Groundwater Flow Research Articles

Modelling the impacts of climate change on agrochemical fate and transport by water on a catchment scale

The export of agrochemicals and their transformation products (TPs) following their application in the agricultural fields poses a threat to water quality. Future changes in climatic conditions (e.g. extreme weather events such as heavy rainfall or extended dry periods) could alter the degradation and mobility of agrochemicals. In this research, we use an integrated modelling framework to understand the impact of extreme climate events on the fate and transport of the agrochemical S-Metolachlor and two of its TPs (M-OXA, Metolachlor Oxanilic Acid and M-ESA, Metolachlor Ethyl Sulfonic Acid). This is done by coupling climate model outputs to the Zin-AgriTra agrochemical reactive transport model in four simulation scenarios. 1) Reference (2015–2018), 2) Very dry (2038–2041), 3) Very wet (2054–2057) and 4) High temperature (2096–2099) conditions of a selected RCP8.5 based regional climate scenario. The modelling framework is tested on an agricultural catchment, Wulka, in Burgenland, Austria. The model results indicate that 13–14 % of applied S-Metolachlor is retained in the soil, and around 85 % is degraded into TPs in the different scenarios. In very dry and high-temperature scenarios, degradation is higher, and hence, there is less S-Metolachlor in the soil. However, a large share of formed M-OXA and M-ESA are retained in the soil, which is transported via overland and groundwater flow, leading to a build-up effect in M-OXA and M-ESA river concentrations over the years. Though a small share of S-Metolachlor and TPs are transported to rivers, their river export is affected by the intensity and amount of rainfall. The very wet and high-temperature scenarios show higher S-Metolachlor and TP concentrations at the catchment outlet due to higher river discharge. The reference scenario shows higher river peak concentrations associated with higher overland flow caused by measured hourly rainfall compared to disaggregated daily precipitation data in the other scenarios.

Understanding Climate Change and Anthropogenic Impacts on the Salinization of Low‐Lying Coastal Groundwater Systems

AbstractFresh coastal groundwater is a valuable water resource of global significance, but its quality is threatened by saltwater intrusion. Excessive groundwater abstraction, sea‐level rise (SLR), land subsidence and other climate‐related factors are expected to accelerate this process in the future. The objective of this study is to (a) quantify the impact of projected climate change and (b) explore the role of individual hydrogeological boundaries on groundwater salinization of low‐lying coastal groundwater systems until 2100 CE. We employ numerical density‐dependent groundwater flow and salt transport modeling for this purpose, using Northwestern Germany as a case. Separate model variants are constructed and forced with climate data, that is, projected SLR and groundwater recharge, as well as likely ranges of other hydrogeological boundaries, including land subsidence, abstraction rates and drain levels. We find that autonomous salinization in the marsh areas, resulting from non‐equilibrium of the present‐day groundwater salinity distribution with current boundary conditions, is responsible for >50% of the salinization increase until 2100 CE. Sea‐level rise, land subsidence and drain levels are the other major factors controlling salinization. We further show that salinization of the water resources is a potential threat to coastal water users, including water suppliers and the agrarian sector, as well as coastal ecosystems. Regional‐scale uplifting of drain levels is identified as an efficient measure to mitigate salinization of deep and shallow groundwater in the future. The presented modeling approach highlights the consequences of climate change and anthropogenic impacts for coastal salinization, supporting the timely development of mitigation strategies.

