Roy, Dilip Kumar (2018) Development of a sustainable groundwater management strategy and sequential compliance monitoring to control saltwater intrusion in coastal aquifers. PhD thesis, James Cook University.

Preview

PDF (Thesis) Download (3MB) | Preview

Abstract

The coastal areas of the world are characterized by high population densities, an abundance of food, and increased economic activities. These increasing human settlements, subsequent increases in agricultural developments and economic activities demand an increasing amount quantity of freshwater supplies to different sectors. Groundwater in coastal aquifers is one of the most important sources of freshwater supplies. Over exploitation of this coastal groundwater resource results in seawater intrusion and subsequent deterioration of groundwater quality in coastal aquifers. In addition, climate change induced sea level rise, in combination with the effect of excessive groundwater extraction, can accelerate the seawater intrusion. Adequate supply of good quality water to different sectors in coastal areas can be ensured by adoption of a proper management strategy for groundwater extraction. Optimal use of the coastal groundwater resource is one of the best management options, which can be achieved by employing a properly developed optimal groundwater extraction strategy. Coupled simulation-optimization (S-O) approaches are essential tools to obtain the optimal groundwater extraction patterns. This study proposes approaches for developing multiple objective management of coastal aquifers with the aid of barrier extraction wells as hydraulic control measure of saltwater intrusion in multilayered coastal aquifer systems. Therefore, two conflicting objectives of management policy are considered in this research, i.e. maximizing total groundwater extraction for advantageous purposes, and minimizing the total amount of water abstraction from barrier extraction wells. The study also proposes an adaptive management strategy for coastal aquifers by developing a three-dimensional (3-D) monitoring network design. The performance of the proposed methodologies is evaluated by using both an illustrative multilayered coastal aquifer system and a real life coastal aquifer study area.

Coupled S-O approach is used as the basic tool to develop a saltwater intrusion management model to obtain the optimal groundwater extraction rates from a combination of feasible solutions on the Pareto optimal front. Simulation of saltwater intrusion processes requires solution of density dependent coupled flow and solute transport numerical simulation models that are computationally intensive. Therefore, computational efficiency in the coupled S-O approach is achieved by using an approximate emulator of the accompanying physical processes of coastal aquifers. These emulators, often known as surrogate models or meta-models, can replace the computationally intensive numerical simulation model in a coupled S-O approach for achieving computational efficiency. A number of meta-models have been developed and compared in this study for integration with the optimization algorithm in order to develop saltwater intrusion management model. Fuzzy Inference System (FIS), Adaptive Neuro Fuzzy Inference System (ANFIS), Multivariate Adaptive Regression Spline (MARS), and Gaussian Process Regression (GPR) based meta-models are developed in the present study for approximating coastal aquifer responses to groundwater extraction. Properly trained and tested meta-models are integrated with a Controlled Elitist Multiple Objective Genetic Algorithm (CEMOGA) within a coupled S-O approach. In each iteration of the optimization algorithm, the meta-models are used to compute the corresponding salinity concentrations for a set of candidate pumping patterns generated by the optimization algorithm. Upon convergence, the non-dominated global optimal solutions are obtained as the Pareto optimal front, which represents a trade-off between the two conflicting objectives of the pumping management problem. It is observed from the solutions of the meta-model based coupled S-O approach that the considered meta-models are capable of producing a Pareto optimal set of solutions quite accurately. However, each meta-modelling approach has distinct advantages over the others when utilized within the integrated S-O approach.

