Contaminant Source Location Research Articles

Breaking the mold of simulation-optimization: Direct forward machine learning methods for groundwater contaminant source identification

Groundwater Contaminant Source Identification (GCSI) is important for addressing environmental concerns. Currently, it is widely achieved using the Simulation/Optimization (S/O) method. However, the utilization of optimization techniques may cause high computation costs and parameter equifinality issues. We introduce two innovative GCSI methods, Direct Forward Machine Learning (DFML) and One-Hot Machine Learning (OHML), utilizing the classical Artificial Neuro Network (ANN) model in the machine learning field. Both new methods eliminate the need for an optimization algorithm in GCSI, thus reducing the construction effort and improving efficiency. The first method, DFML can directly estimate eight parameters, providing valuable insights into the contaminant location, release history, and aquifer properties. The second method, OHML can estimate the spatial probability distribution of the contaminant location through one-hot encoding, addressing uncertainties in the contaminant source location estimations realized by DFML.Evaluations demonstrate that both methods exhibit satisfying performances. DFML can estimate contaminant location and aquifer properties with high accuracy; and estimate the release history information with moderate accuracy. The OHML correctly assigns the higher contaminant probabilities to regions containing true contaminant locations. The combination of DFML and OHML offers a comprehensive framework. This study contributes valid methodologies to the GCSI field.

Computational Fluid Dynamics Simulations to Assess Spatial Variability and Optimal Ventilation Scenarios for Biological Laboratory Exposures

Introduction: A significant amount of uncertainty exists regarding potential human exposure to laboratory biomaterials and organisms in Biosafety Level 2 (BSL-2) research laboratories. Computational fluid dynamics (CFD) modeling is proposed as a way to better understand potential impacts of different combinations of biomaterials, laboratory manipulations, and exposure routes on risks to laboratory workers. Methods: In this study, we use CFD models to simulate airborne concentrations of contaminants in an actual BSL-2 laboratory under different configurations. Results: Results show that ventilation configuration, sampling location, and contaminant source location can significantly impact airborne concentrations and exposures. Depending on the source location and airflow patterns, the transient and time-integrated concentrations varied by several orders of magnitude. Contaminant plumes from sources located near a return vent (or exhaust like a fume hood or ventilated biosafety cabinet) are likely to be more contained than sources that are further from the exhaust. Having a direct flow between the source and the exhaust (through-flow condition) may reduce potential exposures to individuals outside the air flow path. Conclusion: Designing a BSL-2 room with ventilation and airflow patterns that maximize through-flow conditions to the return/exhaust vents and minimize dispersion and mixing throughout the room is, therefore, recommended. CFD simulations can also be used to assist in characterizing the impacts of supply and return vent locations, room layout, and source locations on spatial and temporal contaminant concentrations. In addition, proper placement of particle sensors can also be informed by CFD simulations to provide additional characterization and monitoring of potential exposures in BSL-2 facilities.

Impact of different supply modes of stratum ventilation on airflow and contaminant distribution characteristics

For non-uniform air distribution, the contaminant accumulation occurs in the stagnant zone and leads to high exposure risk. A high supply airflow rate increases the air mixing and contaminant dilution but also the risk of the draft sensation. This study proposes dynamic stratum ventilation with variable supply air direction. The variable supply air direction is achieved by a supply grille with swing flaps. Experiments are conducted with various combinations of the constant/variable supply air direction, high/low supply airflow rate, and different contaminant source locations. The impact of different supply modes of stratum ventilation on the airflow and contaminant distribution characteristics are investigated. For stratum ventilation with the variable supply air direction, the turbulence intensity is quadratic polynomial correlated to the mean air velocity with a coefficient of 0.9853 because the periodically changed supply air direction affects the velocity fluctuation. Stratum ventilation with the variable supply air direction has airflow with the β-value of 1.37–1.59. The different air supply modes influence the horizontal CO2 distribution differently regarding the location of the contaminant source, which provides new insights into indoor air quality control under stratum ventilation. When the contaminant source is in the mixing zone, stratum ventilation with a high supply airflow rate and variable supply air direction can effectively reduce the draft risk and contaminant concentration.

