Aquifer Parameter, Feed Forward Back Propagation, Radial Basis Function, Recurrent Artificial Neural Network, Inverse Modeling, Finite Element Method
The present study focuses on the unexplored area of application of artificial neural network in groundwater hydrology. Three models, each based on artificial neural networks, are applied for prediction of zonal transmissivity. These techniques can be considered as black box models that can predict output values for given range of input values after establishing an acceptable relation which is obtained by training the system. The study is based on coupling of Finite Element Method (FEM) - Artificial Neural Network (ANN) model, which serve as forward (FEM) and inverse (ANN) models. An inverse technique using ANN is considered for estimating parameters of groundwater system. A synthetic problem is examined for two different scenarios, the first one involving the sink and/or sources terms and the second, without these. Inverse model is applied to estimate transmissivity of various zones (64 data pairs involving nodal head and node coordinates) of aquifer domain. The performance evaluation criteria are shown to have good agreement between true transmissivity and estimated transmissivity, both at training and testing stages.
BYU ScholarsArchive Citation
"Artificial Neural Network Application on Estimation of Aquifer Transmissivity,"
Journal of Spatial Hydrology: Vol. 8
, Article 1.
Available at: https://scholarsarchive.byu.edu/josh/vol8/iss2/1