Uncertainties in estimating complex flow and solute transport processes in coastal aquifers demand incorporation of the uncertainties related to some of the model parameters. Multidimensional heterogeneity of aquifer properties such as hydraulic conductivity, compressibility, and bulk density are considered as major sources of uncertainty in groundwater modelling system. Other sources of uncertainty are associated with spatial and temporal variability of hydrologic as well as human interventions, e.g. aquifer recharge and transient groundwater extraction patterns. Different realizations of these uncertain model parameters are obtained from different statistical distributions. FIS based meta-models are advanced to a Genetic Algorithm (GA) tuned hybrid FIS model (GA-FIS), to emulate physical processes of coastal aquifers and to evaluate responses of the coastal aquifers to groundwater extraction under groundwater parameter uncertainty. GA is used to tune the FIS parameters in order to obtain the optimal FIS structure. The GA-FIS models thus obtained are linked externally to the CEMOGA in order to derive an optimal pumping management strategy using the coupled S-O approach. The evaluation results show that the proposed saltwater intrusion management model is able to derive reliable optimal groundwater extraction strategies to control saltwater intrusion for the illustrative multilayered coastal aquifer system. The optimal management strategies obtained as solutions of GA-FIS based management models are shown to be reliable and accurate within the specified ranges of values for different realizations of uncertain groundwater parameters.

One of the major concerns of the meta-model based integrated S-O approach is the uncertainty associated with the meta-model predictions. These prediction uncertainties, if not addressed properly, may propagate to the optimization procedures, and may deteriorate the optimality of the solutions. A standalone meta-model, when used within an optimal management model, may result in the optimization routine producing actually suboptimal solutions that may undermine the optimality of the groundwater extraction strategies. Therefore, this study proposes an ensemble approach to address the prediction uncertainties of meta-models. Ensemble is an approach to assimilate multiple similar or different algorithms or base learners (emulators). The basic idea of ensemble lies in developing a more reliable and robust prediction tool that incorporates each individual emulator's unique characteristic in order to predict future scenarios. Each individual member of the ensemble contains different input -output mapping functions. Based on their own mapping functions, these individual emulators provide varied predictions on the response variable. Therefore, the combined prediction of the ensemble is likely to be less biased and more robust, reliable, and accurate than that of any of the individual members of the ensemble. Performance of the ensemble meta-models is evaluated using an illustrative coastal aquifer study area. The results indicate that the meta-model based ensemble modelling approach is able to provide reliable solutions for a multilayered coastal aquifer management problem.

Relative sea level rise, providing an additional saline water head at the seaside, has a significant impact on an increase in the salinization process of the coastal aquifers. Although excessive groundwater withdrawal is considered as the major cause of saltwater intrusion, relative sea level rise, in combination with the effect of excessive groundwater pumping, can exacerbate the already vulnerable coastal aquifers. This study incorporates the effects of relative sea level rise on the optimized groundwater extraction values for the specified management period. Variation of water concentrations in the tidal river and seasonal fluctuation of river water stage are also incorporated. Three meta-models are developed from the solution results of the numerical simulation model that simulates the coupled flow and solute transport processes in a coastal aquifer system. The results reveal that the proposed meta-models are capable of predicting density dependent coupled flow and solute transport patterns quite accurately. Based on the comparison results, the best meta-model is selected as a computationally cheap substitute of the simulation model in the coupled S-O based saltwater intrusion management model. The performance of the proposed methodology is evaluated for an illustrative multilayered coastal aquifer system in which the effect of climate change induced sea level rise is incorporated for the specified management period. The results show that the proposed saltwater intrusion management model provides acceptable, accurate, and reliable solutions while significantly improving computational efficiency in the coupled S-O methodology.

The success of the developed management strategy largely depends on how accurately the prescribed management policy is implemented in real life situations. The actual implementation of a prescribed management strategy often differs from the prescribed planned strategy due to various uncertainties in predicting the consequences, as well as practical constraints, including noncompliance with the prescribed strategy. This results in actual consequences of a management strategy differing from the intended results. To bring the management consequences closer to the intended results, adaptive management strategies can be sequentially modified at different stages of the management horizon using feedback measurements from a deigned monitoring network. This feedback information can be the actual spatial and temporal concentrations resulting from the implementation of actual management strategy. Therefore, field-scale compliance of the developed coastal aquifer management strategy is a crucial aspect of an optimally designed groundwater extraction policy. A 3-D compliance monitoring network design methodology is proposed in this study in order to develop an adaptive and sequentially modified management policy, which aims to improve optimal and justifiable use of groundwater resources in coastal aquifers. In the first step, an ensemble meta-model based multiple objective prescriptive model is developed using a coupled S-O approach in order to derive a set of Pareto optimal groundwater extraction strategies. Prediction uncertainty of meta-models is addressed by utilizing a weighted average ensemble using Set Pair Analysis. In the second step, a monitoring network is designed for evaluating the compliance of the implemented strategies with the prescribed management goals due to possible uncertainties associated with field-scale application of the proposed management policy. Optimal monitoring locations are obtained by maximizing Shannon's entropy between the saltwater concentrations at the selected potential locations. Performance of the proposed 3-D sequential compliance monitoring network design is assessed for an illustrative multilayered coastal aquifer study area. The performance evaluations show that sequential improvements of optimal management strategy are possible by utilizing saltwater concentrations measurements at the proposed optimal compliance monitoring locations.

