Groundwater Quality Monitoring Network Design Research Articles

A GIS-expert-based approach for groundwater quality monitoring network design in an alluvial aquifer: a case study and a practical guide.

Groundwater quality monitoring is a critical part of water management in all groundwater basins. In order to be effective and to meet the required needs, groundwater quality monitoring networks (GQMNs) must be designed to be able to operate long-term and economically without minimal disruption. The analytical hierarchical process (AHP), a multi-criteria decision-making program, was used to design a GQMN for an alluvial aquifer located in the Islam Abad plain west of Kermanshah province, Iran. This semi-arid area is subject to groundwater depletion and water quality changes. The model used 8 primary criteria sub-divided with 5 sub-criteria based on a combination of empirical data and expert opinion. The primary criteria included density of wells, well discharge, well depth, water quality (conductivity), flow direction, annual groundwater extraction, water level declines, and accessibility. The model showed that 59 of 254 production wells in the basin could provide optimal monitoring locations. When a second screening of the wells was used to determine constraints (physical conditions of the wells and pumps, owner permission of use, type of the pump, etc.), the number of wells was reduced to 13 wells. An initial round of water sampling and chemical analysis demonstrated that the design of the GQMN met the goals of the water management agency of the region.

Groundwater quality monitoring network design and optimisation based on measured contaminant concentration and taking solute transit time into account

Groundwater quality monitoring networks are designed to give the best possible overview of the quality status and trend of aquifers from a limited number of measurement stations. Ideally, the stations of the network should (i) allow the detection of significant changes in water quality, either improvements due to successful mitigation measures, or deteriorations caused for instance by land use change or other increasing pressure and (ii) approximate closely the true unknown mean concentration of the considered aquifer. If these two features are met, the monitoring network is representative of the entire aquifer. In this article, we present a method with which to assess and improve the representativeness of a regional groundwater monitoring network. The method is designed specifically to optimize the selection of monitoring stations from a larger choice of fixed locations such as springs or existing observation wells using one contaminant for the optimisation procedure. Firstly, the empirical distribution of selected solutes (contaminants and age tracers) is calculated from a large dataset consisting of as many sampling points as possible and compared to the distributions obtained solely from the stations of the existing or planned monitoring network. If differences are large, the network is modified by adding new stations or replacing some of them, using one contaminant for the optimisation. Lastly, the distribution of the optimised network is compared to the total distribution for other contaminants and one groundwater age tracer.The use of the method is illustrated for the main aquifer of the country of Luxembourg, where it is shown that the representativeness of the current monitoring network can be improved by replacing two stations. In the case study, the optimisation was based on desethyl atrazine, a ubiquitous pesticide transformation compound which was assumed to be an indicator of diffuse agricultural pollution. Although the largest improvement was obviously obtained for the optimised compound, improvement was also observed for other contaminants, including nitrate, probably showing that at the regional scale, the pattern of diffuse pollution of widely used agrochemicals is similar between compounds. The temporal evolution of the major contaminants could also be reproduced by the optimised network, even though the range of tritium values used as age tracer was narrower for the modified network than for the original network, thus indicating some bias towards faster reaction times.

Development of an Entropy Method for Groundwater Quality Monitoring Network Design

A new approach is presented based on the entropy theory to design an optimal groundwater quality monitoring network. First, the unrecorded values of Total Dissolved Solids (TDS) are estimated in the study area using Artificial Neural Network (ANN) and K-Nearest Neighbor (KNN) methods, and then, using Differential Evolution (DE) optimization algorithm aimed at maximizing the pure information gained by stations. The number and location of monitoring stations are selected optimally from the active wells located in the study area. The spatial distribution of chosen wells provides adequate coverage of the whole aquifer and, at the same time, gives the maximum useful information about the water quality status. It also avoids selecting any redundant station considering the operational and temporal costs. The proposed entropy method is compared with “error minimization” and “K-Means clustering” methods and is shown to be superior. The proposed approach is used for optimizing Eshtehard aquifer monitoring network in central part of Iran. As a result of its implementation, 20 wells are selected among 79 active monitoring wells in the study area, so that by sampling the wells, the maximum qualitative information can be obtained without needing to sample all the wells.

A method for selecting monitoring wells and measured water-quality characteristics with application to the Liaohe River (China) groundwater system

A new method for regional groundwater quality monitoring network design is presented in this study. The monitoring design is based on a limited number of preexisting monitoring wells and water-quality data from a previous monitoring. An arithmetic overlay model for the multiple environmental factor score index (MEFSI) was created principle component scores representing factors affected by different natural and anthropogenic processes. The number of constituents per well to be analyzed does not only depend on the MEFSI but also on the hydrogeological settings. We apply the proposed method to design a groundwater quality monitoring network for the lower Liaohe River Plain in NE China. The results showed that the relationship between the monitoring cost and the total sum of standard deviation of the MEFSI can be depicted by the trade-off curve obtained from the MEFSI kriging variance analysis experiments. The reduction monitoring well experiments show that removing the redundancy wells from preexisting monitoring network is specially suitable for the quaternary aquifers system. Our proposed method has strong potential application in designing a groundwater quality monitoring network for the regional quaternary aquifers with relatively homogeneous properties.

