论文信息
论文解读
AI 辅助整理- 方法
- PRISMA流程筛选103篇文献,系统分析ANN、RF、SVM、XGBoost、
- 发现
- 多技术融合可实现地下水水质预测、污染制图与智能决策,但数据、可解释性与传感器可靠性仍是瓶颈。
- 机会点
- 可探索XAI、数字孪生、边缘计算与联邦学习在农业地下水污染实时预警与灌溉水质决策中的落地。
论文摘要
地下水是支撑全球生活、农业和工业活动的重要资源。快速城市化、工业化、集约化农业实践以及气候变化,加剧了地下水污染与枯竭的风险。基于野外采样和实验室分析的传统监测方法往往耗时、成本高且时空覆盖有限,制约了及时评估和有效的地下水管理。人工智能(AI)、机器学习(ML)、地理信息系统(GIS)、遥感(RS)和物联网(IoT)的最新进展,为构建集成监测网络提供了新机遇,可通过高效的地下水质量评估、污染预测、污染物追踪和可持续水资源管理来保障长期水安全。按照基于PRISMA的筛选与合格性评估流程,共选取103项相关研究并对其进行系统分析,以评估AI、ML、GIS、RS和IoT应用在地下水质量监测与管理方面的最新进展。本综述对近期有关AI、ML、GIS、RS和IoT用于地下水监测、污染制图、健康风险评估和智能决策的研究与进展进行了更新概述。文章主要强调多种机器学习算法的应用,即人工神经网络(ANN)、随机森林(RF)、支持向量机(SVM)、极端梯度提升(XGBoost)和深度学习在地下水质量和水质指数(WQI)预测中的应用。本综述还重点关注利用GIS和RS进行空间分析、脆弱性评估、污染热点检测,以及通过基于IoT的网络开展地下水质量在线监测等方面。此外,还批判性地讨论了数据可得性、模型可解释性、传感器可靠性、网络安全和技术实施等主要挑战。文章还介绍了可解释AI(XAI)、数字孪生、边缘计算、联邦学习和智能决策支持系统(IDSS)等潜在未来应用。研究结论认为,环境科学、地理空间技术和人工智能的跨学科融合能够带来稳健、可扩展且智能的地下水。
全文
Asian Journal of ChemistryREVIEW
Emerging Soft Computing Technologies for Groundwater Quality
Monitoring, Prediction and Management: A Comprehensive Review1
SANTOSH TUKARAM SANAP1, , SHRIKANT BHAUSAHEB RANDHAVANE1, and SANDHYA SANTOSH SANAP2, ,*
1Department of Civil Engineering, SVKM NMIMS Global University, Dhule-424001, India
2Department of Artificial Intelligence & Data Science Engineering, SNJB’s Late Sau. K.B. Jain College of Engineering, Chandwad, Nashik-
423101, India
*Corresponding author: E-mail: sandhya.aghav070@gmail.com
Received: 5 August 2026 Accepted: 13 October 2026 Published online: 13 November 2026 AJC-22515
Groundwater is a vital resource for domestic, agricultural and industrial activities worldwide. Rapid urbanization, industria lization,
intensive agricultural practices and climate change have increased the risks of groundwater pollution and depletion. Conventional
monitoring methods based on field sampling and laboratory analysis are often time-consuming, costly and limited in spatial and temporal
coverage, restricting timely assessment and effective groundwater management. Recent advances in artificial intelligence (AI), machine
learning (ML), geographic information systems (GIS), remote sensing (RS) and the Internet of Things (IoT) offer new opportuni ties to
develop integrated monitoring networks to sustain long -term water security through efficient groundwater quality assessment, pollution
prediction, contaminant tracking and sustainable water resource management . Following a PRISMA -based screening and eligibility
assessment process, 103 relevant studies were selected and systematically analyzed to evaluate recent advances in AI, ML, GIS, RS and
IoT applications for groundwater quality monitoring and management This review provides an updated overview of various recent studies
and developments concerning the use of AI, ML, GIS, RS and IoT in groundwater monitoring, pollution mapping, health-risk evaluation
and intelligent decision-making. It primarily emphasizes on applications of various machine learning algorithms namely artificial neural
network (ANN), random forest (RF), support vector machine (SVM), extreme gradient boosting (XGBoost) and deep learning in
prediction of groundwater quality and water quality index (WQI). This review also focuses on aspects of spatial analysis using GIS and
RS, vulnerability assessment and detection of pollution hot-spot and on-line groundwater quality monitoring through IoT-based networks.
In addition, some main challenges such as data availability, interpretability of models, reliability of sensors, cyber security and technology
implementation are discussed critically. Potential future applications such as explainable AI (XAI), Digital Twin s, edge computing,
federated learning and intelligent decision support system (IDSS) are also presented . The study concludes that inter -disciplinary
integration of environmental science, geospatial technologies and artificial intelligence can bring about robust, scalable an d intelligent
groundwater.
Keywords: Groundwater quality, Artificial intelligence, Geographic information systems, Smart groundwater management.
INTRODUCTION
Groundwater is a vital component of global freshwater
resources and a major source of water for domestic supply,
irrigation and industrial applications [1]. Nearly half of the
global population depends directly or indirectly on ground -
water for daily water needs [2-5]. Population growth,
urbanization, industrialization, intensive agriculture and climate
change have placed increasing pressure on groundwater
resources, with deterioration in water quality emerging as a
This is an open access journal, and articles are distributed under the terms of the Attribution 4.0 International (CC BY 4.0) License. This
license lets others distribute, remix, tweak, and build upon your work, even commercially, as long as they credit the author for the original
creation. You must give appropriate credit, provide a link to the license, and indicate if changes were made.
major environmental concern [6-8]. From a chemical persp -
ective, contamination by fluoride, arsenic, nitrate, heavy metals,
pesticides, pharmaceuticals and other emerging pollutants poses
serious risks to human health and ecosystems [9-12]. Pro -
longed exposure to contaminated groundwater may cause
fluorosis, arsenicosis, kidney diseases, neurological disorders
and other adverse health effects [12,13].
Conventional groundwater monitoring relies on field
sampling and laboratory-based chemical analysis. Although
these methods provide reliable measurements of physico -
Asian Journal of Chemistry; Vol. 38, No. 11 (2026), 2839-2853
https://doi.org/10.14233/ajchem.2026.36967
2840 Tukaram et al. Asian J. Chem.
chemical properties and contaminant concentrations, they are
labour-intensive, time-consuming, expensive and dependent
on discrete sampling. Such approaches may fail to capture
seasonal variations, sudden contamination events and changes
in hydrochemical composition. Limited spatial coverage and
inadequate forecasting capabilities further restrict their effec-
tiveness in groundwater management [14,15]. Systematic
assessment of pH, electrical conductivity, total dissolved
solids, hardness, major ions, nutrients and potentially toxic
elements is essential for identifying contamination sources,
understanding geochemical processes and evaluating ground-
water suitability.
Recent developments in artificial intelligence (AI), machine
learning (ML), geographic information systems (GIS), remote
sensing (RS) and the Internet of Things (IoT) offer opport -
unities to integrate hydrochemical measurements with environ-
mental data for improved groundwater assessment. Machine
learning techniques can identify complex nonlinear relation -
ships among physico-chemical parameters, geological charact-
eristics, land -use patterns and contaminant concentrati ons.
Artificial neural networks (ANN), random forest (RF), support
vector machines (SVM), extreme gradient boosting (XGBoost)
and deep learning models have been applied to groundwater
quality prediction, contamination assessment and water quality
index (WQI) estimation [16-18]. These methods support the
interpretation of chemical datasets, identification of potential
pollution sources and prediction of groundwater quality
variations.
GIS and remote sensing facilitate spatial assessment by
integrating geological, hydrogeological, land-use and environ-
mental information. Their application enables the identifica -
tion of contamination hotspots, evaluation of land-use impacts
and preparation of groundwater quality maps across extensive
geographical areas [19]. Geostatistical techniques, particu -
larly inverse distance weighting (IDW) and kriging, are widely
used to estimate the spatial distribution of chemical consti -
tuents and delineate contamin ation-prone regions [20]. The
integration with hydrochemical data improves the interpre -
tation of spatial variations in groundwater composition.
IoT-based monitoring systems enable continuous, real -
time measurement of selected water quality parameters using
chemical sensors and automated devices. Wireless communi-
cation and cloud-based platforms facilitate rapid data collec-
tion, transmission, and analysis. Automated monitoring and
early-warning systems can detect abnormal changes in water
quality parameters and support timely intervention [21-23].
Similarly, the i ntegration with AI and ML models further
enables contamination prediction and informed groundwater
management.
Important challenges remain concerning hydrochemical
data quality, model interpretability, prediction uncertainty,
sensor calibration and reliability, data security and monitoring
network scalability [5,8]. Addressing these li mitations is
necessary to establish reliable and sustainable groundwater
monitoring frameworks. Existing research often examines
ML, GIS, RS and IoT separately, while fewer reviews integrate
these technologies into a unified framework encompassing
chemical characterization, contamination prediction, health
risk assessment, and decision -making. Applications of expl -
ainable artificial intelligence (XAI), digital twins, edge com-
puting and federated learning in groundwater management
remain comparatively limit ed. Integrating these emerging
technologies with chemical analysis and geochemical inter -
pretation can strengthen monitoring networks and sustain
long-term water security.