Effects of groundwater pumping on pore water flow and salt transport in tide‐controlled unconfined coastal aquifers

AbstractUnconfined coastal aquifers are a main pathway for land‐sourced solutes to enter the oceans. The migration of these solutes in aquifers is highly affected by the groundwater flow and salt transport processes, which are, to a great extent, controlled by tides. While many studies have examined how tidal oscillations would influence the subsurface hydrodynamics in coastal aquifers, most of them ignored the potential impact of groundwater pumping, a common practice in coastal areas to satisfy the demand for freshwater. This study, by means of laboratory experiments and numerical simulations, explored the combined effects of tides and groundwater pumping on the pore water flow and salinity distributions in an unconfined coastal aquifer. The results show that, in a tide‐controlled aquifer, the addition of groundwater pumping would exacerbate the degree of seawater intrusion and lead to wider spreading and deeper penetration of the upper saline plume. Moreover, groundwater pumping would enhance the tide‐driven circulation in the upper saline plume and weaken the density‐driven circulation in the saltwater wedge, ultimately leading to the reduction in total submarine groundwater discharge. These findings may promote a deep insight into the complex coastal groundwater systems experiencing human activities, and provide guidance for better evaluating the environmental impact of groundwater pumping.

Similarity solutions of a class of unsteady laminar boundary layer

The study of laminar unsteady boundary layer flows is essential for understanding the transition from laminar to turbulent flow, as well as the origins of turbulence. However, finding solutions to this phenomenon poses significant challenges. In this study, we introduce a novel method that employs a similarity transformation to convert the two-dimensional unsteady laminar boundary layer equations into a single partial differential equation with constant coefficients. By applying this transformation, we successfully derive similarity solutions for flat plate boundary layer flow, expressed in terms of Kummer functions. For convergent boundary layer flow, we derive an approximate analytical solution that includes both shock wave and soliton wave solutions. The superposition of these solutions provides evidence for the existence of solitons or soliton-like coherent structures (SCS) within boundary layers. Additionally, this paper explores two- and three-dimensional laminar flows, as well as three-dimensional turbulent flow equations, revealing that they all incorporate third-order derivatives with respect to spatial coordinates. This finding suggests that all viscous fluid motions have the potential to exhibit solitons/like coherent structures (SCS).

New Insights into the Velocity of Groundwater Flow in Geothermal Systems

New Insights into the Velocity of Groundwater Flow in Geothermal Systems

Theoretical evaluation of high-permeability wellbore skin effect on aquifer response under pumping

Skin zones with distinct hydraulic properties often form around wellbores during drilling operations. Conventional skin-factor models, developed for low-permeability (positive) skin, assume discontinuous hydraulic heads between the wellbore and aquifer. However, these models may be inadequate for high-permeability (negative) skin scenarios resulting from intensive well development. This study derives a new skin-factor approach for negative skin approaches based on the two-zone model approach. We derive a new skin-factor approach that maintains continuous heads between the wellbore and aquifer while allowing for discontinuous fluxes. Using Laplace and finite cos-Fourier transformations, we develop a groundwater flow model for aquifer pumping under a partially penetrating well, incorporating this new approach. An asymptotic expansion method is employed to obtain a large-time solution. The proposed model is also compared to the existing two-zone model. Sensitivity analysis reveals that high-permeability skin significantly affects drawdown, particularly when there is a sudden increase in porosity within the skin zone. This revised approach provides a more accurate framework for modeling various skin effects, offering valuable insights for groundwater management and well hydraulic modeling in both hydrogeology and petroleum engineering applications.

Hydrogeological conceptual model of Stampriet transboundary aquifer system in Southern Africa

Stampriet Transboundary Aquifer System (STAS) is shared between Botswana, Namibia and South Africa with massive utilization especially in Namibia. However, the understanding of the comprehensive hydrogeological framework of the aquifer system is limited. This study aimed to develop a comprehensive hydrogeological conceptual model of the aquifer system which mainly entailed 3D hydrostratigraphic modelling using Rockworks 3D geological modelling software, analysis of regional groundwater occurrence and flow system, sources and sinks, hydraulic parameters and definition of STAS boundaries. Six hydrostratigraphic units composed of four aquifers, including a new one (the Dwyka), and two aquitards were identified. Analysis of the regional groundwater flow system revealed that groundwater flows in two directions, indicating presence of a regional groundwater divide. 3D hydrostratigraphic geometry revealed that groundwater flow in STAS is influenced by topography but also by faults such as the Zoetfontein Fault, which displaces the pre-Kalahari rocks vertically creating a barrier to groundwater flow. Hydrostratigraphic and 3D geometry work also led to a 42% increase of the size of STAS compared to the boundary initially delineated by UNESCO. The proposed hydrogeological conceptual model forms a basis for development of numerical integrated hydrological model to simulate surface-groundwater interactions, regional groundwater flow and aquifer development potential.