The integrated S-O approach is used to develop a saltwater intrusion management model for a real world coastal aquifer system in the Barguna district of southern Bangladesh. The aquifer processes are simulated by using a 3-D finite element based combined flow and solute transport numerical code. The modelling and management of seawater intrusion processes are performed based on very limited hydrogeological data. The model is calibrated with respect to hydraulic heads for a period of five years from April 2010 to April 2014. The calibrated model is validated for the next three-year period from April 2015 to April 2017. The calibrated and partially validated model is then used within the integrated S-O approach to develop optimal groundwater abstraction patterns to control saltwater intrusion in the study area. Computational efficiency of the management model is achieved by using a MARS based meta-model approximately emulating the combined flow and solute transport processes of the study area. This limited evaluation demonstrates that a planned transient groundwater abstraction strategy, acquired as solution results of a meta-model based integrated S-O approach, is a useful management strategy for optimized water abstraction and saltwater intrusion control. This study shows the capability of the MARS meta-model based integrated S-O approach to solve real-life complex management problems in an efficient manner.

Item ID:	56083
Item Type:	Thesis (PhD)
Keywords:	coastal aquifer management, coastal aquifer, compliance monitoring, controlled elitist multi-objective genetic algorithm, ensemble models, ensemble, fuzzy c-means clustering, fuzzy inference system, fuzzy logic, genetic algorithm, linked simulation-optimization, mathematical modelling, monitoring network design, multivariate adaptive regression spline, parallel computing parameter uncertainty, salinity concentration, saltwater intrusion, simulation-optimization, surrogate model
Related URLs:	https://researchonline.jcu.edu.au/50403/ https://researchonline.jcu.edu.au/50399/ https://researchonline.jcu.edu.au/53015/ https://researchonline.jcu.edu.au/50401/
Copyright Information:	Copyright © 2018 Dilip Kumar Roy
Additional Information:	For this thesis, Dilip Roy received the Dean's Award for Excellence 2019. Publications arising from this thesis are available from the Related URLs field. The publications are: Chapter 2: Roy, Dilip Kumar, and Datta, Bithin (2017) Saltwater intrusion processes in coastal aquifers - modelling and management: a review. Desalination and Water Treatment, 78. pp. 57-89. Chapter 4: Roy, Dilip Kumar, and Datta, Bithin (2017) Fuzzy c-mean clustering based inference system for saltwater intrusion processes prediction in coastal aquifers. Water Resources Management, 31 (1). pp. 355-376. Chapter 6: Roy, Dilip Kumar, and Datta, Bithin (2017) Genetic algorithm tuned fuzzy inference system to evolve optimal groundwater extraction strategies to control saltwater intrusion in multi-layered coastal aquifers under parameter uncertainty. Modeling Earth Systems and Environment, 3 (4). pp. 1707-1725. Chapter 7: Roy, Dilip Kumar, and Datta, Bithin (2017) Multivariate adaptive regression spline ensembles for management of multilayered coastal aquifers. Journal of Hydrologic Engineering, 22 (9).
Date Deposited:	07 Nov 2018 22:50
FoR Codes:	09 ENGINEERING > 0905 Civil Engineering > 090509 Water Resources Engineering @ 100%
SEO Codes:	96 ENVIRONMENT > 9609 Land and Water Management > 960999 Land and Water Management of Environments not elsewhere classified @ 100%
Downloads:	Total: 269 Last 12 Months: 16
	More Statistics