Nickel and Chromium Origin in Fluvisols of the Petruševec Well Field, Zagreb Aquifer

Soil plays an important role in the accumulation and transport of potentially toxic elements (PTEs), from surface into aquifer. PTEs can get to the environment naturally, but also from different kinds of contamination sources. In this study, a soil profile located in the vicinity of well field Petruševec, one of the most important well fields related to the public water supply of the City of Zagreb, was analyzed. The main aim of this study was to determine soil properties which can influence retention/mobilization of Ni and Cr in alluvial soil, as well as to define their origin in the investigated soil profile. Results suggest that Cr is geogenic, while Ni is probably of dominantly anthropogenic origin. Observed concentrations, enrichment factors and Igeo values showed no enrichment for Cr, while for Ni, they showed minor to very severe enrichment, i.e., that in some soil horizons, moderate to strong pollution exists. Evaluation of wind directions and location of possible contamination sources that prevail in the study area suggest that Ni can come by aerodeposition from different sources. Results showed that mineral composition can have important influence on retention of analyzed PTEs. Soil horizons, which have very high concentrations of Ni, in general have higher proportion of clay minerals, especially chlorites, as well as Fe oxyhydroxides which can act as an adsorption phase for the investigated PTEs. Results suggest that more detailed research about the investigated PTEs presents a necessity if measures for soil and groundwater protection want to be effectively implemented.

Simultaneous identification of the number, location and release intensity of groundwater contamination sources based on simulation optimization and ensemble surrogate model

Abstract In most instances, the number, location and release intensity of groundwater contamination sources (GCSs) are all unknown. The 0-1 mixed integer nonlinear programming optimization model (0-1 MINPOM) used previously could only identify the location and release intensity for GCSs. Thus a 0-1 MINPOM was improved and applied to simultaneously identify the number, location and release history of GCSs. In addition, to reduce the time of calling the simulation model for iterative calculation, the deep belief neural network (DBNN), long short-term memory network (LSTM) and Kriging method were applied to establish an ensemble surrogate model of the simulation model to replace the simulation model for iterative calculation. The results show that compared with the three single surrogate models, the ensemble surrogate model has the highest approximation accuracy to the simulation model, and could save 99% of the calculation time when iterative calculation was performed in place of the simulation model. The improved 0-1 MINPOM could identify the number, location, and release history of GCSs simultaneously. The accuracy of identifying the number of GCSs was 100%, and the accuracies of identifying the locations and release history were above 95.12% and 83.51%, respectively.

Aquifer heterogeneity controls to quality monitoring network performance for the protection of groundwater production wells

A groundwater monitoring network surrounding a pumping well (such as a public water supply) allows for early contaminant detection and mitigation where possible contaminant source locations are often unknown. This numerical study investigates how the contaminant detection probability of a hypothetical sentinel-well monitoring network consisting of one to four monitoring wells is affected by aquifer spatial heterogeneity and dispersion characteristics, where the contaminant source location is randomized. This is achieved through a stochastic framework using a Monte Carlo approach. A single production well is considered that results in converging non-uniform flow close to the well. Optimal network arrangements are obtained by maximizing a weighted risk function that considers true and false positive detection rates, sampling frequency, early detection, and contaminant travel time uncertainty. Aquifer dispersivity is found to be the dominant parameter for the quantification of network performance. For the range of parameters considered, a single monitoring well screening the full aquifer thickness is expected to correctly and timely identify at least 12% of all incidents resulting in contaminants reaching the production well. This proportion increases to a global maximum of 96% for a network consisting of four wells and very dispersive transport conditions. Irrespective of network size and sampling frequency, more dispersive transport conditions result in higher detection rates. Increasing aquifer heterogeneity and decreasing aquifer spatial continuity also lead to higher detection rates, though these effects are diminished for networks of 3 or more wells. Statistical anisotropy has no effect on the network performance. Earlier detection, which is critical for remedial action and supply safety, comes with a significant cost in terms of detection rate, and should be carefully considered when a monitoring network is being designed.

