A nonlinear rainfall-runoff model using neural network technique: example in fractured porous media

One of the more advanced approaches for simulating groundwater flow in karstic and fractured porous media is the combination of a linear and a nonlinear model. The paper presents an attempt to determine outflow influencing parameters in order to simulate aquifer outflow. Our approach in this study i...

Full description

Saved in:
Bibliographic Details
Published in:Mathematical and computer modelling 2003-05, Vol.37 (9), p.1047-1061
Main Authors: Lallahem, S., Mania, J.
Format: Article
Language:English
Subjects:
Applied sciences
Artificial intelligence
Artificial neural networks
Chalk media
Computer science
control theory
systems
Connectionism. Neural networks
Exact sciences and technology
Flows through porous media
Fluid dynamics
Fractured media
Fundamental areas of phenomenology (including applications)
Groundwater flow
Karst aquifer
Mathematical analysis
Mathematics
Nonhomogeneous flows
Partial differential equations
Physics
Sciences and techniques of general use
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:One of the more advanced approaches for simulating groundwater flow in karstic and fractured porous media is the combination of a linear and a nonlinear model. The paper presents an attempt to determine outflow influencing parameters in order to simulate aquifer outflow. Our approach in this study is to create a productive interaction system between expert, mathematical model, MERO, and artificial neural networks (ANNs). The proposed method is especially suitable for the problem of large-scale and long-term simulation. In the present project, the first objective is to determine aquifer outflow influencing parameters by the use of MERO model, which gave a good results in a fissured and chalky media, and then introduce these parameters in neural network (NN). To determine outflow influencing parameters, we propose to test the NN under fourth different external input scenarios. The second objective is to investigate the effect of temporal information by taking current and past data sets. The good found results reveal the merit of ANNs-MERO combination and specifically multilayer perceptron (MLP) models. This methodology provided that the network with lower, lag and number hidden layer, consistently produced better performance.
ISSN:0895-7177
1872-9479
DOI:10.1016/S0895-7177(03)00117-1