Quantum vortex stability in draining fluid flows

Quantum vortices with more than a single circulation quantum are usually unstable and decay into clusters of smaller vortices. One way to prevent the decay is to place the vortex at the center of a convergent (draining) fluid flow, which tends to force vortices together. It is found that while the primary splitting instability is suppressed in this way (and completely quenched for strong enough flows) a secondary instability can emerge in circular trapping geometries. This behavior is related to an instability of rotating black holes when superradiantly amplified waves are confined inside a reflective cavity. The end state of the secondary instability is dramatic, manifesting as a shock wave that propagates round the circular wall and nucleates many more vortices. Published by the American Physical Society 2024

Regulated Inductively Coupled Plasma–Optical Emission Spectrometry Detectible Elements in Utah Lake: Characterization and Discussion

During the 2021 (n = 15) and 2022 (n = 13) summers, we measured the total and dissolved (<0.45 μm) concentration of 25 elements in Utah Lake using Inductively Coupled Plasma–Optical Emission Spectrometry (ICP-OES) with detection limits in the order of a few parts-per-billion (ppb). This resulted in 1400 measurements, which is a unique dataset in terms of sensitivity and temporal resolution. Regulated elements are not commonly measured at the ppb-level; thus, these data provide insight into both the behavior and existence of these elements in an aquatic environment and have implications for both the management and regulation of the lake. Utah regulates twelve of these elements. While ICP-OES has ppb-level sensitivity, it is not the approved regulatory analysis method for these elements. All regulations are for dissolved concentrations, except aluminum (Al) and phosphorus (P), which are for total recovery. We found total Al above the allowable concentration, but dissolved concentrations were well below allowable concentrations. We attribute high total concentrations to suspended clays. This suggests that regulatory methods should be reviewed for lakes with a high suspended-solid content. Dissolved copper (Cu) concentrations were below regulatory levels in 2021, but some samples were above regulatory levels in 2022. This could be related to the use of Cu-based algaecide treatments, or from other sources. Lead (Pb) data were inconclusive; dissolved Pb concentrations were well below the acute (1 h average) limit, but the chronic concentration limit (4 h average) was below the ICP-OES minimal detection limit. Arsenic (As) concentrations exhibited a seasonal trend that we attribute to groundwater inflows—they were below regulatory levels for aquatic environments but around the levels for drinking water. This ppb-level study with high temporal resolution provides insight into regulated elements in Utah Lake previously not available due to the high sensitivity of the method and measurements of both total and dissolved concentrations.

Experimental Investigation on a Water Diversion Shield Segment Using a Newly Developed Model Test Chamber

Jinan City in China is known for its abundant spring water, and the protection of underground spring water transport channels is a key issue that needs to be paid attention to in Jinan subway construction. In areas where underground spring water transport channels are easily disturbed, when the shield tunnel passes through a composite formation with a permeable layer in the middle and an impermeable layer in the upper and lower parts, the underground spring water transportation channel is blocked, the spring veins are damaged, and the underground spring water is blocked in the upstream, resulting in an increase in the water pressure difference on both sides of the tunnel. In this paper, based on the working principle of a new type of pipe segment with a diversion function, a diversion function pipe model test box is designed and fabricated. Using this model box, the working state of the diversion function pipe segments is simulated. An earth pressure box and pore water pressure gauges are embedded in the model box to measure the earth pressure and pore pressure around the pipe under seepage unsymmetrical pressure. Strain gauges are attached in a circular shape on the inner side of the pipe to measure and calculate the strain and stress of the pipe. Through model tests, the effective repair of the diversion pipe on blocking groundwater flow and the relief effect of the unsymmetrical pressure of the pipe are verified.