Methodology of review
The present study systematically reviews the growing
application of artificial intelligence (AI), machine learning
(ML), geographic information systems (GIS), remote sensing
(RS), and the Internet of Things (IoT) in groundwater quality
monitoring, contamination prediction and sustainable ground-
water management. A systematic review approach was adop-
ted to identify, critically evaluate and the prepare existing
research while minimizing selection bias. The review meth -
odology followed the PRISMA guidelines to maintain transp-
arency, consistency, reproducibility in literature identification,
screening, eligibility assessment and study selection [24-26].
The review primarily covers studies published between
2015 and 2026, with selected seminal publications from 2008
to 2014 included to establish the theoretical, methodological,
and technological foundations. Literature was retrieved from
six scientific databases such as Scopus, Web of Science, IEEE
Xplore, ScienceDirect, SpringerLink and Google Scholar.
The search stra tegy covered groundwater hydrochemistry,
water quality assessment, contamination monitoring, predic -
tive modelling, geospatial analysis and intelligent monitoring
systems. The several keywords and search phrases were used
for example, Groundwater quality monitoring; Groundwater
contamination prediction ; Artificial intelligence in water
quality; Machine learning groundwater; Deep learning ground-
water quality ; GIS groundwater contamination ; Remote
sensing groundwater assessment; IoT water quality monitoring;
Smart groundwater management ; and Water Quality Index
prediction.
Boolean operators (AND and OR) were used to combine
search terms, with database-specific syntax adapted to improve
retrieval accuracy. The search strategy targeted studies add -
ressing groundwater chemical composition, physico-chemical
parameters, contaminant distribution, pollution prediction,
and technology-assisted groundwater management [27,28].
Study selection criteria: Following the literature search,
the identified studies were screened using predefined inclu -
sion and exclusion criteria to ensure consistency and mini -
mize selection bias throughout the review process [27]. The
review included peer-reviewed journal articles and conference
papers published in English, primarily between 2015 and
2026, with relevant studies from 2008 to 2014 considered
where necessary to establish the theoretical, methodological,
and technological foundations. Eligible studies focused on
groundwater quality assessment, chemical characterization,
contamination prediction, monitoring or management using
artificial intelligence (AI), machine learning (ML), geographic
information systems (GIS), remote sensing (RS), the Internet
of Things (IoT), cloud computing, or related computational
approaches. Studies were required to provide adequate metho-
dological details and report relevant model evaluation metrics.
Publications addressing surface water quality, lacking suffi -
Vol. 38, No. 11 (2026) Emerging Soft Computing Methods for Groundwater Quality Prediction: A Comprehensive Review 2841
cient methodological information or not primarily concerned
with groundwater quality were excluded, along with duplicate
records, abstracts, editorials, non-peer-reviewed publications
and reports. These criteria helped maintain the relevance,
transparency and consistency of study selection.
Screening and eligibility assessment: A total of 724
records were initially identified through database searches of
Scopus, Web of Science, ScienceDirect, SpringerLink, IEEE
Xplore and Google Scholar. After removing 156 duplicate
records, 568 records remained for title and abstract screening.
Of these, 352 records were excluded as they were not directly
related to groundwater quality monitoring, contamination fore-
casting, AI technologies, GIS, remote sensing or IoT -based
groundwater management systems. The remaining 216 full -
text articles were assessed for eligibility. Of these, 113 articles
were excluded due to insufficient methodological details,
limited relevance to groundwater quality assessment, repli -
cated results, lack of peer review or a focus on surface water
rather than groundwater. The remaining 1 03 articles met the
inclusion criteria and were retained for further review and
synthesis. These studies were examined to investigate the
applications of AI/ML, GIS/RS and IoT techniques in ground-
water quality prediction, contamination -related health -risk
assessment and intelligent groundwater management. The
literature selection process comprised four stages viz. (i) iden-
tification of records through database searching; (ii) removal
of duplicate records; (iii) screening of titles and abstracts; and
(iv) full-text eligibility assessment, as illustrated in the PRISMA
flow diagram (Fig. 1).
Fig. 1. PRISMA-style flow diagram showing literature identification,
screening, eligibility assessment and final study selection process
adopted in the present review
Data extraction: Data were extracted from the selected
studies using a structured approach to facilitate systematic
comparison of the reported research findings. The extracted
information included publication year, country or study area,
groundwater quality parameters, AI/ML algorithms employed,
GIS and remote-sensing techniques, IoT monitoring compo -
nents, prediction performance metrics, key findings and
study limitations. In addition, the comparative assessment
considered dataset sources and sizes, prediction targets, model
validation strategies and the evaluation metrics reported in
individual studies, wherever this information was available.
For AI/ML-based studies, the analysis considered the algo -
rithms employed, data partitioning and validation procedures
and reported performance measures such as the coefficient of
determination (R 2), root mean square error (RMSE), mean
absolute error (MAE) and classification accuracy, where
applicable. GIS and remote -sensing studies were examined
in terms of spatial data inputs, groundwater-quality mapping,
contamination assessment and validation approaches. IoT
based studies were assessed in relation to monitored water
quality parameters, sensor systems, data-transmission appro-
aches and reported field -validation evidence. A qualitati ve
comparative analysis was adopted since the reviewed studies
differed in their datasets, groundwater characteristics, predic-
tion targets, modelling approaches and performance indicators.
Consequently, reported performance values were interpreted
in the context of the methods and validation procedures used
in each study rather than being treated as directly comparable.
Missing information was identified as not reported and
findings were interpreted according to the evidence available
in the original publi cations. The extracted information was
comparatively analyzed to identify common methodological
approaches, variations in reported performance, limitations
in validation practices and research gaps in the application of
AI/ML, GIS, remote sensing and IoT to groundwater quality
monitoring, contamination prediction, health-risk assessment
and intelligent groundwater management. The analysis further
examined challenges related to hydrochemical dataset avail -
ability, model generalizability, sensor reliability a nd techno-
logical integration. These findings provided a basis for identi-
fying research priorities and opportunities to develop more
reliable, chemically informed and integrated groundwater
monitoring systems [29].
Review scope and novelty: Existing reviews on the
groundwater quality have largely focused on contamination
sources, hydrogeochemical processes, health-risk assessment
or individual monitoring technologies. The present review
adopts an integrated, cross-disciplinary approach to examine
the applications of artificial intelligence (AI), machine learn-
ing (ML), geographic information systems (GIS), remote
sensing (RS), and the Internet of Things (IoT) in groundwater
quality assessment, contamination prediction and sustainable
management. Based on a critical analysis of 10 3 selected
studies, the review explores the complementary roles of these
technologies in hydrochemical monitoring, predictive model-
ling, spatial contamination assessment, health-risk evaluation
and d ecision support. Particular attention is given to the
chemical parameters investigated, datasets used, prediction
targets, validation procedures and reported performance indi-
cators, where available, to assess methodological differences
and limitations. Th e review examines challenges related to
data availability, model interpretability and generalizability,
sensor reliability, cybersecurity, technological integration,
and practical implementation. Emerging approaches, such as
explainable artificial intellig ence (XAI), cloud computing,
2842 Tukaram et al. Asian J. Chem.
edge computing, Digital Twins and federated learning, are
discussed in terms of their potential applications, limitations
and relevance to groundwater management, with due consi -
deration of the limited groundwater-specific evidence for some
technologies. The distinctive contribution of this review lies
in connecting chemical characterization, monitoring techno -
logies, predictive methods, spatial analysis and health -risk
assessment within a unified framework. The identified rese -
arch gaps and methodological challenges provide a basis for
defining future research priorities and developing reliable,
transferable and practically applicable groundwater monitor-
ing systems to sustain long-term water security.
Limitations of the review : Despite the efforts made to
produce an accurate, thorough and systematic review, certain
limitations should be acknowledged. This review included
peer-reviewed articles published in English between 2008
and 2026 and indexed in Scopus, ScienceDirect, Web of
Science, SpringerLink, IEEE Xplore and Google Scholar.
Although the overall review period covered 2008 -2026, the
primary focus was on studies published between 2015 and
2026, with earlier studies included where necessary to establish
methodological, the oretical and technological foundations.
Consequently, relevant studies published in languages other
than English, unpublished articles, technical reports and other
grey literature were excluded. The reviewed studies covered
diverse aquifer types, groundwat er quality parameters, data -
sets, modelling techniques and evaluation methods, which
may limit direct comparisons across studies.
Research objectives and scope: The continuous enhance-
ment of application of AI, ML, GIS, RS and IoT technologies
have led to considerable improvement of groundwater quality
monitoring, prediction of contaminations and management of
water resource [30]. As such, this review attempts to critically
discuss the utilization of emerging technologies in the assess-
ment of groundwater quality and its sustainable management,
based on the objectives of the present review as follows: (i)
Review of groundwater quality parameters and contaminants
indicator; (ii) Assessment of AI and ML methods in the
prediction of groundwater quality; (iii) Review of application
of GIS and RS in contaminant mapping, (iv) Examination of
IoT-based groundwater monitoring system and (v) Identi -
fying the research g aps and future scope in smart water
management. The present review covers peer -reviewed
studies published between 2008 and 2026, with a primary
focus on research published between 2015 and 2026. Relevant
earlier studies were included where necessary to pro vide
methodological, theoretical and technological foundations.
