Prediction of aquifer properties from vertical electrical sounding data using artificial neural network: case study of Ibadan Metropolis, South-western Nigeria

Authors

Abstract

Adequate estimates of aquifer properties are of utmost importance for proper management of groundwater resources. In an effort to provide alternative way of estimating aquifer properties at minimum cost, Artificial Neural Network (ANN) model was generated for the prediction of Transmissivity (T) and Hydraulic Conductivity (K). Six hundred and thirty eight vertical electrical soundings data were acquired and interpreted to obtain Dar-Zarrouk parameters. The diagnostic relationship between the K values measured in reference wells and electrical soundings data was combined with Dar-Zarrouk parameters to estimate T and K. Transverse resistance (R), thickness (h), resistivity (ρ), K and T were subjected to ANN analysis using SPSS software and Python. The results showed that R ranges from 40 to 21552 Ωm2, h ranges from 1.4 to 40.0 m, ρ ranges from 4 to 754 Ωm. K varies from 0.004 to 0.800 m/day and T varies from 0.04 to 18.20 m2/day within the study area. ANN model was able to predict K and T values with an accuracy ranging from 97 to 99%. RMSE values for the prediction ranged from 0.063 to 0.250. The ANN model generated was able to predict K and T of the aquifer from geo-electrical data at minimum cost.

Keywords

NRIAG Journal of Astronomy and Geophysics
Volume 9, Issue 1
January 2020
Pages 598-612

Files

Share

How to cite

Statistics