Extent of gross underestimation of precipitation in India

Abstract. The underestimation of precipitation (UoP) in the hilly and mountainous parts of South Asia is estimated by some studies to be as large as the observed precipitation (P). However, UoP has been analyzed to only a limited extent across India. To help bridge this gap, watershed-scale UoP was analyzed using various P datasets within a water imbalance analysis. Among these P datasets, the often-used Indian Meteorological Department (IMD) dataset is of primary interest. The gross UoP was identified by analyzing the extent of the imbalance in the annual water budget of watersheds corresponding to 242 river gauging stations for which quality-controlled data on catchment boundaries and streamflow are available. The water year (WY)-based volume of observed annual P was compared against the observed annual streamflow (R) and the satellite-based actual evapotranspiration (ET). Across many watersheds of both Northern and Peninsular India, spurious water imbalance scenarios (P≤R or P≪R+ET) were realized. It is shown that the management of water, such as groundwater extraction, reservoir storage and water diversion, is generally minimal compared to the annual P in such watersheds. It is also shown that annual changes in terrestrial water storage are minimal compared to the annual P in such watersheds. Assuming that data on R (and, to a lesser extent, ET) are reliable, it is concluded that UoP is very likely the cause of this imbalance. Inter-watershed groundwater flow (IGF) is assumed to be negligible. While the effect of IGF on R is unknown, examples are provided which show that IGF is unlikely to be the cause of the observed imbalance in certain watersheds. All 12 of the P datasets analyzed here suffer from UoP, but the extent of the UoP varies by dataset and region. The reanalysis-based datasets ERA5-Land and IMDAA are less affected by UoP than the IMD dataset. Based on the 30-year period of WY 1985–2014, P for the whole of India could be as much as 19 % (ERA5-Land) to 37 % (IMDAA) higher than that from the IMD, with substantial variability within years and river basins. The actual magnitude of UoP is speculated to be even greater. Moreover, trends seen in the IMD's P are not always present in ERA5-Land and IMDAA. Studies using IMD should exercise caution since UoP could lead to the misrepresentation of water budgets and long-term trends. Limitations of this study are discussed.

Potential contamination of groundwater quality in west Algeria by non-biodegradable effluents from public discharges

The aim of this work is to study the environmental impact of the municipal landfill in the town of Sidi Bel Abbes (western Algeria), in particular on the water table, which circulates at a shallow depth (Max. 20 m) under permeable soil. The huge quantities of putrescible waste discharged into this landfill generate highly polluting leachate water, rich in organic matter (COD up to 1300 mg/L, 209 mg/L in BOD5), salts and micro-organisms. A physico-chemical study of well water in the region reveals a significant deterioration in the quality of the underlying groundwater. In fact, electrical conductivities (on average 5.75 mS/cm) are common in wells, especially those located close to the landfill site. This indicates heavy contamination of the water in these wells by leachate. As for the organic and metallic load, abnormally high levels (82 mg/l in COD, 40 mg/l in BOD₅, 0.74 mg/L in Cr, 0.45 mg/L in Cd, 0.24 mg/L in Cu, 0.85 mg/L in Ni) were observed, mainly in the wells located near and downstream of the landfill, in the direction of groundwater flow.