The reviewed studies address groundwater quality moni -
toring, contamination assessment, forecasting, predictive
modelling and smart groundwater management through the
application of AI, ML, GIS, RS and IoT technologies.
Groundwater quality parameters and contamination
indicators: Physico-chemical and biological characteristics
describe drinking, agricultural and industrial quality of ground-
water. Several parameters like pH, electrical conductivity,
total dissolved solids, total hardness, major ions, nutrients
and trace elements such as fluoride, nitrate, arsenic and heavy
metals are generally studied. All these parameter value could
possibly relate to the nature of geological or anthropogenic
processes which are influencing the groundwater chemistry.
Monitoring and assessment of these parameter values is an
integral part of establishing the quality of the water calcul -
ating Water Quality Inde x (WQI) establishing risk analysis
on human health and developing of machine learning -based
models for the prediction of these parameters [31-33].
Determination of groundwater quality is based on the contin-
uously monitored pa rameters for the efficient management
and control of contamination and for sustainable use of
groundwater [33-35].
pH: pH is the degree of acidic and alkaline nature of
water. The pH varies depending on the minerals content a nd
chemical reaction in the aquifer. The value of the pH reco -
mmended by WHO (2022) [5] for drinking water is between
6.5 and 8.5 other wises scaling, corrosion or bad taste,
variation of solubility of the metals, nutrients would occur in
water.
Electrical conductivity (EC): EC gives an indication of
the total amount of salts dissolved in water and is generally
used to represent the complete mineralization of groundwater.
High electrical conductivity values can be related to high
salinity and mineralization levels and are usually caused by
natural phenomena like weathering of rocks and man -made
causes like discharge of industrial wastes, drain water from
irrigation and infiltration of sea water [9,36].
Total dissolved solids (TDS): TDS refers to all the
inorganic and organic constituents present in a solution. A
high level of TDS in water causes water to taste unpleasant
and renders it unfit for domestic, agricultural or industrial
consumption. TDS in drinking water should preferably not
exceed 500 mg/L as recommended [9].
Total hardness (TH): TH is most closely related to
calcium and magnesium ions and will comm only form scale
within pipes, appliances and equipment; also, in higher con-
centrations the effectiveness of soaps and detergents is
decreased. The source of TH can be linked to the dissolution
of carbonate rock such as in limestone or dolomite aquifers
[37,38]. The status of groundwater should be checked for TH
as water quality when in use:
Fluoride: In the groundwater, fluoride is found naturally
due to weathering of fluoride bearing rocks through geolo -
gical processes. The benefit of fluoride consumption in the
optimum concentration is preventive tooth decay and dental
diseases. In higher concentration, fluoride can cause dental
fluorosis and skeletal fluorosis. The WHO recommended
drinking water with 1.5 mg/L fluoride level is safe for drink-
ing purposes [9,39].
Nitrate: Nitrate is one of the commonly found conta -
minants of groundwater in agriculture regions due to appli -
cation of fertilizer for agriculture, livestock waste and sewage
discharge. Excessive intake of nitrate concentration poses
severe health risks to children, leading to methemoglobi -
nemia, which is also termed as blue baby syndrome and the
WHO guideline value of drinking water quality for nitrate is
50 mg/L [9,12,40].
Arsenic: Arsenic is one of the significant human induced
health hazards, wherein groundwater acts as a source of drin-
king water. Chronic exposure to arsenic contaminated water
could lead to skin lesions, cardiovascular diseases and several
Vol. 38, No. 11 (2026) Emerging Soft Computing Methods for Groundwater Quality Prediction: A Comprehensive Review 2843
cancer types. The WHO guideline value of arsenic concentra-
tion for drinking water is 0.01 mg/L [9,41,42].
Heavy metals: Various toxic heavy metals such as lead
(Pb), cadmium (Cd), chromium (Cr), mercury (Hg), iron (Fe)
and manganese (Mn) contamination in the groundwater
through anthropogenic sources like industries, mining opera-
tions, agriculture and due to natural geological causes may
contribute to health disorders such as neurological disorders,
kidney diseases and carcinoma to human [37]. The examined
parameters relevant to contaminant detection, water quality
index (WQI) assessment, human health risk evaluation and
machine learning (ML) -based prediction of ground water
quality are presented, providing essential information for
evaluating groundwater suitability and supporting its safe and
sustainable utilization [37,43].
Table-1 presents the key groundwater quality parameters
along with their allowable limits and health or environmental
implication. These parameters are considered to be the key
determinants of the groundwater quality, assessment of con -
taminants and the predictors for the development of machine
learning model.
GIS and remote sensing applications for groundwater
quality assessment: GIS and remote sensing (RS) provide
significant information in the study of groundwater quality
assessment, contamination zones, sustainable water resources
management. By means of acquisition, integration, interpre -
tation, depiction of spatial information using GIS and RS
technology, groundwater status can be explored and sound
management plan can be devised by experts and research ers
[19,20,46,47]. GIS provides a powerful method to store,
analyze and interpret the groundwater quality information
obtained from field survey, monitoring well and laboratory
test. The spatial distribution of the groundwater quality
parameters as well as their relationsh ip with the other
environment-logical, hydrological and geological data can be
analyzed and hotspots contamination areas can be delineated
and estimate its spatial extent at local and regional level
[19,48-50]. Among the spatial interpolation methods widely
used in groundwater research, inverse distance weighting
(IDW), ordinary kriging, universal kriging and spline inter -
polation are commonly applied to estimate groundwater quality
parameters at unsampled locations and generat e continuous
spatial distribution maps [20,46]. Remote sensing, conversely,
presents land-use/land-cover conditions, land surface charac-
teristics, vegetation vigor, potential ground water recharge
zone, relation between surface water and ground water, all of
which impact the ground water quality significantly. Landsat,
Sentinel and MODIS satellite data are well established as
sources for analysis of ground water resources and contami -
nation issues over large areas [19,47,51,52]. Combining of
remote sensing data with GIS based analysis would enhance
the ground water vulnerability and contamination hazard
mapping. Several studies related to groundwater quality were
performed using GIS and Remote sens ing techniques to
detect contamination by different pollutants such as nitrate,
fluoride, arsenic, seawater intrusion and heavy metals, to
know about the vulnerability zones for further treatment,
management or remediation activities [53-55]. The newest
technologies have also improved studies on the quality of
groundwater by the integration of GIS, RS with AI. The asso-
ciation of machine learning algorithm with geo -referenced
information has assisted to estimate pollution hazar d and
identify polluted areas precisely and determine water vulner-
ability accurateness [18,56-58]. Such integrated methodo -
logies have also extensively used in the smart water manage-
ment plans to handle the massive data quickly in decision
support system [18,56-58]. However, by and large GIS and
remote sensing technologies are capable to act as a good tool
to monitor the groundwater quality, mapping and a nalyzing
the contamination risk, water management for the resources
in a sustainable manner. Further enhancement of existing water
management plan using future emerging technology and
smart system like IoT along with GIS and remote sensing can
also improve water management
Table-2 summarizes the major applications of GIS and
remote sensing in groundwater quality assessment, emphasi-
zing the role of spatial analysis, interpolation techniques, and
satellite-derived observations in contaminant distribution
mapping, groundwater vulnerability assessment and sustain -
able resource management.
Fig. 2 illustrates the GIS -based groundwater quality
assessment framework, beginning with data collection and
processing, followed by spatial interpolation, contamination
mapping and decision support for groundwater management.
IoT-based groundwater monitoring systems: The im-
plementation of the IoT into groundwater quality monitoring
has transformed conventional sampling-based approaches by
enabling continuous, real-time measurement of water quality
parameters. Traditional monitoring relies on the collection of
the groundwater samples and analyzing them in laboratories.
TABLE-1
LIST OF GROUNDWATER QUALITY INDICATORS, STANDARD PERMISSIBLE VALUES AND THEIR HEALTH IMPLICATIONS
Parameter Unit Permissible limit Health/environmental significance Ref.
pH – 6.5-8.5 Controls acidity, alkalinity and contaminant mobility [9,44]
Electrical conductivity (EC) S/cm No specific limit Indicates salinity and dissolved ionic content [9]
Total dissolved solids (TDS) mg/L 500 Affects taste and drinking water suitability [9,44]
Total hardness (TH) mg/L as CaCO3 500 Causes scaling and affects domestic water use [9,37,38]
Fluoride mg/L 1.5 Excess exposure may cause dental and skeletal fluorosis [9,39,45]
Nitrate mg/L 50 High concentrations may cause methemoglobinemia [9,12,40]
Arsenic mg/L 0.01 Carcinogenic and associated with chronic health effects [9,41,42]
Lead (Pb) mg/L 0.01 Neurological and developmental disorders [9,13,43]
Cadmium (Cd) mg/L 0.003 Kidney damage and toxic effects [9,13,43]
Chromium (Cr) mg/L 0.05 Potential carcinogenic effects [9,13,37]
2844 Tukaram et al. Asian J. Chem.
Although this approach provides accurate and reliable mea -
surements, however, it is time -consuming, labour-intensive,
and often unable to detect sudden changes in groundwater
composition [21,22,56,59]. IoT-based systems automate data
collection, transmission, storage and analysis, which improve
the monitoring efficiency and data accessibility.
