Estimation of Groundwater Levels With Surface Observations via Genetic Programming

Abstract

Surface water levels alone are indicators of both surface and groundwater storage in which the surface water and groundwater are highly interactive; such situations exist in most parts of Florida. Forecasting groundwater-level fluctuations by means of easily measured surface water levels using a groundwater-surface water model is an important requirement for planning conjunctive use in any basin. This article investigates the potential of artificial intelligence (AI) approaches in forecasting groundwater-level fluctuations in an aquifer using the measured water levels of two lakes in North Central Florida along with monthly averaged precipitation and evaporation. Relationships among lake levels, groundwater levels, rainfall, and evaporation were determined using different AI approaches-namely a multi-layer perceptron, radial basis neural network, multi-gene genetic programming, and multi-linear and multi-nonlinear regression models. A comparison of the results revealed that the proposed multi-gene genetic programming model produced more accurate predictions than those of the other approaches.