Heuristic estimation of river bathymetry in braided streams using digital image processing

AbstractMeasurement of river bathymetry has been revolutionized by high‐resolution remote sensing that combines UAV platforms with SfM‐MVS photogrammetry. Mapping inundated and exposed areas simultaneously are possible using either two‐media refraction correction or some form of the Beer–Lambert Law to estimate water depths. If, as in turbid glacially‐fed braided streams, the bed is not visible then traditional survey techniques (e.g. differential GPS systems) are required. As an alternative, here we test whether the spatial distribution of water depths in a shallow braided stream can be modelled from basic planimetric data and used to estimate inundated zone bathymetry. We develop heuristic rules using; (1) distance from the nearest river bank; (2) total inundated width along a line tangential to the local flow direction; (3) local curvature magnitude and direction; and distance from the nearest flow (4) divergence and (5) convergence regions. We parameterize them using a sample of measured water depths in stepwise multiple linear regressions and validate them using independent data. Resulting water depth distribution maps explain between 50% and 60% of the measured water depth spatial variability when compared to independent data. After incorporating modelled water depths into digital elevation models (DEMs) of exposed areas, we show that the developed method is suitable for volumetric change calculations in both dry and inundated areas.

Groundwater flow regime shapes nitrogen functional traits by affecting microbial community assembly processes in the subsurface

The complex nitrogen (N) cycle in groundwater systems is affected by both biological and environmental factors. The interactions between hydrogeological conditions and the microbial community assembly processes that impact N-cycling processes remain poorly understood. We explored the assembly patterns of N-cycling microbial communities along the groundwater flow path. The environmental heterogeneity in different hydrological phases increased along the flow path (mean Ed: 0.16–0.49), accompanied by different microbial community assembly patterns. The assembly patterns that engaged in dissimilatory nitrate reduction to ammonium (DNRA) and denitrification changed across the water-sediment phases. Nitrifying microorganisms in the discharge area were mainly influenced by heterogeneous selection (41–69 %), and were closely correlated with dissolved oxygen (DO) concentrations. Homogeneity along flow-through increased stochastic assemblies, such as downstream drift of anammox bacterial (AnAOB) communities. Thus, the N removal pathway changed from “nitrification-denitrification” in the recharge area to “partial nitrification-anammox” in the discharge area. The increasing environmental heterogeneity brought more deterministic assembly patterns of N-cycling communities, linked to higher community turnover along the groundwater flow path. This study indicated that groundwater flow regime determined microbial community assembly patterns, providing valuable insight into the response of N transitions to environmental variations in groundwater systems.

The water cycle of small catchments impacted with tailings mudflows: A study in the Ferro-Carvão watershed after the breakup of B1 dam in Brumadinho

The B1 tailings dam of Córrego do Feijão iron-ore mine owned by Vale, S.A. company collapsed in 25 January 2019 releasing to the Ferro-Carvão stream watershed (32.6 km2) as much as 11.7 Mm3 of mine waste. A major share (8.9 Mm3) has been deposited along the stream channel and margins forming a 2.7 km2 patch. The main purpose of this study was to question whether the tailings deposit impacted the local water cycle and how. Using the Soil and Water Assessment Tool (SWAT) hydrologic model, the water balance components of 36 hydrologic response units (HRU) were calculated for pre- (S1) and post- (S2) B1 dam rupture scenarios represented by appropriate soil, land use and tailings cover. The results revealed an increase of evapotranspiration from S1 to S2, related to the sudden removal of vegetation from the stream valley and replacement with a blanket of mud, which raised the exposure of Earth's surface to sunlight and hence soil evaporation. For 11 HRU (10.3 km2) located around the tailings deposit, a decrease in lateral flow was observed, accompanied by an increase in percolation and a slight increase in groundwater flow. In this case, the water balance changes observed between S1 and S2 reflected a barrier effect imposed to the lateral flows by the tailings, which shifted the flows towards the vertical direction (percolation). Thus, the water followed an easier vertical route until reaching the shallow aquifer and being converted into groundwater flow. As per the modelling outcomes, the hydrologic impacts of B1 dam rupture are relevant because they affected 1/3 of Ferro-Carvão stream watershed, and hence claim for the complete removal of the tailings.