An IoT-based groundwater monitoring system typically
comprises chemical sensors, microcontrollers, communication
modules and cloud-based data processing platforms. Sensors
measure physico-chemical parameters such as pH, electrical
conductivity (EC), total dissolved solids (TDS), temperature,
dissolved oxygen, and turbidity, alongside selected chemical
contaminants such as nitrate and heavy metals. The measure-
ments are transmitted to central servers through wireless
communication technologies su ch as Wi -Fi, GSM, ZigBee,
LoRaWAN and NB-IoT for subsequent processing and inter-
pretation [21,22,60]. Proper sensor selection and calibration
are necessary to maintain measurement accuracy, particularly
for specific ions and potentially toxic elements.
Cloud-based platforms enable data storage, visualization
and trend analysis, with automated alerts generated when water
quality parameters exceed predefined thresholds. Integration
of AI and ML models supports the detection of abnormal
changes in hydrochem ical parameters and the prediction of
potential contamination events, allowing timely intervention
[23,56,61]. Continuous data collection expands monitoring
coverage, lowers operational costs and provides researchers,
environmental authorities and policymakers with reliable
information for informed groundwater management [60,62].
Integrating IoT with AI, GIS and remote sensing further
strengthens groundwater monitoring by combining real -time
measurements, predictive analysis and spatial assessment.
Such integrated systems can identify unusual changes in water
quality, estimate contam ination risks and generate spatial
distribution maps to locate potentially polluted groundwater
sources. These capabilities offer considerable potential for
intelligent water management and digital environmental moni-
toring [16,56,63]. Despite these advantages, the practical implem-
entation of IoT-based groundwater monitoring faces challenges
related to sensor calibration, maintenance, measurement accu-
racy, communication reliability, network coverage, power supply,
cybersecurity and equipment costs. Current research focuses
on developing cost-effective, durable sensors, energy-efficient
monitoring devices, edge-computing systems and reliable data
processing techniques to improve the accuracy, security and
long-term performance of groundwater monitoring networks
[60,64-66].
Table-3 summarizes the main components of IoT -based
groundwater monitoring systems and their respective funct -
ions. The integration of sensors, communication networks,
cloud computing and data analytics enables real -time moni-
toring and informed groundwater management. Fig. 3 presents
the system architecture, covering data acquisition, wireless
transmission, cloud storage, data processing and decision support.
TABLE-2
LIST OF GIS AND REMOTE SENSING TECHNOLOGIES, THEIR APPLICATIONS AND
CORRESPONDING OUTPUTS IN GROUNDWATER QUALITY ASSESSMENT
Technology Application Output Ref.
GIS Groundwater quality mapping Spatial distribution maps [19,48,49]
GIS Contamination hotspot identification Risk maps [20,50]
IDW Spatial interpolation of groundwater parameters Distribution maps [20,46]
Ordinary kriging Spatial prediction and uncertainty assessment Continuous contamination surface [20,46]
Remote sensing Land-use and land-cover analysis Pollution source identification [47,48]
Landsat/sentinel imagery Groundwater recharge and vulnerability
assessment
Vulnerability maps [19,47,52,55]
GIS + Machine learning Groundwater contamination prediction Intelligent decision-support systems [18,57,58]
GIS + Remote sensing + AI Integrated groundwater management Risk assessment and monitoring framework [18,54,56]
Fig. 2. Flow chart of GIS-based groundwater quality assessment framework
TABLE-3
LIST OF ESSENTIAL HARDWARE AND SOFTWARE COMPONENTS USED IN IoT-BASED GROUNDWATER MONITORING
Component Function Examples Ref.
Sensors Measure water quality parameters pH, EC, TDS, turbidity sensors [21,22,60]
Microcontrollers Data acquisition and processing Arduino, ESP32, raspberry Pi [21,23]
Communication modules Data transmission Wi-Fi, GSM, LoRaWAN, ZigBee [21,22,60]
Cloud platforms Data storage and analytics AWS, Azure, Google Cloud [23,56,61]
Analytics engine Prediction and decision support AI/ML models [16,56,63]
Vol. 38, No. 11 (2026) Emerging Soft Computing Methods for Groundwater Quality Prediction: A Comprehensive Review 2845
Machine learning (ML) models for groundwater qua-
lity prediction: ML is an effective approach for assessing
groundwater quality and predicting contamination. These
models use historical physico-chemical parameters, contami-
nant concentrations, water quality index (WQI) values and
relevant environmental variables to predict groundwater quality
and assess potential health risks [16,30,67]. Most of the tradi-
tional statistics have difficulty in managing the complex non-
linear relationship between the hydrogeology, environment
and quality variables. However, machine learning could use the
countless amounts of data to analyze, identify obscure patterns
and build reliable predicted models for groundwater quality
assessment [16,30,68]. Groundwater quality prediction utilizes
the recorded values for physico -chemical parameters, cont -
aminant concentrations, WQI and possible health risks on the
basis of observed past and related environmental variables
[16,18,69]. Enhanced availability of ground water monitoring
data and computational power have led to the proliferation of
machine learning algorithms for groundwater investigations
[16,18,63].
Artificial neural networks (ANN): Among the differ -
ent kinds of ML models that have been successfully used to
predict groundwater quality (GWQ), ANNs are one of the
most frequently used. ANNs are designed with inspiration
from the structure of the human brain and have the ability to
capture complex nonlinear relationships between groundwater
quality parameters. Many investigations on using ANN for
prediction of different parameters of GWQ such as pH, TDS,
EC, fluoride, nitrate and WQI are available and most of them
have achieved high prediction accuracy [56,67,70,71]. On the
contrary to the high accuracy of the models, ANNs usually
need big data size and enough computation time in order to
train. However, overfitting and limited interpretability are the
additional limitations, which must be addressed before appl-
ying these models to GWQ management [59,71].
Random forest (RF): RF is an ensemble learning tech -
nique that improves predictive accuracy and model robust -
ness. It performs well with noisy and incomplete datasets and
has been applied to groundwater contamination classification,
water quality index (WQI) prediction, arsenic occurrence
assessment and groundwater vulnerability mapping [57,58,72].
RF can estimate feature importance and help in identify the
variables, which is the most strongly influence groundwater
quality and supporting result interpretation. Compared with
artificial neural networks (ANNs), RF generally requires less
parameter tuning and offers a more transparent modelling
process [72].
Support vector machines (SVM): SVM is a supervised
learning algorithm widely used for classification and regre -
ssion tasks. It is suitable for modelling high-dimensional and
nonlinear data and has been applied to p ollutant concen -
tration prediction, groundwater suitability classification and
water quality index (WQI) estimation [57,73,74]. SVM offers
good generalization performance, resistance to overfitting when
properly tuned and reli able predictions with small datasets
[73]. Its limitations include sensitivity to kernel selection, the
need for parameter optimization and high computational costs
when processing large datasets [75].
Extreme gradient boosting (XGBoost): The XGBoost
is an ensemble machine learning algorithm widely used for
environmental prediction. It captures nonlinear relationships
by sequentially correcting errors from previous iterations,
often achieving high predictive accuracy in groundwater
quality modelling [76-78]. Its key advantages include com -
putational efficiency, the ability to handle missing values and
feature selection through feature-importance analysis. These
capabilities make XGBoost suitable for groundwater quality
prediction, contamination assessment, water quality index
(WQI) estimation and environmental risk assessment [78].
Deep learning models: Deep learning uses neural net -
works with multiple hidden layers to identify complex patterns
in data. Convolutional neural networks (CNNs), recurrent
neural networks (RNNs), long short -term memory (LSTM)
networks and hybrid CNN –LSTM models are increasingly
used for groundwater quality predic tion and spatiotemporal
contamination assessment [68,79,80]. LSTM networks are
particularly useful for time -series forecasting, as they can
capture long-term dependencies in groundwater monitoring
data. Deep learning models can achieve high predictive accu-
racy, although their performance depends on data availability,
computational resources and careful hyperparameter tuning
[79,80]. Hybrid CNN-LSTM models can capture spatial and
temporal dependenc ies simultaneously, which improve the
groundwater quality forecasting [68,79,80].
Comparative evaluation of machine learning models:
The selection of an appropriate ML model depends on dataset
size, required prediction accura cy, computational resources
and interpretability. ANNs and deep learning models can cap-
ture complex nonlinear relationships, whereas RF and XGBoost
offer a practical balance between predictive performance,
computational efficiency and model interpretation. SVMs are
suitable for relatively small datasets and classification tasks
[72,73,78]. Ensemble and hybrid AI approaches have reported
improved performance in groundwater quality prediction
compared with individual models [67,71,77,81]. Integrating
ML with GIS, RS and IoT can strengthen groundwater moni-
toring and assessment [16,18,56]. These methods support con-
tamination prediction, WQI estimation and human health risk
assessment. Model performance varies with data quality,
hydrogeological conditions, spatial heterogeneity and para -
Fig. 3. Flow chart of IoT-based groundwater monitoring architecture
2846 Tukaram et al. Asian J. Chem.
meter selection. Careful model selection, uncertainty analysis
and independent validation are essential before applying ML
models to groundwater management. Advances in intelligent
technologies are expected to expand their role in environ -
mental assessment and decision-making [16,68,69,71].
Table-4 compares machine learning models used for ground-
water quality prediction, outlining their advantages, limita -
tions and applications. Fig. 4 presents the workflow for deve-
loping and evaluating these models, while Fig. 5 illustrates
their integration with GIS and IoT technologies for real -time
groundwater monitoring, spatial analysis and management
decisions.