Contaminant source characterization in a coastal aquifer influenced by tidal forces and density-driven flow

Contaminant source characterization (CSC) plays an important role in environmental forensics and pollution control. Past studies commonly solve the CSC problem in case studies where simple groundwater velocity fields (with e.g. parallel velocity vectors) prevail under steady state conditions. A few studies have addressed CSC in cases where the effect of heterogeneity or wells causes the velocity vectors to deviate from a parallel alignment. More complex transient velocity fields are yet to be addressed in groundwater CSC studies. One key example is coastal aquifers, where density-driven flow and the effect of oceanic forces (e.g. tides and waves) create complex transient velocity fields that lead to deformation of contaminant plumes and enhance mixing processes. The present paper aims to address this challenge by developing and validating a novel solution method for CSC in a coastal aquifer under the effect of tidal forces and density-driven flow. For this purpose, the study combines a numerical model of density-dependent flow and multiple-species solute transport, artificial neural networks, and a customized Kalman filtering technique, termed the ‘constrained restart dual ensemble Kalman filter’ (CRD-EnKF). In this approach, the time series of contaminant concentration downgradient of the source, and salinity concentrations in the transition zone between fresh and saline groundwater is employed for the estimation of contaminant source location and strength. We validated the proposed methodology by using a benchmark problem of a coastal aquifer with sloping beach and tide-induced pressure oscillation at the sea boundary.

Using data mining techniques to isolate chemical intrusion in water distribution systems.

The security of water distribution systems has become the subject of an increasing volume of research over the last decade. Data analysis and machine learning are linked to hydraulic and quality modeling for improving the capacity of water utilities to save lives when faced with the contamination of water networks. This research applies k-nearest neighbor and random forest algorithms to estimate the location of contamination sources at near-real time. Epanet and Epanet-MSX software are used to simulate intrusions of pesticide into water distribution system and the interaction with compounds already present in water bulk. Different pesticide concentrations are considered in the simulations, and chlorine monitoring occurs through placed quality sensors. The results show that random forest can localize [Formula: see text] of contamination scenarios, while the KNN algorithm found [Formula: see text]. Finally, an assessment of contamination spread is made for a better understanding of the impacts of non-localized contamination.

Transport-based source tracking of contaminants in a karst aquifer: Model implementation, proof of concept, and application to event-based field data

Identification and location of contamination sources is crucial for water resource protection — especially in karst aquifers which provide 25% of the world´s population with water but are highly vulnerable to contamination. Transport-based source tracking is proposed and verified here as a complementary approach to microbial and chemical source tracking in karst aquifers for identifying and locating such sources of contamination and for avoiding ambiguities that might arise from using one method alone. The transport distance is inversely modelled from contaminant breakthrough curves (BTC), based on analytical solutions of the 1D two-region non-equilibrium advection dispersion equation using GNU Octave. Besides the BTC, the model requires reliable estimates of transport velocity and input time. The model is shown to be robust, allows scripted based, automated 2D sensitivity analyses (interplay of two parameters), and can be favourable when distributed numerical models are inappropriate due to insufficient data. Sensitivity analyses illustrate that the model is highly sensitive to the input time, the flow velocity, and the fraction of the mobile fluid region. A conclusive verification approach was performed by applying the method to synthetic data, tracer tests, and event-based field data. Transport distances were correctly modelled for a set of artificial tracer tests using a discharge-velocity relationship that could be established for the respective karst catchment. For the first time such an approach was shown to be applicable to estimate the maximum distance to the contamination source for coliform bacteria in karst spring water combined with microbial source tracking. However, prediction intervals for the transport distance can be large even in well-studied karst catchments mainly related to uncertainties in the flow velocity and the input time. Using a maximum transport distance is proposed to account for less permeable, “slower” pathways. In general, transport-based source tracking might be used wherever transport can be described by the 1D two-region non-equilibrium model, e.g. rivers and fractured or porous aquifers.