Convergence of online flow and AI-enabled services: the impact on awe experience in e-tail customer journeys

ABSTRACT This study investigates how control, concentration, and enjoyment impact customers’ awe experience in e-tailing, and examines how AI-enabled services may alter online flow dynamics. The research employed cutting-edge data collection techniques using an AI-driven platform developed by an Indian online survey firm. Participants were recruited via Amazon Mechanical Turk and snowball sampling. Variables were measured using a Likert-type scale, with survey credibility and validity ensured through rigorous measures, including feedback collection and analysis. Through a novel research model and the ‘broaden-and-build theory,’ the study reveals that customers’ awe experience can be significantly influenced by control and concentration. Surprisingly, AI-powered e-retail services negatively affected two flow dimensions and subsequent awe experience. However, customers’ awe experience positively correlated with AI service quality, indicating emotions’ criticality. These findings provide new insights for future research and e-tailers to design effective marketing strategies enhancing customer journeys. The study underscores the importance of fashion retailers embracing AI technology to meet dynamic customer expectations, yielding enhanced efficiency and cost-effectiveness through process automation and waste minimization. By integrating AI, the industry undergoes a significant transformation, effectively redefining the shopping experience.

Iron redox states in closed-basin lakes on early Mars: Its sensitivity to water chemistry, hydrology, and atmospheric composition

Mars has experienced drastic environmental evolution in terms of pH, redox, and desiccation. NASA’s Curiosity rover discovered that the sediments of the Pahrump Hills member of the Murray formation in Gale Crater contained different redox states of Fe oxides. To interpret the observations, understanding the factors that control iron redox states within early Gale lakes on Mars is needed. Here, we present the results of a one-dimensional geochemical model of a closed-basin lake that considers the Fe(II) photo-oxidation reaction. Depending on the source of Fe(II) (groundwater vs. river flow), lake pH, advection rate, and atmospheric composition, we defined four types of chemical profiles in the lake; namely, Fe(II) dominated, Fe(III) dominated, carbonate precipitating, and redox-stratified profiles. The Fe(III) dominated profile corresponds to the end member case in which photo-oxidation effectively oxidizes the supplied Fe(II). The redox-stratified profile, interpreted as an intermediate condition between Fe(III) dominated profile and Fe(II) dominated profile, appears in broad parameter set when Fe(II) is supplied from the groundwater, typically with a low to moderate Fe(II) input flux at acidic to alkaline pH (pH 5–9) conditions; however, the low water flux due to the nature of the groundwater may be unable to explain the measured abundance of Fe oxides in the sediments. Another endmember case, the Fe(II) dominated profile, which would also be expected to generate mixed valence Fe oxides, occurs when the pH is acidic (pH 4–5); nevertheless, the acidic pH is inconsistent with the mineral assemblages of the sediments. Carbonate precipitating profiles were limited to the case in which Fe(II) is supplied by groundwater with high Fe input flux and alkaline pH of 8–9. These results may imply either the Fe(II) input flux was high enough for the redox-stratified case or Fe(II) photo-oxidation was suppressed by turbidity or by dusty/cloudy atmosphere, or possibly that Fe(II) oxides found in the Pahrump Hills member are, at least partly, of detrital origin. Although numerous other factors (ex., effect of lake depth, variable pH throughout the lake water, detailed chemical speciation) remain unconsidered, the diverse profiles in Fe redox resulting from Fe(II) photo-oxidation reaction sheds light on interpreting the variety and the evolution of redox condition on early Mars.