AI-based health risk assessment: Continuous consum-
ption of contaminated groundwater leads to the formation of
various diseases like dental and skeletal fluorosis, methemo -
globinemia, neurological diseases, cardiovascular diseases ,
kidney failure, various cancers, etc. [9,12,41,45]. Therefore,
health-risk assessment has become an important part of ground-
water quality assessment and public health management [31].
Traditional methods of health risk assessment consider con -
centration of the contaminant and human exposure in the
calculation of non-carcinogenic risk (HQ, HI) and carcino -
genic risk (CR). The health risk (non-carcinogenic) should be
less than one (HQ < 1); otherwise, possible health effect is
occurred. However, carcinogenic health risk (of specific
compounds) must be less than the acceptable risk or a high
carcinogenic risk would occur [82,83]. The effectiveness and
accuracy of health risk assessment is considerably enhanced
with the utilization of sophisticated approaches based on AI
and ML. Health ris k models have capacity to manage enor -
mous volumes of water quality data, model non-linear relation-
ship between water quality parameters and predict health risk
more precisely as compared to conventional statistical models
[16,84,85]. Machine learning methods, such as random forest,
artificial neural networks, support vector machines and deep
learning, have been applied to predict health risk indicators,
classify pollution levels and identify population groups vulner-
able to g roundwater contamination [84-86]. AI combined
with GIS and remote sensing applications could improve the
spatial health risk assessment.
Using all the described methods, location and extent of
contaminated water as well as health risks among populations
are effectively identified, whereas health risk map provided
using GIS is used by policy makers to prioritize risk manage-
ment measures and to manage resources effectively [53,87,88].
In addition , IoT -based groundwater monitoring system has
ability to provide simultaneous, real-time and in situ measure-
ments on groundwater quality parameter. Integration of ML
with this IoT system would allow an early health risk warning
mechanism, which contribute to mitigating emerging health
risks [16,56,60]. Nevertheless, there were limitations to the wide
applicability of AI based health risk assessment such as data
availability, quantity of uncertainties and AI model transp -
arency have to be further improved. The application of explain-
able artificial intelligence (XAI) and an overall analysis of
uncertainties, combined with health-risk assessment platform
TABLE-4
LIST OF COMMONLY USED MACHINE LEARNING ALGORITHMS WITH THEIR APPLICATIONS,
STRENGTHS AND LIMITATIONS IN GROUNDWATER QUALITY PREDICTION
Model Major Application Advantages Limitations Ref.
ANN WQI prediction, contaminant
forecasting
Handles nonlinear relationships Overfitting, less interpretable [59,70,71]
Random forest Classification, contamination
prediction
High accuracy, feature importance
analysis
Computationally intensive for
large datasets
[57,58,72]
SVM Classification and regression Good performance with small datasets Parameter sensitivity and kernel
selection
[57,73,74]
XGBoost Advanced prediction modeling High predictive performance and
feature selection
Model complexity [76-78]
CNN Spatial contamination analysis Strong feature extraction capability Requires large datasets [68,79]
LSTM Time-series forecasting Captures temporal patterns effectively High computational cost [68,80]
Fig. 4. Machine learning framework for groundwater quality prediction
Fig. 5. Flow chart of Integration of AI, GIS and IoT for smart groundwater management
Vol. 38, No. 11 (2026) Emerging Soft Computing Methods for Groundwater Quality Prediction: A Comprehensive Review 2847
including environmental monitoring, machine learning, geo -
spatial analysis etc., will ensure the reliability of the system
and an effective decision -making can be implemented in
future [16,89,90].
Table-5 summarizes major groundwater contaminants, their
sources, associated health impacts and key risk indicators used
in public health assessment. Building on these parameters, Fig. 6
presents a conceptual framework for AI -based groundwater
health-risk assessment, linking groundwater quality monitoring
with machine learning analysis, health risk estimation and
decision support. This integrated approach supports the identi-
fication of contamination -related health hazards and helps
prioritize interventions to protect vulnerable population groups.
Decision support systems (DSS) for smart groundwater
management: Decision support systems play an important
role in groundwater quality assessment, contamination manage-
ment and sustainable water resource planning. They integrate
data acquisition, storage, analysis, visualization and decision
making tools to support policymakers, scientists and water
resource managers in addressing complex groundwater prob-
lems [92-96]. Integrating AI, ML, GIS, RS and IoT into DSS
enables large-scale monitoring data analysis, contamination
trend assessment and groundwater quality prediction [56,60,
94,97]. ML algorithms, s uch as ANN, RF, SVM and XG -
Boost, support the prediction of water quality parameters,
WQI and contamination risks, which allow to authorities to
plan timely interventions [16,30,76,94]. GIS -based maps
identify contamination hotspots and vulnerable areas, which
help to prioritize monitoring, remediation and resource allo -
cation [53,58,98]. IoT networks provide continuous ground -
water quality measurements, while cloud computing supports
centralized data storage, processing and access across moni -
toring locations [56,60,61]. Key challenges include data inte-
gration, sensor reliability, cybersecurity, model uncertainty
and interoperability among different technologies [65,89,99].
Addressing these limitations will strengthen AI -based DSS
and support informed groundwater management and long -
term water resource sustainability [56,60,94].
Table-6 outlines the key co mponents of smart ground -
water decision support systems and their functions. Fig. 7
shows their integration within a unified architecture, linking
IoT-based data acquisition with the AI-driven analysis, GIS-
TABLE-5
LIST OF MAJOR GROUNDWATER CONTAMINANTS, THEIR SOURCES,
ASSOCIATED HEALTH EFFECTS AND RISK ASSESSMENT INDICATORS
Contaminant Major source Health effects Risk indicator Ref.
Fluoride Geological formations Dental and skeletal fluorosis HQ, HI [9,45,88]
Nitrate Fertilizers, sewage Methemoglobinemia HQ [9,12,42,87]
Arsenic Geogenic processes Skin lesions, cancer HQ, CR [9,41,42,91]
Lead Industrial discharge Neurological disorders HQ, CR [9,13,37]
Cadmium Mining, industrial waste Kidney damage HQ, CR [9,13,43]
Chromium Industrial effluents Carcinogenic effects CR [9,13,16]
Fig. 6. Flow chart of AI-based health risk assessment framework
TABLE-6
LIST OF MAJOR COMPONENTS, THEIR FUNCTIONS AND RELATED
TECHNOLOGIES IN SMART GROUNDWATER DECISION SUPPORT SYSTEMS
Component Function Technology Ref.
Data acquisition Continuous groundwater quality monitoring IoT sensors [21,22,56,60]
Data storage Centralized data management and processing Cloud computing [56,61]
Data analysis Groundwater quality prediction and assessment Artificial intelligence and machine learning [16,30,76,77]
Spatial analysis Mapping, visualization and hotspot identification GIS and remote sensing [19,46,47,98]
Risk assessment Contamination and health-risk evaluation ML-Based risk models [53,84,85]
Decision support Sustainable groundwater management recommendations Integrated DSS platforms [92-96]
Fig. 7. Flow chart of smart groundwater decision support framework groundwater sensors (IoT)
2848 Tukaram et al. Asian J. Chem.
based spatial visualization and cloud computing t o support
groundwater quality assessment, contamination management
and sustainable resource planning.
Challenges and research gaps: Despite advances in AI,
ML, GIS, RS and IoT for groundwater quality assessment and
contamination prediction, several challenges limit their large-
scale implementation. Addressing these issues is essential for
developing reliable intelligent monitoring systems and sust -
ainable groundwater management strategies. Data availability
and quality remain major concerns for AI and ML a pplica-
tions. Groundwater monitoring data are often limited in many
regions due to sparse monitoring networks, incomplete records,
missing values, inconsistencies in sampling and data collec -
tion methods. These limitations can reduce model accuracy
and prediction reliability, which affect the effectiveness of AI-
based groundwater quality assessment systems [30,97,99].
Reliability and calibration of sensors: Water quality
sensors in IoT-based groundwater monitoring networks enable
continuous data acquisition, but their performance can be
affected by calibration errors, sensor drift, environmental
interference and device malfunction. These factors may com-
promise measurement accuracy and the reliability of subse -
quent predictions. Maintaining sensor stability over extended
periods is particularly challenging in harsh environments and
remote monitoring locations, where regular inspection and
maintenance may be difficult. The development of robust,
cost-effective and self -calibrating sensors is a key research
priority for improving the accuracy and long -term reliability
of groundwater monitoring systems [21,22,60].
Model interpretability: Although most of the high -
performance ML and DL models excel at the prediction tasks,
most of the state-of-the-art ML/DL systems function as “black-
box” models, making them not very acceptable and usable for
decision making. The requirements and necessity of having
trustworthy AI-based water quality prediction and manage -
ment system are increasingly becoming prominent, the
predictions outputs are often un explainable, Therefore, XAI
technique is expected to overcome this limitation [89,90].
Model transferability: Although AI/ML models have
shown promise in groundwater quality prediction, their trans-
ferability across different hydrogeological and climatic settings
remains a major challenge. Models developed for a specific
aquifer or region may lose predictive accuracy when applied
to areas with different geological formations, land-use patterns,
climatic conditions and pollution sources. Developing transfer-
able models requires multi -region datasets, cross -validation
across hydrogeological settings and domain adaptation tech -
niques to improve performance in new environments.