Contaminant Source Identification from Finite Sensor Data: Perron–Frobenius Operator and Bayesian Inference

Sensors in the built environment ensure safety and comfort by tracking contaminants in the occupied space. In the event of contaminant release, it is important to use the limited sensor data to rapidly and accurately identify the release location of the contaminant. Identification of the release location will enable subsequent remediation as well as evacuation decision-making. In previous work, we used an operator theoretic approach—based on the Perron–Frobenius (PF) operator—to estimate the contaminant concentration distribution in the domain given a finite amount of streaming sensor data. In the current work, the approach is extended to identify the most probable contaminant release location. The release location identification is framed as a Bayesian inference problem. The Bayesian inference approach requires considering multiple release location scenarios, which is done efficiently using the discrete PF operator. The discrete PF operator provides a fast, effective and accurate model for contaminant transport modeling. The utility of our PF-based Bayesian inference methodology is illustrated using single-point release scenarios in both two and three-dimensional cases. The method provides a fast, accurate, and efficient framework for real-time identification of contaminant source location.

Prompt location of indoor instantaneous air contaminant source through multi-zone model-based probability method by utilizing airflow data from coarse-grid CFD model

In order to ensure indoor air quality safety, locating the airborne contaminant sources accurately and quickly is extremely important so that timely measures can be taken to undermine the spread of pollutant and even eliminate the negative effects. Previous studies have shown that the multi-zone model can greatly reduce the inverse calculation time. However, the multi-zone model cannot describe the details of the indoor velocity field, which limits its application in complex multi-zone or large space buildings. On the premise of the accuracy and computational speed, based on the joint probability method, this study adopted the coarse-grid CFD method to speed up the process of acquiring the indoor airflow field, together with the multi-zone model method, to perform the inverse calculation of indoor airborne contaminant source location. In the backward calculation process, we conducted the ‘transpose' of the velocity field to obtain adjoint matrix, instead of computing ‘negative' of the velocity vector to save the calculation time. A two-dimensional ventilation model was utilized to validate the method, which proved the accuracy and time-saving potential of it. This study provides theoretical and practical prospect for the real-time inverse calculation of locating the indoor airborne contaminant sources.

Application of the Layered Algorithm in search of an airborne contaminant source

The paper presents a new method of optimization by the Layered Algorithm (LA). The proposed algorithm reduces the initial area to the sub-area containing the optimum. The proposed technique is based on the classification of the optimized function values (data sampled by sensors). The classification method uses a two-dimensional three-state Cellular Automata (CA). The CA classifies all area points ascribed to the CA cells based on their values. Specification of the categorization layers to the data gives a possibility to identify the different levels areas. Consequently, after analysis, a sub-area containing the optimum can be designated. In this paper, the proposed algorithm is applied to find the location of the airborne contaminant source by analyzing the concentration of released substances reported by mobile sensors distributed over the domain of interest. The Gaussian dispersion model simulation of the contaminant dispersion in the urbanized area is applied to generate the data used to verify the efficiency of the proposed Layered Algorithm. The LA successfully estimates the sub-area of the considered domain where the contamination source is located, taking to account data from sensors solely.

Tracing indoor contaminant release location based on local mean residual-life-time of air

Even in case ventilation system is operating in optimal fashion, indoor air quality could be varied depending on the indoor contaminant release occasions. Thus, accurate and rapid identification of contaminant source locations in indoor environment is critical for occupants’ health as well as building safety and integrity. This study presented a novel method to locate indoor contaminant source using local mean residual-life-time (LMR). Based on the theoretical definition of LMR, a characteristic quantitative relationship between the LMR at source location and the contaminant concentration profile of a room was derived and assessed through two stages of studies: experimental and numerical explorations. In the experiments, by changing contaminant release location under limited condition, the LMR at the source location and the contaminant concentration profile of a test chamber were measured. Then, the outcomes were examined to verify whether the derived equation was established in the chamber. Computational Fluid Dynamics (CFD) technique was also used to make up for the experimental limitations. CFD simulations were carried out to analyze the validity of the equation in response to different release locations of the contaminant in a more complex physical environment by obtaining high-resolution information. Both experimental and numerical results showed that the presented equation was well established and could be used to locate indoor contaminant source even in non-ideal situation or in complicated spaces.

