Prediction of spatial fluoride concentrations by a hybrid artificial neural network in complex aquifers

Document Type

Article

Publication Date

1-1-2019

Abstract

Due to the health risks by fluoride concentrations exceeding the permissible limits, an investigation is presented in this chapter to develop models for the prediction of spatial fluoride concentration in a complex aquifer. According to previous research the mean fluoride concentration in the study area was 2.8 mg/L and the maximum was 8.1 mg/L. Therefore the local population is exposed to health risks and hence this investigation. The chapter presents a modeling strategy based on the use of 10 different variations of artificial neural networks (ANN). The impact of the complexity of the aquifer system on the predictive model is investigated through demarcating the aquifer into basaltic, nonbasaltic and their interface zones; as well as comparing these results with a learning vector quantization (LVQ) neural network. A further hybrid ANN (HANN) model is developed that integrates LVQ and ANN capabilities and the results presented in the chapter show that HANN offers significant improvements.

Publication Source (Journal or Book title)

GIS and Geostatistical Techniques for Groundwater Science

First Page

269

Last Page

281

This document is currently not available here.

Share

COinS