Model reliability depends on appropriate data partiti on-
ing and validation strategies. Data leakage occurs when test
data influence model training, preprocessing, feature selection
or hyperparameter tuning, leading to overestimated predict -
ive performance. Random data splitting may also provide mis-
leading r esults when observations from nearby monitoring
wells or repeated measurements from the same locations are
divided between training and test sets. Spatially independent
validation and chronological data splitting can provide a more
realistic assessment of model performance at unobserved
locations and during future periods. Overfitting remains a
concern when datasets are small, incomplete or imbalanced.
Regularization, control of model complexity and independent
testing are important for evaluating predictiv e reliability.
These methodological considerations help determine whether
reported model accuracy represents genuine predictive capa-
bility and whether the models can be applied reliably across
diverse groundwater systems.
Challenges in integration: Integrating AI, GIS, RS,
IoT, cloud computing and decision support systems remains
challenging due to differences in communication protocols,
data formats, software architectures and computational require-
ments. Developing standardized, interoperable frameworks is
essential for seamless data exchange, efficient processing and
reliable integration of these technologies in intelligent ground-
water monitoring and management systems [56,61,99].
Security and privacy concerns: Cloud-based platforms
and large volumes of data generated by IoT sensors pose signi-
ficant cybersecurity and privacy challenges in groundwater
monitoring systems. Data breaches, unauthorized access and
network vulnerabilities may compromise sensitive water
quality information and undermine management decisions.
Strong encryption, secure communication protocols, access
controls, regular security audits and reliable data storage
mechanisms are essential to protect monitoring networks and
maintain the integrity and confidentiality of groundwater data
[61,65].
Real-time decision support systems: Most of the exis-
ting groundwater quality monitoring systems focus on data
collection, storage, retrieval and visualization, with limited
capabilities for real-time automated decision-making and inter-
active resource management. Developing intelligent decision
support systems that integrate continuous monitoring, predic-
tive analytics, automated alerts and timely management recom-
mendations remains an important research priority [92-94].
Research gaps: The literature review identifies several
research gaps that need to be addressed to improve ground -
water quality monitoring and management: (i) limited integ -
ration of AI, GIS, RS, and IoT within unified monitoring
frameworks; (ii) insufficient application of explainable artificial
intelligence (XAI) to improve the interpretability of ground -
water quality predictions; (iii) inadequate real-time monitoring
networks capable of supporting large-scale applications; (iv)
limited investigation of Digital Twins for groundwater
modelling, simulation and management; (v) insufficient integ-
ration of uncertainty quantification methods such as Bayesian
modelling, Monte Carlo simulation, ensemble modelling and
probabilistic forecasting, to assess prediction reliability; (vi)
inadequate attention to cybersecurity and data privacy in smart
water management systems; and (vii) limited use of edge
computing for local data processi ng and real -time assess -
ment. Federated learning offers another potential approach
for collaborative model development across monitoring
networks while limiting the need to centralize sensitive data
[66,89,90,100-103]. Table-7 shows the major research chall-
enges and limitations in smart groundwater monitoring and
management, which identify the priority areas for developing
reliable, scalable and efficient intelligent groundwater mana-
gement systems.
Future research directions: Advances in AI, ML, GIS,
RS and IoT are expected to strengthen future groundwater
quality monitoring and management. Integrated intelligent
Vol. 38, No. 11 (2026) Emerging Soft Computing Methods for Groundwater Quality Prediction: A Comprehensive Review 2849
systems are needed to support real-time monitoring, contami-
nation forecasting, health-risk assessment and informed decision
making for sustainable groundwater resource management
[56,60,92,94]. Several emerging technologies offer potential
solutions to current limitations, although their practical effec-
tiveness requires further investigation.
Explainable artificial intelligence (XAI) is a promising
research direction for improving the interpretability of ML
models used in groundwater quality assessment. By identi -
fying the contribution of input variables to model predictions,
XAI can help researchers and groundwater managers under -
stand the factors influencing groundwater quality. Its reliability
must be assessed using suitable explanation methods and
groundwater-specific validation [89,90]. Digital Twin techno-
logy offers another potential approach through the creation of
virtual representations of groundwater systems. Integrating
Digital Twins with IoT sensor networks, GIS databases a nd
ML algorithms could support real -time simulation, contami-
nation forecasting, scenario analysis and adaptive groundwater
management [100,101].
Future research should focus on validating these frame-
works across different aquifer types and hydrogeological set-
tings. Linking real-time sensor observations with groundwater
flow and contaminant transport models could improve the
simulation of groundwater conditions and the predictio n of
contamination events. Scenario -based simulations may help
evaluate groundwater abstraction strategies, pollution control
measures, and remediation options. Reliable implementation
will depend on sensor data quality, interoperability, model
calibration, computational requirements and validation against
field observations. Quantifying predictive uncertainty under
changing climatic and environmental conditions is essential for
assessing model reliability. Validated Digital Twin systems
could support early-warning mechanisms, infrastructure main-
tenance, and evidence -based groundwater policy decisions
[100,101].
Edge computing offers opportunities to process moni -
toring data close to the point of collection, reducing comm -
unication delays and supporting timely assessment of ground-
water quality. Its suitability depends on device capabilities,
computational resources, network connectivity and the relia -
bility of locally generated predictions [66]. Federated learning
provides a potential framework for combined model develop-
ment when groundwater data are distributed across institutions
and cannot readily be shared. Although this approach reduces
the need to transfer raw data, its effectiveness depends on data
compatibility, communication requirements, model perfor -
mance across participating locations and appropriate privacy
safeguards. The groundwater-specific studies are needed to
establish its practical benefits and limitations [102,103].
Future research should prioritize unified AI -GIS-IoT
platforms that integrate real-time sensor data, spatial analysis,
machine-learning predictions, cloud computing and decision
support. Such platforms could improve contamination fore -
casting, identify the targeted and vulnerable areas and reme-
diation [56,61,98]. Smart city and smart village initiatives
should incorporate groundwater quality monitoring into their
environmental management programmes through sensor net-
works, cloud-based platforms, geospatial tools and AI-driven
forecasting [56,60,61]. Achieving these objectives will require
collaboration among hydrologists, geologists, environmental
engineers, computer scientists, data analysts and policymakers.
Field-based validation of integrated systems is essential to
establish their reliability, scalability and contribution to sus -
tainable groundwater management [100-103].
Table-8 summarizes emerging technologies and future
research opportunities in groundwater quality monitoring and
management, with emphasis on improving prediction accu -
TABLE-7
LIST OF MAJOR CHALLENGES, THEIR IMPACTS, AND FUTURE RESEARCH NEEDS IN SMART GROUNDWATER MANAGEMENT
Challenge Impact Future research need Ref.
Data scarcity Reduced prediction accuracy Large-scale monitoring databases [30,97,99]
Sensor reliability Poor data quality Self-calibrating sensors [21,22,60]
Black-box AI models Limited trust and adoption Explainable AI (XAI) [89,90]
Technology integration Interoperability issues Unified frameworks [56,61,99]
Cybersecurity Data vulnerability Secure IoT architecture [65]
Limited DSS capability Delayed decisions Real-time intelligent DSS [92-94]
Prediction uncertainty Reduced reliability of AI predictions Uncertainty quantification frameworks [89,90]
Real-time processing limitations Delayed contamination response Edge computing integration [66]
Decentralized data management Data privacy concerns Federated learning approaches [102,103]
Lack of virtual groundwater systems Limited scenario analysis capability Digital twin development [100,101]
TABLE-8
LIST OF EMERGING TECHNOLOGIES AND FUTURE RESEARCH OPPORTUNITIES
Emerging technology Potential application Expected benefit Ref.
Explainable AI (XAI) Model interpretation Improved transparency [89,90]
Digital Twins Virtual groundwater systems Real-time simulation [100,101]
Edge computing Local data processing Reduced latency [66]
Federated learning Distributed AI modeling Data privacy protection [102,103]
Smart IoT networks Continuous monitoring Real-time assessment [21,22,60]
AI–GIS–IoT integration Intelligent groundwater management Enhanced decision support [56,61,98]
Digital Twin + AI integration Predictive groundwater simulation Improved planning and management [100,101]
Edge-AI systems Real-time contamination detection Faster response and reduced latency [66,102]
2850 Tukaram et al. Asian J. Chem.
racy, model transparency, real-time monitoring and decision
making. Fig. 8 presents a future-oriented framework integra-
ting AI, GIS, IoT, Digital Twins, edge computing and fede -
rated learning for intelligent groundwater monitoring and
sustainable resource management. Table -9 compares the
principal technologies reviewed in this study, outlining their
strengths, limitations and major applications in groundwater
quality monitoring, contamination forecasting and intelligent
management.
Conclusions
Groundwater quality is under growing pressure from
urbanization, industrial activities, agricultural expansion, and
climate change, creating a need for more efficient monitoring
and management approaches. Conventional sampling and
laboratory analysis remai n important, but their cost, limited
monitoring coverage and lack of predictive capability restrict
their effectiveness for large-scale and continuous assessment.
This review examines the role of AI, ML, GIS, RS and IoT
in advancing groundwater quality monitoring, contamination
forecasting, health-risk assessment and resource management.