Integrated hydrogeophysical modelling and data assimilation for geoelectrical leak detection

Time-lapse electrical resistivity tomography (ERT) measurements provide indirectobservations of hydrological processes in the Earth's shallow subsurface at high spatial and temporal resolution. ERT has been used in the past decades to detect leaks and monitor the evolution of associated contaminant plumes. Specifically, inverted resistivity images allow visualization of the dynamic changes in the structure of the plume. However, existing methods do not allow the direct estimation of leak parameters (e.g. leak rate, location, etc.) and their uncertainties. We propose an ensemble-based data assimilation framework that evaluates proposed hydrological models against observed time-lapse ERT measurements without directly inverting for the resistivities. Each proposed hydrological model is run through the parallel coupled hydro-geophysical simulation code PFLOTRAN-E4D to obtain simulated ERT measurements. The ensemble of model proposals is then updated using an iterative ensemble smoother. We demonstrate the proposed framework on synthetic and field ERT data from controlled tracer injection experiments. Our results show that the approach allows joint identification of contaminant source location, initial release time, and solute loading from the cross-borehole time-lapse ERT data, alongside with an assessment of uncertainties in these estimates. We demonstrate a reduction in site-wide uncertainty by comparing the prior and posterior plume mass discharges at a selected image plane. This framework is particularly attractive to sites that have previously undergone extensive geological investigation (e.g., nuclear sites). It is well suited to complement ERT imaging and we discuss practical issues in its application to field problems.

Groundwater contamination sources identification based on kernel extreme learning machine and its effect due to wavelet denoising technique.

Measurements of contaminant concentrations inevitably contain noise because of accidental and systematic errors. However, groundwater contamination sources identification (GCSI) is highly dependent on the data measurements, which directly affect the accuracy of the identification results. Thus, in the present study, the wavelet hierarchical threshold denoising method was employed to denoise concentration measurements and the denoised measurements were then used for GCSI. A 0-1 mixed-integer nonlinear programming optimization model (0-1 MINLP) based on a kernel extreme learning machine (KELM) was applied to identify the location and release history of a contamination source. The results showed the following. (1) The wavelet hierarchical threshold denoising method was not very effective when applied to concentration measurements observed every 2months (the number of measurements is small and relatively discrete) compared with those obtained every 2days (the number of measurements is large and relatively continuous). (2) When the concentration measurements containing noise were employed for GCSI, the identifications results were further from the true values when the measurements contained more noise. The approximation of the identification results to the true values improved when the denoised concentration measurements were employed for GCSI. (3) The 0-1 MINLP based on the surrogate KELM model could simultaneously identify the location and release history of contamination sources, as well reducing the computational load and decreasing the calculation time by 96.5% when solving the 0-1 MINLP.

Aggregated GP-based Optimization for Contaminant Source Localization

Recently a new simulation-based optimization benchmark of groundwater contaminant source localization problems has been introduced to the hydrogeological science community. Given information on contaminant concentration levels at each monitoring well and each time step, its objective is to identify the location of contaminant source. In this work, we analyze and look at the problem from different angles to gain more insights on this class of groundwater problems. To tackle the problem, a novel simulation-based optimization algorithm relying on an aggregated Gaussian process model, and the expected improvement criterion is introduced. Results from this study show that the proposed algorithm, though relying on an approximated Gaussian process model, demonstrates superior efficiency and reliability than a traditional expected improvement-based algorithm. The location of the monitoring wells was confirmed to play a crucial role in assisting the optimization algorithm to accurately localize the contaminant source. Additional monitoring wells, while adding more knowledge of the space-time mapping of concentration levels, could nevertheless slow down convergence of the algorithm due to the increase in problem complexity.

