Assessment of runoff via precipitation using neural networks: watershed modelling for developing environment in arid region

Samantaray Sandeep, . and Sahoo Abinash, . and Ghose Dillip Kumar, . Assessment of runoff via precipitation using neural networks: watershed modelling for developing environment in arid region. pp. 2245-2263. ISSN 2231-8526

This work describes the application of three different neural network (i) Back propagation neural network (BPNN) (ii) Layer Recurrent Neural Network (LRNN) and (iii) Radial Basis Fewer Network (RBFN) model to predict runoff. Here two scenarios were considered for developing the models. Scenario 1 exclusive of evapotranspiration and Scenario 2 with evapotranspiration are considered for experiencing the impact on runoff. Performance indicators entailed Scenario 2 performed best as compared to Scenario 1. Two watersheds Loisingha and Saintala were considered for study. In Loisingha watershed LRNN performed best with architecture 4-3-1 following tangential sigmoid transfer function. At Saintala both LRNN and BPNN performed in parallel with small deviation of prediction and LRNN performed best among three networks with model architecture 4-2-1 using Log-sig transfer function for predicting runoff.

Full text available from: Official URL: http://www.pertanika.upm.edu.my/resources/files/Pe...

Abstract

This work describes the application of three different neural network (i) Back propagation neural network (BPNN) (ii) Layer Recurrent Neural Network (LRNN) and (iii) Radial Basis Fewer Network (RBFN) model to predict runoff. Here two scenarios were considered for developing the models. Scenario 1 exclusive of evapotranspiration and Scenario 2 with evapotranspiration are considered for experiencing the impact on runoff. Performance indicators entailed Scenario 2 performed best as compared to Scenario 1. Two watersheds Loisingha and Saintala were considered for study. In Loisingha watershed LRNN performed best with architecture 4-3-1 following tangential sigmoid transfer function. At Saintala both LRNN and BPNN performed in parallel with small deviation of prediction and LRNN performed best among three networks with model architecture 4-2-1 using Log-sig transfer function for predicting runoff.

[error in script]

Item Type:	Article
AGROVOC Term:	Watersheds
AGROVOC Term:	Arid zones
AGROVOC Term:	Watershed management
AGROVOC Term:	Runoff
AGROVOC Term:	Neural networks
AGROVOC Term:	Evapotranspiration
AGROVOC Term:	Precipitation
AGROVOC Term:	Temperature
AGROVOC Term:	Rainfall
AGROVOC Term:	Water resources
Depositing User:	Mr. AFANDI ABDUL MALEK
Last Modified:	24 Apr 2025 00:55
URI:	http://webagris.upm.edu.my/id/eprint/9464

Actions (login required)

View Item