Estimation of Hydraulic Conductivity and Its Uncertainty from Grain-Size Data Using GLUE and Artificial Neural Networks

Abstract

Various approaches exist to relate saturated hydraulic conductivity (K s ) to grain-size data. Most methods use a single grain-size parameter and hence omit the information encompassed by the entire grain-size distribution. This study compares two data-driven modelling methods-multiple linear regression and artificial neural networks-that use the entire grain-size distribution data as input for K s prediction. Besides the predictive capacity of the methods, the uncertainty associated with the model predictions is also evaluated, since such information is important for stochastic groundwater flow and contaminant transport modelling.Artificial neural networks (ANNs) are combined with a generalised likelihood uncertainty estimation (GLUE) approach to predict K s from grain-size data. The resulting GLUE-ANN hydraulic conductivity predictions and associated uncertainty estimates are compared with those obtained from the multiple linear regression models by a leave-one-out cross-validation. The GLUE-ANN ensemble prediction proved to be slightly better than multiple linear regression. The prediction uncertainty, however, was reduced by half an order of magnitude on average, and decreased at most by an order of magnitude. This demonstrates that the proposed method outperforms classical data-driven modelling techniques. Moreover, a comparison with methods from the literature demonstrates the importance of site-specific calibration. The data set used for this purpose originates mainly from unconsolidated sandy sediments of the Neogene aquifer, northern Belgium. The proposed predictive models are developed for 173 grain-size K s -pairs. Finally, an application with the optimised models is presented for a borehole lacking K s data.

Experimental and statistical study of saturated hydraulic conductivity and relations with other soil properties of a desert soil

“…() used PCA to identify relations between K , bulk density and porosity, Rogiers et al . () used it for estimating K from particle‐size data and Wang et al . () used it to identify soil properties affecting K in loessial soils of China.…”

“…Consequently, it is used to reduce the dimensionality of the data and to determine the strength of the relations among these variables (Jolliffe, 2002). For instance, Duffera et al (2007) used PCA to identify relations between K, bulk density and porosity, Rogiers et al (2012) used it for estimating K from particle-size data and Wang et al (2013) used it to identify soil properties affecting K in loessial soils of China.…”

Simulation of the Spatial Distribution of Hydraulic Conductivity in Porous Media through Different Methods

“…Vienken and Dietrich performed a comparative analysis to identify the application conditions of seven formulas [12]. Rogiers et al used particle size distribution to estimate hydraulic conductivity through artificial neural networks and estimated the uncertainty [13]. The distribution of hydraulic conductivity can also be obtained by these methods.…”

Minimising Backbreak at the Dewan Cement Limestone Quarry Using an Artificial Neural Network

“…Empirical methods and traditional modelling techniques such as multivariate regression fail to derive the cause of backbreak in cases where it is affected by numerous parameters nonlinearly. Artificial intelligence has been diversely applied in earth sciences recently, using fuzzy logic (Demicco & Klir, 2004;Muhammad & Glass, 2011), neural networks (Bonaventura et al, 2017;Chatterjee et al, 2010;Izadi et al, 2017;Rogiers et al, 2012;Roslin & Esterle, 2016;Muhammad et al, 2014), and neuro-fuzzy modelling techniques (Cherkassky et al, 2006;Kar et al, 2014;Valdés & Bonham-Carter, 2006;Yegireddi & Uday Bhaskar, 2009;Yurdakul et al, 2014;Zoveidavianpoor et al, 2013). Recently, several researchers have solved backbreak problems through applying neural networks (Jang & Topal, 2013;Monjezi & Dehghani, 2008;Monjezi et al, 2013;Saadat et al, 2014;Sayadi et al, 2013;Ebrahimi et al, 2016), neuro-fuzzy techniques (Ghasemi et al, 2016), stochastic optimisation (Sari et al, 2013), and machine learning techniques (Khandelwal & Monjezi, 2012;Mohammadnejad et al, 2013).…”