Use of a relevance vector machine for groundwater quality monitoring network design under uncertainty

This paper presents a methodology for groundwater quality monitoring network design. This design takes into account uncertainties in aquifer properties, pollution transport processes, and climate. The methodology utilizes a statistical learning algorithm called relevance vector machines (RVM), which is a sparse Bayesian framework that can be used for obtaining solutions to regression and classification tasks. Application of the methodology is illustrated using the Eocene Aquifer in the northern part of the West Bank, Palestine. The procedure presented in this paper utilizes a Monte Carlo (MC) simulation process to capture the uncertainties in recharge, hydraulic conductivity, and nitrate reaction processes through the application of a groundwater flow model and a nitrate fate and transport model. This MC modeling approach provides several thousand realizations of nitrate distribution in the aquifer. Subsets of these realizations are then used to design the monitoring network. This is done by building a best-fit model of nitrate concentration distribution everywhere in the aquifer for each Monte Carlo subset using RVM. The outputs from the RVM model are the distribution of nitrate concentration everywhere in the aquifer, the uncertainty in the characterization of those concentrations, and the number and locations of “relevance vectors” (RVs). The RVs form the basis of the optimal characterization of nitrate throughout the aquifer and represent the optimal locations of monitoring wells. In this paper, the number of monitoring wells and their locations where chosen based on the performance of the RVM model runs. The results from 100 model runs show the consistency of the model in selecting the number and locations of RV‘s. After implementing the design, the data collected from the monitoring sites can be used to estimate nitrate concentration distribution throughout the entire aquifer and to quantify the uncertainty in those estimates.

Optimal Monitoring Network and Ground-Water–Pollution Source Identification

A methodology combining an optimal ground-water–quality monitoring network design and an optimal source-identification model is presented. In the first step of the three-step methodology, an embedded nonlinear optimization model is utilized for preliminary identification of pollutant sources (magnitude, location, and duration of activity) based on observed concentration data from arbitrarily located existing wells. The second step utilizes these preliminary identification results and a simulation optimization approach to design an optimal monitoring network that can be implemented in the subsequent time periods. In the third step, the observed concentration data at the designed monitoring well locations are utilized for more accurate identification of the pollutant sources. The design of the monitoring network can be dynamic in nature, with sequential installation of monitoring wells during subsequent time periods. The monitoring network can be implemented in stages, in order to utilize the updated information in the form of observed concentration data from a time-varying (dynamic) network. The performance evaluation of the proposed methodology demonstrates the potential applicability of this methodology and shows significant improvement in the identification of unknown ground-water–pollution sources with limited observation data.

APPLICATION OF GEOGRAPHIC INFORMATION SYSTEMS TO GROUNDWATER MONITORING NETWORK DESIGN1

ABSTRACT: Effective monitoring configurations for contaminant detection in groundwater can be designed by analyzing the spatial relationships between candidate sampling sites and aquifer zones susceptible to contamination. Examples of such zones are the domain underlying the contaminant source, zones of probable contaminant migration, and areas occupied by water supply wells. Geographic information systems (GIS) are well‐suited to performing key groundwater monitoring network design tasks, such as calculating values for distance variables which quantify the proximity of candidate sites to zones of high pollution susceptibility, and utilizing these variables to quantify relative monitoring value throughout a model domain. Through a case study application, this paper outlines the utility of GIS for detection‐based groundwater quality monitoring network design. The results suggest that GIS capabilities for analyzing spatially referenced data can enhance the field‐applicability of established methodologies for groundwater monitoring network design.

Review of Ground‐Water Quality Monitoring Network Design

Groundwater quality monitoring network design is defined as the selection of sampling sites and temporal sampling frequency to determine physical, chemical, and biological properties of ground wate...

An optimization approach for groundwater quality monitoring network design

The optimal sampling plan for groundwater quality monitoring is formulated as a mixed integer programming (MIP) problem. A sampling plan consists of the number and locations of sampling sites as well as the temporal sampling frequency. The MIP network problem is defined by the minimization of the variance of estimation error subject to resource and unbiasedness constraints. The mean and covariance of the spatial/temporal variable (chloride concentration measurements) are derived from the advection‐dispersion equation governing mass transport. The solution for the optimal sampling proceeds in two stages: (1) parameter estimation and (2) network optimization. The MIP model was successfully tested with a network design problem in a buried valley aquifer in Butler County, Ohio. The application illustrates the role of objective function, resource constraint, mass transport processes, and hydrogeologic setting in groundwater quality monitoring network design.