Feng1, Hylke Beck2, Kathryn Lawson1, and Chaopeng Shen1 Dapeng Feng et al.
  • 1Civil and Environmental Engineering, The Pennsylvania State University
  • 2Joint Research Centre of the European Commission, Ispra, Italy

Abstract. Differentiable, learnable process-based hydrologic models (abbreviated as δ or delta models) with regionalized parameterization pipelines were recently shown to provide daily streamflow prediction performance that closely approach state-of-the-art long short-term memory (LSTM) deep networks. Meanwhile, δ models provide a full suite of diagnostic physical variables and guaranteed mass conservation. Due to their physical constraints, we hypothesize that they are suitable for making extrapolated predictions. Here, we ran experiments to test (1) their ability to extrapolate to regions far from streamflow gauges; and (2) their ability to make credible projections of long-term (decadal-scale) change trends. We evaluate the models based on daily hydrograph metrics (Nash-Sutcliffe model efficiency coefficient, etc.), as well as projected decadal streamflow trends. The results show that, for spatial interpolation (test in randomly sampled ungauged basins, or PUB), δ models have mixed comparisons with LSTM, presenting better trends for annual mean flow and high flow but slightly worse for low flow. For spatial extrapolation (test in regionally held out basins, or PUR, representing a highly data-scarce scenario), δ models’ advantages in mean and high flows are more prominent. In addition, an untrained variable, evapotranspiration, retained good seasonality even for extrapolated cases. δ models’ parameterization pipeline produces parameter fields that maintain remarkably stable spatial patterns even in highly data-scarce scenarios, which explains their robustness. Combined with their interpretability and ability to assimilate multi-source observations, δ models are strong candidates for regional and global scale hydrologic simulations for climate change impact assessment.

Dapeng Feng et al.

Status: open (until 26 Oct 2022)

Comment types: AC – author | RC – referee | CC – community | EC – editor | CEC – chief editor | : Report abuse
  • CC1: 'Comment on hess-2022-245', John Ding, 12 Aug 2022 reply

Dapeng Feng et al.

Dapeng Feng et al.

Viewed

Total article views: 741 (including HTML, PDF, and XML)
HTML PDF XML Total BibTeX EndNote
543 189 9 741 4 4
  • HTML: 543
  • PDF: 189
  • XML: 9
  • Total: 741
  • BibTeX: 4
  • EndNote: 4
Views and downloads (calculated since 08 Aug 2022)
Cumulative views and downloads (calculated since 08 Aug 2022)

Viewed (geographical distribution)

Total article views: 671 (including HTML, PDF, and XML) Thereof 671 with geography defined and 0 with unknown origin.
Country # Views %
  • 1
1
 
 
 
 
Latest update: 28 Sep 2022
Download
Short summary
Hybrid models (we call δ models) that embrace the fundamental learning capability of AI but can explain all the physical processes can be powerful. In this paper we assess how they perform when applied in regions not in the training data. δ models rivaled the accuracy of state-of-the-art AI models under the data-dense scenario and even surpassed them for the data-sparse one. They generalize well due to the physical structure. δ models could be ideal candidates for global hydrologic assessments.