Preprints
https://doi.org/10.5194/hess-2021-176
https://doi.org/10.5194/hess-2021-176

  14 Apr 2021

14 Apr 2021

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

Deep, Wide, or Shallow? Artificial Neural Network Topologies for Predicting Intermittent Flows

Farhang Forghanparast, Elma Annette Hernandez, and Venkatesh Uddameri Farhang Forghanparast et al.
  • TTU Water Resources Center, Department of Civil, Environmental and Construction Engineering, Texas Tech University

Abstract. Intermittent Rivers and Ephemeral Streams (IRES) comprise 60 % of all streams in the US and about 50 % of the streams worldwide. Furthermore, climate-driven changes are expected to force a shift towards intermittency in currently perennial streams. Most modeling studies have treated intermittent streamflows as a continuum. However, it is better to envision flow data of IRES as a “mixture-type”, comprised of both flow and no-flow regimes. It is therefore hypothesized that data-driven models with both classification and regression cells can improve the streamflow forecasting abilities in these streams. Deep and wide Artificial Neural Networks (ANNs) comprising of classification and regression cells were developed here by stacking them in series and parallel configurations. These deep and wide network architectures were compared against the commonly used single hidden layer ANNs (shallow), as a baseline, for modeling IRES flow series under the continuum assumption. New metrics focused on no-flow persistence and transitions between flow and no-flow states were formulated using contingency tables and Markov chain analysis. Nine IRES across the state of Texas, US, were used as a wide range of testbeds with different hydro-climatic characteristics. Model overfitting and the curse-of-dimensionality were reduced using extreme learning machines (ELM), and balancing training data using the synthetic minority oversampling technique (SMOTE), greedy learning and Least Absolute Shrinkage and Selection Operator (LASSO). The addition of classifier cells greatly improved the ability to distinguish between no-flow and flow states, in turn, improving the ability to capture no-flow persistence (dryness) and transitions to and from flow states (dryness initiation and cessation). The wide network topology provided better results when the focus was on capturing low flows and the deep topology did well in capturing extreme flows (zero and > 75th percentile).

Farhang Forghanparast et al.

Status: open (until 09 Jun 2021)

Comment types: AC – author | RC – referee | CC – community | EC – editor | CEC – chief editor | : Report abuse
  • RC1: 'Comment on hess-2021-176', Anonymous Referee #1, 07 May 2021 reply

Farhang Forghanparast et al.

Farhang Forghanparast et al.

Viewed

Total article views: 269 (including HTML, PDF, and XML)
HTML PDF XML Total BibTeX EndNote
221 42 6 269 1 1
  • HTML: 221
  • PDF: 42
  • XML: 6
  • Total: 269
  • BibTeX: 1
  • EndNote: 1
Views and downloads (calculated since 14 Apr 2021)
Cumulative views and downloads (calculated since 14 Apr 2021)

Viewed (geographical distribution)

Total article views: 248 (including HTML, PDF, and XML) Thereof 248 with geography defined and 0 with unknown origin.
Country # Views %
  • 1
1
 
 
 
 
Latest update: 07 May 2021
Download
Short summary
Intermittent flows are ecologically important yet poorly studied. New machine learning modeling methods and modeling evaluation schemes have been developed and tested over a wide range of climatic conditions to address this important knowledge gap in hydrology. The results demonstrate the superiority of these new algorithms to properly capture intermittent flow characteristics and evaluate no-flow persistence and flow regime transitions important for eco-hydrological assessments.