Instability characteristics and heat transfer of a separator assisted dual-evaporators loop thermosyphon under heating power maldistribution condition

Thermosyphon loop with multiple evaporators has been widely used in the data centers thanks to its compact structure and local hot spot avoiding. However, under the heating power maldistribution condition, on account of the separation and convergence of flow steams from different evaporators, the instability of the system can be exacerbated which leads to premature failure and damage of servers. Thus, a separator assisted loop thermosyphon with dual evaporators had been proposed to improve its heat transfer performances and stability. An experimental study of the instability characteristics and heat transfer of the system are carried out under different filling ratios, cooling water temperatures and heating powers in this paper. To fully understand the interaction mechanism of two evaporators behind the oscillation, the visualization of flow patterns in the liquid tube and two-phase tube are observed as well. Results showed that during the oscillatory operation, since the evaporator2 (high heat load) circuit will circulate first, the variation of fluid temperature at evaporator1(low heat load) outlet lags others. Besides, when the inlet water temperature increases, the operating states of the system transfer from oscillatory operation to stable operation. When the filling ratio varies from 50 %-70 %, instability occurs in the system, the oscillation period increases from 115 s to 911 s and amplitude increases from 2.5 °C to 11.1 °C with an increase of the filling ratio. It should be noted that in our new system, it still can operate stably even when the heating power difference between two evaporators are as large as 350 W. In addition, alternatively oscillation in the two evaporators, maximum heat transfer coefficient of evaporator2 and minimum heat transfer coefficient of evaporator1 can be found when the heat load varies due to the mass competition between two circuits. According to above results, it is expected that the system could be beneficial to the application in the data center cooling.

Experimental study of hydrodynamics of groundwater flow in karst aquifers

ABSTRACT A large-scale laboratory model is constructed to simulate a karst aquifer. Three groups of experiments are conducted to study the matrix-conduit exchange and how head difference, conduit size, and matrix/conduit gradients affect it. The findings can be summarized as follows; (1) In experiments with higher matrix head, a counterintuitive head inversion, where conduit head becomes higher than matrix head, occurs when the matrix-conduit head difference is small (less than 9.3 cm and 11.5 cm for the 1-inch and 2-inch conduits, respectively); (2) Flow exchange is positively related to rescaled head difference (RSHD), a parameter combining the effect of matrix and conduit gradients and the matrix-conduit head difference; (3) When the conduit has the higher head, the flow exchange is on average 2.8 times higher than when the matrix has the higher head; (4) Doubling the conduit size doubled the flow exchange and increased the length of the head inversion.

Adaptive machine learning surrogate based multiobjective optimization for scavenging residual saltwater behind subsurface dams

High-fidelity physically based groundwater flow and solute transport models have been limited for seawater intrusion remediation design because of computationally intensive evolutionary algorithms. Data-driven machine learning approaches are promising to substitute expensive-cost groundwater numerical models within optimization due to computing efficiency. However, machine learning surrogates may accumulate error of forecasting and thereby result in infeasible optimal solutions. To achieve desired fidelity level, this study proposes a novel adaptive machine learning surrogate based multiobjective optimization method for coastal aquifer desalination. An adaptive modeling algorithm is newly introduced to iteratively retrain poorly-performed machine learning models and enhance predicting accuracy. The method is demonstrated to seek optimal extraction and injection strategies for scavenging residual saltwater trapped in an upstream aquifer behind a subsurface dam. Two conflicting objectives of minimizing the total extraction-injection rate and maximizing saltwater removal effectiveness are considered. Three machine learning models including artificial neural network (ANN), Gaussian process (GP) and response surface regression model (RSR) are developed to replace a high-fidelity seawater intrusion model for predicting chloride concentration and salinity mass. Non-dominated Sorting Genetic Algorithm II (NSGA-II) is employed to derive Pareto fronts. Pareto optimal solutions obtained from machine learning models are compared against those from the seawater intrusion model. Results indicate that the developed machine learning models do not only have strong predicting capability, but also maintain good quality of Pareto optimal solutions, while achieving substantial computational saving up to 95%. Especially, the adaptively retrained RSR model inhibits error accumulation of forecasting and accelerates correct convergence to the true Pareto front. The proposed method is found superior performance in accurate convergence, widely-spread diversity as well as high computing efficiency than conventional methods.