산업단지 지하수 오염원 위치 및 누출 이력 규명 기술의 특성 분석

산업단지 지하수 오염원 위치 및 누출 이력 규명 기술의 특성 분석

An iterative strategy for contaminant source localisation using GLMA optimization and Data Worth on two synthetic 2D Aquifers

A contaminant source localisation strategy was developed considering unknown heterogeneous hydraulic conductivity field, unknown dispersivity and unknown location of a continuous contaminant source. The Gauss-Levenberg-Marquardt algorithm is combined with a data worth analysis to estimate the unknown parameters and identify the best locations of additional measurements. The data collection strategy is iterative, based on the ability of the additional dataset to decrease the uncertainties on the contaminant source location. Two 2D synthetic models are considered. The method is first illustrated with a simple model and a more complex model is then considered to evaluate the ability of the approach to locate the contaminant source from hydraulic heads and concentration data. This approach is parsimonious in terms of model runs and applicable to real cases. The results give a good estimate of the source location and the dispersivity, with acceptable NRMSE for each case. New observations introduced at each iteration decrease the standard deviation of the source location and improve the NRMSE. The estimated hydraulic conductivity field presents the same features as the original field.

Heavy Metals in Bottom Sediments of Lake Kenon (The Trans-Baikal Territory, Russia).

Concentrations of heavy metals in bottom sediments of Lake Kenon in descending order are distributed as follows: Mn > Zn > Pb > Mo > Cd. Spatial distribution of metals in bottom sediments of Lake Kenon depends on composition of bottom deposits (sands in shallow water and sapropel silt in the deep part), location of contamination sources (thermal power station, residential area), as well as density and duration of growth of aquatic plants. The greatest pollution of bottom sediments was observed in the area of TPP-1. Due to the intense and all-year-round process of aquatic vegetation growth in the area of TPP-1 contaminants are being accumulated in the bottom sediments of this part of the lake. However, plants that absorb metals in excess amounts and are passively moved by currents through of the system become a source of contamination of bottom sediments in relatively clean parts of the water reservoirs.

Contaminant source localization via Bayesian global optimization

Abstract. Contaminant source localization problems require efficient and robust methods that can account for geological heterogeneities and accommodate relatively small data sets of noisy observations. As realism commands hi-fidelity simulations, computation costs call for global optimization algorithms under parsimonious evaluation budgets. Bayesian optimization approaches are well adapted to such settings as they allow the exploration of parameter spaces in a principled way so as to iteratively locate the point(s) of global optimum while maintaining an approximation of the objective function with an instrumental quantification of prediction uncertainty. Here, we adapt a Bayesian optimization approach to localize a contaminant source in a discretized spatial domain. We thus demonstrate the potential of such a method for hydrogeological applications and also provide test cases for the optimization community. The localization problem is illustrated for cases where the geology is assumed to be perfectly known. Two 2-D synthetic cases that display sharp hydraulic conductivity contrasts and specific connectivity patterns are investigated. These cases generate highly nonlinear objective functions that present multiple local minima. A derivative-free global optimization algorithm relying on a Gaussian process model and on the expected improvement criterion is used to efficiently localize the point of minimum of the objective functions, which corresponds to the contaminant source location. Even though concentration measurements contain a significant level of proportional noise, the algorithm efficiently localizes the contaminant source location. The variations of the objective function are essentially driven by the geology, followed by the design of the monitoring well network. The data and scripts used to generate objective functions are shared to favor reproducible research. This contribution is important because the functions present multiple local minima and are inspired from a practical field application. Sharing these complex objective functions provides a source of test cases for global optimization benchmarks and should help with designing new and efficient methods to solve this type of problem.