Application of a recurrent neural network to rainfall-runoff modeling

Kuo lin Hsu, Hoshin Vijai Gupta, Soroosh Sorooshian

Research output: Chapter in Book/Report/Conference proceedingConference contribution

10 Citations (Scopus)

Abstract

The lumped daily rainfall-runoff process for the Leaf River Basin in Mississippi was modeled using two different Artificial Neural Network (ANN) model structures. Our results indicate that both structures, the popular Three Layer Feedforward Neural Network (TLFNN) and the Recurrent Neural Network (RNN), perform well. However, the TLFNN requires trial-and-error testing to identify the appropriate number of time-delayed input variables to the model. Further, it is not suitable for distributed watershed modeling; i.e., when distributed precipitation information (multiple gages or radar images) is available. The RNN structure provides a representation of the dynamic internal feedbacks loops in the system, thereby eliminating the need for lagged inputs and resulting in a reduction in the number of network weights (and hence training time). The suitability of RNN's for distributed watershed modeling is discussed.

Original languageEnglish (US)
Title of host publicationProceedings of the Annual Water Resources Planning and Management Conference
EditorsD.H. Merritt
PublisherASCE
Pages68-73
Number of pages6
StatePublished - 1997
EventProceedings of the 1997 24th Annual Water Resources Planning and Management Conference - Houston, TX, USA
Duration: Apr 6 1997Apr 9 1997

Other

OtherProceedings of the 1997 24th Annual Water Resources Planning and Management Conference
CityHouston, TX, USA
Period4/6/974/9/97

Fingerprint

rainfall-runoff modeling
Recurrent neural networks
Feedforward neural networks
Watersheds
Runoff
Rain
watershed
Model structures
Catchments
artificial neural network
modeling
Gages
gauge
Radar
river basin
Rivers
radar
runoff
Neural networks
Feedback

ASJC Scopus subject areas

  • Civil and Structural Engineering
  • Water Science and Technology

Cite this

Hsu, K. L., Gupta, H. V., & Sorooshian, S. (1997). Application of a recurrent neural network to rainfall-runoff modeling. In D. H. Merritt (Ed.), Proceedings of the Annual Water Resources Planning and Management Conference (pp. 68-73). ASCE.

Application of a recurrent neural network to rainfall-runoff modeling. / Hsu, Kuo lin; Gupta, Hoshin Vijai; Sorooshian, Soroosh.

Proceedings of the Annual Water Resources Planning and Management Conference. ed. / D.H. Merritt. ASCE, 1997. p. 68-73.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Hsu, KL, Gupta, HV & Sorooshian, S 1997, Application of a recurrent neural network to rainfall-runoff modeling. in DH Merritt (ed.), Proceedings of the Annual Water Resources Planning and Management Conference. ASCE, pp. 68-73, Proceedings of the 1997 24th Annual Water Resources Planning and Management Conference, Houston, TX, USA, 4/6/97.
Hsu KL, Gupta HV, Sorooshian S. Application of a recurrent neural network to rainfall-runoff modeling. In Merritt DH, editor, Proceedings of the Annual Water Resources Planning and Management Conference. ASCE. 1997. p. 68-73
Hsu, Kuo lin ; Gupta, Hoshin Vijai ; Sorooshian, Soroosh. / Application of a recurrent neural network to rainfall-runoff modeling. Proceedings of the Annual Water Resources Planning and Management Conference. editor / D.H. Merritt. ASCE, 1997. pp. 68-73
@inproceedings{321718953902494a8b40ce1ec09943de,
title = "Application of a recurrent neural network to rainfall-runoff modeling",
abstract = "The lumped daily rainfall-runoff process for the Leaf River Basin in Mississippi was modeled using two different Artificial Neural Network (ANN) model structures. Our results indicate that both structures, the popular Three Layer Feedforward Neural Network (TLFNN) and the Recurrent Neural Network (RNN), perform well. However, the TLFNN requires trial-and-error testing to identify the appropriate number of time-delayed input variables to the model. Further, it is not suitable for distributed watershed modeling; i.e., when distributed precipitation information (multiple gages or radar images) is available. The RNN structure provides a representation of the dynamic internal feedbacks loops in the system, thereby eliminating the need for lagged inputs and resulting in a reduction in the number of network weights (and hence training time). The suitability of RNN's for distributed watershed modeling is discussed.",
author = "Hsu, {Kuo lin} and Gupta, {Hoshin Vijai} and Soroosh Sorooshian",
year = "1997",
language = "English (US)",
pages = "68--73",
editor = "D.H. Merritt",
booktitle = "Proceedings of the Annual Water Resources Planning and Management Conference",
publisher = "ASCE",

}

TY - GEN

T1 - Application of a recurrent neural network to rainfall-runoff modeling

AU - Hsu, Kuo lin

AU - Gupta, Hoshin Vijai

AU - Sorooshian, Soroosh

PY - 1997

Y1 - 1997

N2 - The lumped daily rainfall-runoff process for the Leaf River Basin in Mississippi was modeled using two different Artificial Neural Network (ANN) model structures. Our results indicate that both structures, the popular Three Layer Feedforward Neural Network (TLFNN) and the Recurrent Neural Network (RNN), perform well. However, the TLFNN requires trial-and-error testing to identify the appropriate number of time-delayed input variables to the model. Further, it is not suitable for distributed watershed modeling; i.e., when distributed precipitation information (multiple gages or radar images) is available. The RNN structure provides a representation of the dynamic internal feedbacks loops in the system, thereby eliminating the need for lagged inputs and resulting in a reduction in the number of network weights (and hence training time). The suitability of RNN's for distributed watershed modeling is discussed.

AB - The lumped daily rainfall-runoff process for the Leaf River Basin in Mississippi was modeled using two different Artificial Neural Network (ANN) model structures. Our results indicate that both structures, the popular Three Layer Feedforward Neural Network (TLFNN) and the Recurrent Neural Network (RNN), perform well. However, the TLFNN requires trial-and-error testing to identify the appropriate number of time-delayed input variables to the model. Further, it is not suitable for distributed watershed modeling; i.e., when distributed precipitation information (multiple gages or radar images) is available. The RNN structure provides a representation of the dynamic internal feedbacks loops in the system, thereby eliminating the need for lagged inputs and resulting in a reduction in the number of network weights (and hence training time). The suitability of RNN's for distributed watershed modeling is discussed.

UR - http://www.scopus.com/inward/record.url?scp=0030645841&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0030645841&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:0030645841

SP - 68

EP - 73

BT - Proceedings of the Annual Water Resources Planning and Management Conference

A2 - Merritt, D.H.

PB - ASCE

ER -