Geospatial technologies support the identification of conta -
mination hotspots and vulnerable areas, while ML models
such as ANN, RF, SVM, XGBoost and deep learning help
predict groundwater quality parameters, water quality indices
and contamination risks. IoT -based sensor networks enable
continuous data collection and timely alerts, while the integ -
ration of these technologies offers a practical foundation for
intelligent groundwater management. The widespread appli-
cation of these systems remains constrained by inadequate
datasets, sensor reliability issues, limited model transparency,
uncertainty in predictions, cybersecurity risks and difficulties
in integrating technologi es across different scales. Research
on XAI, Digital Twins, edge computing, federated learning,
and integrated AI–GIS–IoT platforms may help address these
limitations. Future efforts should prioritize field validation,
standardized data practices, model re liability and practical
implementation under diverse hydrogeological conditions.
This review provides a consolidated understanding of current
technological developments, their applications and unresolved
challenges. The insights can guide researchers, environmental
agencies, engineers and policymakers in designing effective
groundwater monitoring programmes and developing evidence
based management strategies to protect groundwater quality and
ensure the long-term sustainability of this essential resource.
ACKNOWLEDGEMENTS
The authors sincerely thank SVKM NMIMS Global
University, Dhule and all individuals whose valuable contri -
butions and support facilitated the successful completion of
this research.
CONFLICT OF INTEREST
The authors declare that there is no conflicts of interest
regarding the publication of this article.
DECLARATION OF AI-ASSISTED TECHNOLOGIES
During the preparation of this manuscript, the authors
used an AI -assisted tool(s) to improve the language and no
Fig. 8. Flow chart of future intelligent groundwater management framework
TABLE-9
LIST OF COMPARATIVE SUMMARIES OF REVIEWED TECHNOLOGIES
Technology Strengths Limitations Major applications Ref.
GIS Effective spatial analysis and
mapping
Depends on quality of
spatial data
Groundwater quality mapping,
vulnerability assessment
[19,20,46]
Remote sensing Large-area coverage and
rapid assessment
Indirect measurement of
groundwater conditions
Land-use analysis, recharge
assessment, contamination mapping
[47,52,55]
Artificial neural
networks (ANN)
Handles complex nonlinear
relationships
Overfitting, limited
interpretability
WQI prediction, groundwater quality
forecasting
[59,71,75]
Random forest (RF) High accuracy and feature
importance analysis
Computationally intensive
for large datasets
Contamination prediction,
groundwater potential mapping
[57,58,72]
Support vector machine
(SVM)
Effective with small and
high-dimensional datasets
Sensitive to kernel and
parameter selection
Classification and regression of
groundwater quality
[73,74]
XGBoost High predictive performance
and robustness
Model complexity and
tuning requirements
Water quality prediction and
contamination assessment
[76-78]
Deep learning
(CNN/LSTM)
Captures spatial and temporal
patterns
Requires large datasets and
high computational power
Time-series forecasting and
contamination prediction
[68,79,80]
IoT Real-time monitoring and
automated data collection
Sensor maintenance and
communication issues
Smart groundwater monitoring
systems
[21,22,60]
AI–GIS–IoT integration Comprehensive intelligent
groundwater management
Integration and
interoperability challenges
Decision support, risk assessment,
smart monitoring
[16,18,56]
Vol. 38, No. 11 (2026) Emerging Soft Computing Methods for Groundwater Quality Prediction: A Comprehensive Review 2851
data, results or scientific conclusions were generated by AI.
All analyses, interpretations and conclusions are the authors’
own. The authors reviewed and edited the content and take
full responsibility for the published work.
REFERENCES
- U. Lall, L. Josset and T. Russo, Annu. Rev. Environ. Resour., 45, 171
(2020);
https://doi.org/10.1146/annurev-environ-102017-025800 - T. Gleeson, Y. Wada, M.F.P. Bierkens and L.P.H. van Beek, Nature,
488, 197 (2012);
https://doi.org/10.1038/nature11295 - R. Taylor, B. Scanlon, P. Döll, M. Rodell, R. van Beek, M. Leblanc,
L. Longuevergne, Y. Wada, J.S. Famiglietti, M. Edmunds, L. Konikow,
T.R. Green, J. Chen, M. Taniguchi, M.F.P. Bierkens, A. MacDonald,
Y. Fan, R.M. Maxwell, Y. Yechieli, J.J. Gurdak, M. Shamsudduha,
D.M. Allen, K. Hiscock, P.J.-F. Yeh, I. Holman and H. Treidel, Nat.
Clim. Chang., 3, 322 (2013);
https://doi.org/10.1038/nclimate1744 - J. Famiglietti, Nat. Clim. Chang., 4, 945 (2014);
https://doi.org/10.1038/nclimate2425 - M. Giordano, Annu. Rev. Environ. Resour., 34, 153 (2009);
https://doi.org/10.1146/annurev.environ.030308.100251 - Y. Wada, L.P.H. van Beek, C.M. van Kempen, J.W.T.M. Reckman, S.
Vasak and M.F.P. Bierkens, Geophys. Res. Lett., 37, 2010GL044571
(2010);
https://doi.org/10.1029/2010GL044571 - X. Xie, J. Shi, K. Pi, Y. Deng, B. Yan, L. Tong, L. Yao, Y. Dong, J. Li,
L. Ma and C. Zheng, Annu. Rev. Environ. Resour., 48, 395 (2023);
https://doi.org/10.1146/annurev-environ-112321-114701 - Y. Zhang and P. Li, Environ. Sci. Pollut. Res. Int., 30, 51234 (2023);
https://doi.org/10.1007/s11356-023-26000-0 - Y. Jurczynski, R. Passos and L.C. Campos, Sustainability, 16, 7107 (2024);
https://doi.org/10.3390/su16167107 - D.J. Lapworth, N. Baran, M.E. Stuart and R.S. Ward, Environ. Pollut.,
163, 287 (2012);
https://doi.org/10.1016/j.envpol.2011.12.034 - M.E. Stuart, D.J. Lapworth, E.J. Crane and A. Hart, Sci. Total
Environ., 416, 1 (2012);
https://doi.org/10.1016/j.scitotenv.2011.11.072 - M.H. Ward, R.R. Jones, J.D. Brender, T.M. de Kok, P.J. Weyer, B.T.
Nolan, C.M. Villanueva and S.G. van Breda, Int. J. Environ. Res.
Public Health, 15, 1557 (2018);
https://doi.org/10.3390/ijerph15071557 - Y. Aryan, T. Pon, B. Panneerselvam and A.K. Dikshit, J. Water Health,
22, 235 (2024);
https://doi.org/10.2166/wh.2023.082 - R. Rojas, L. Feyen and A. Dassargues, Water Resour. Res. , 44,
W12418 (2008);
https://doi.org/10.1029/2008WR006908 - M. Sophocleous, Hydrogeol. J., 10, 52 (2002);
https://doi.org/10.1007/s10040-001-0170-8 - M.S. Hanoon, A.N. Ahmed, C.M. Fai, A.H. Birima, A. Razzaq, M.
Sherif, A. Sefelnasr and A. El-Shafie, Water Air Soil Pollut., 232, 411
(2021);
https://doi.org/10.1007/s11270-021-05311-z - S. Singha, S. Pasupuleti, S.S. Singha, R. Singh and S. Kumar, Chemosphere,
276, 130265 (2021);
https://doi.org/10.1016/j.chemosphere.2021.130265 - B. Rammohan, P. Partheeban, R. Ranganathan and S. Balaraman,
Sustainability, 16, 9848 (2024);
https://doi.org/10.3390/su16229848 - M.K. Jha, V.M. Chowdary and A. Chowdhury, Hydrogeol. J., 18, 1713
(2010);
https://doi.org/10.1007/s10040-010-0631-z - P.P. Adhikary, C.J. Dash, H. Chandrasekharan, T.B.S. Rajput and S.K.
Dubey, Arab. J. Geosci., 5, 1423 (2012);
https://doi.org/10.1007/s12517-011-0330-7 - M.V. Storey, B. van der Gaag and B.P. Burns, Water Res., 45, 741 (2011);
https://doi.org/10.1016/j.watres.2010.08.049 - Z. Rasin and M.R. Abdullah, IACSIT Int. J. Eng. Technol., 9, 24 (2009).
- R.W. Gullick, J. AWWA, 95, 58 (2003);
https://doi.org/10.1002/j.1551-8833.2003.tb10493.x - M.J. Page, J.E. McKenzie, P.M. Bossuyt, I. Boutron, T.C. Hoffmann,
C.D. Mulrow, L. Shamseer, J.M. Tetzlaff, E.A. Akl, S.E. Brennan, R.
Chou, J. Glanville, J.M. Grimshaw, A. Hróbjartsson, M.M. Lalu, T. Li,
E.W. Loder, E. Mayo-Wilson, S. McDonald, L.A. McGui nness, L.A.
Stewart, J. Thomas, A.C. Tricco, V.A. Welch, P. Whiting and D. Moher,
BMJ, 372, n71 (2021);
https://doi.org/10.1136/bmj.n71 - D. Moher, A. Liberati, J. Tetzlaff, D.G. Altman, The PRISMA Group,
PLoS Med., 6, e1000097 (2009);
https://doi.org/10.1371/journal.pmed.1000097 - B. Kitchenham and S. Charters, Guidelines for Performing Systematic
Literature Reviews in Software Engineering, EBSE Techn ical Report
EBSE-2007-001, Keele University and Durham University (2007). - H. Snyder, J. Bus. Res., 104, 333 (2019);
https://doi.org/10.1016/j.jbusres.2019.07.039 - Y. Xiao and M. Watson, J. Plann. Educ. Res., 39, 93 (2019);
https://doi.org/10.1177/0739456X17723971 - R.J. Torraco, Hum. Resour. Dev. Rev., 15, 404 (2016);
https://doi.org/10.1177/1534484316671606 - A. Mosavi, P. Ozturk and K. Chau, Water, 10, 1536 (2018);
https://doi.org/10.3390/w10111536 - F. Chen, L. Yao, G. Mei, Y. Shang, F. Xiong and Z. Ding, Water, 13,
783 (2021);
https://doi.org/10.3390/w13060783 - L. Pei-Yue, Q. Hui and W. Jian-Hua, J. Chem., 7, 451304 (2010);
https://doi.org/10.1155/2010/451304 - M. Bahir, S. Ouhamdouch, D. Ouazar and A. Chehbouni, Carbonates
Evaporites, 35, 27 (2020);
https://doi.org/10.1007/s13146-020-00564-x - A. Nayak, G. Matta and D.P. Uniyal, Environ. Dev. Sustain., 25, 14229
(2022);
https://doi.org/10.1007/s10668-022-02661-4 - P. Li, J. Wu and H. Qian, Earth Sci. Res. J., 18, 27 (2014);
https://doi.org/10.15446/esrj.v18n1.34048 - B. Sarkar, A. Islam and A. Majumder, Regional Stud. Marine Sci., 44,
101751 (2021);
https://doi.org/10.1016/j.rsma.2021.101751 - S. He and J. Wu, Expo. Health, 11, 125 (2019);
https://doi.org/10.1007/s12403-018-0289-7 - S. Akram and F. Rehman, J. Chem. Appl., 4, 4 (2018).
- Y.S. Solanki, M. Agarwal, A.B. Gupta, S. Gupta and P. Shukla, Sci.
Total Environ., 807, 150601 (2022);
https://doi.org/10.1016/j.scitotenv.2021.150601 - R.F. Spalding and M.E. Exner, J. Environ. Qual., 22, 392 (1993);
https://doi.org/10.2134/jeq1993.00472425002200030002x - J. Podgorski and M. Berg, Science, 368, 845 (2020);
https://doi.org/10.1126/science.aba1510 - P.L. Smedley and D.G. Kinniburgh, Appl. Geochem., 17, 517 (2002);
https://doi.org/10.1016/S0883-2927(02)00018-5 - M. Jaishankar, T. Tseten, N. Anbalagan, B.B. Mathew and K.N.
Beeregowda, Interdiscip. Toxicol., 7, 60 (2014);
https://doi.org/10.2478/intox-2014-0009 - S. Pandey, G. Mohapatra and R. Arora, J. Clean. Prod., 427, 139149
(2023);
https://doi.org/10.1016/j.jclepro.2023.139149 - S. Li, Y. Wei, J. Zhou, Y. Li, L. Qiao, D. You, Y. Jiang, Z. Jiang, X.
Wei, and X. Ma, Cell Rep. Med., 7, 102588 (2026);
https://doi.org/10.1016/j.xcrm.2026.102588 - P.K. Srivastava, D. Han, M. Gupta and S. Mukherjee, Hydrol. Sci. J.,
57, 1453 (2012);
https://doi.org/10.1080/02626667.2012.716156 - A. Chowdhury, M.K. Jha, V.M. Chowdary and B.C. Mal, Int. J.
Remote Sens., 30, 231 (2009);
https://doi.org/10.1080/01431160802270131
2852 Tukaram et al. Asian J. Chem.
48. S.S. Asadi, P. Vuppala and M.A. Reddy, Int. J. Environ. Res. Public
Health, 4, 45 (2007);
https://doi.org/10.3390/ijerph2007010008
49. J. Krishnamurthy, A. Mani, V. Jayaraman and M. Manivel, Int. J.
Remote Sens., 17, 1867 (1996);
https://doi.org/10.1080/01431169608948744
50. E. Sener, A. Davraz and M. Ozcelik, Hydrogeol. J., 13, 826 (2005);
https://doi.org/10.1007/s10040-004-0378-5
51. A. Jasrotia, A. Kumar and R. Singh, Arab. J. Geosci., 9, 304 (2016);
https://doi.org/10.1007/s12517-016-2326-9
52. N.S. Magesh, N. Chandrasekar and J.P. Soundranayagam, Geoscience
Front., 3, 189 (2012);
https://doi.org/10.1016/j.gsf.2011.10.007
53. A. Rashid, M. Ayub, Z. Ullah, A. Ali, T. Sardar, J. Iqbal, X. Gao, J.
Bundschuh, C. Li, S.A. Khattak, L. Ali, H.A. El -Serehy, P. Kaushik
and S. Khan, Int. J. Environ. Res. Public Health, 20, 2113 (2023);
https://doi.org/10.3390/ijerph20032113
54. O. Rahmati, A. Nazari Samani, M. Mahdavi, H.R. Pourghasemi and H.
Zeinivand, Arab. J. Geosci., 8, 7059 (2015);
https://doi.org/10.1007/s12517-014-1668-4
55. H.-F. Yeh, Y. -S. Cheng, H. -I. Lin and C. -H. Lee, Sustain. Environ.
Res., 26, 33 (2016);
https://doi.org/10.1016/j.serj.2015.09.005
56. S.R. Krishnan, M.K. Nallakaruppan, R. Chengoden, M. Iyapparaja, S.
Koppu, J. Sadhasivam and S. Sethuraman, Sustainability, 14, 13384
(2022);
https://doi.org/10.3390/su142013384
57. S.A. Naghibi, K. Ahmadi and A. Daneshi, Water Resour. Manage., 31,
2761 (2017);
https://doi.org/10.1007/s11269-017-1660-3
58. S.A. Naghibi, H.R. Pourghasemi and B. Dixon, Environ. Monit.
Assess., 188, 44 (2016);
https://doi.org/10.1007/s10661-015-5049-6
59. H.R. Maier and G.C. Dandy, Environ. Model. Softw., 15, 101 (2000);
https://doi.org/10.1016/S1364-8152(99)00007-9
60. C.Z. Zulkifli, S. Garfan, M. Talal, A.H. Alamoodi, A. Alamleh, I.Y.Y.
Ahmaro, S. Sulaiman, A.B. Ibrahim, B.B. Zaidan, A.R. Ismail, O.S.
Albahri, A.S. Albahri, C.F. Soon, N.H. Harun and H.H. Chiang, Water,
14, 3621 (2022);
https://doi.org/10.3390/w14223621
61. R. Buyya, C.S. Yeo, S. Venugopal, J. Broberg and I. Brandic, Future
Gener. Comput. Syst., 25, 599 (2009);
https://doi.org/10.1016/j.future.2008.12.001
62. J.O. Ighalo, A.G. Adeniyi and G. Marques, Internet of Things for
Water Quality Monitoring and Assessment: A Comprehensive Review,
In: Artificial Intelligence for Sustainable Development: Theory, Practice
and Future Applications, Studies in Computational Intelligence, Cham,
Switzerland: Springer, vol. 912, pp. 245-259 (2021).
63. M. Shaikh and F. Birajdar, Int. J. Sci. Res. Arch., 11, 502 (2024);
https://doi.org/10.30574/ijsra.2024.11.1.0105
64. M.E.E. Alahi and S.C. Mukhopadhyay, Sensors, 17, 1885 (2017);
https://doi.org/10.3390/s17081885
65. S. Sicari, A. Rizzardi, L.A. Grieco and A. Coen -Porisini, Comput.
Netw., 76, 146 (2015);
https://doi.org/10.1016/j.comnet.2014.11.008
66. W. Shi, J. Cao, Q. Zhang, Y. Li and L. Xu, IEEE Internet Things J., 3,
637 (2016);
https://doi.org/10.1109/JIOT.2016.2579198
67. V. Nourani, M. Alami and M. Aminfar, Eng. Appl. Artif. Intell. , 22,
466 (2009);
https://doi.org/10.1016/j.engappai.2008.09.003
68. F. Sham, A. El-Shafie, W.Z.B.W. Jaafar, S. Adarsh and A.N. Ahmed,
Arch. Comput. Methods Eng., 32, 3593 (2025);
https://doi.org/10.1007/s11831-025-10248-1
69. E.A. Hussein, C. Thron, M. Ghaziasgar, A. Bagula and M. Va ccari,
Algorithms, 13, 300 (2020);
https://doi.org/10.3390/a13110300
70. P. Coulibaly, F. Anctil and B. Bobée, J. Hydrol. Eng., 6, 367 (2001);
https://doi.org/10.1061/(ASCE)1084-0699(2001)6:5(367)
71. T. Rajaee, H. Ebrahimi and V. Nourani, J. Hydrol., 572, 336 (2019);
https://doi.org/10.1016/j.jhydrol.2018.12.037
72. L. Breiman, Mach. Learn., 45, 5 (2001);
https://doi.org/10.1023/A:1010933404324
73. V.N. Vapnik, Introduction: Four Periods in the Research of the Learning
Problem, In: The Nature of Statistical Learning Theory, Statistics for
Engineering and Information Science. New York, NY, USA: Springer,
pp. 1-15 (2000).
74. O. Kisi and K.S. Parmar, J. Hydrol., 534, 104 (2016);
https://doi.org/10.1016/j.jhydrol.2015.12.014
75. I.N. Daliakopoulos, P. Coulibaly and I.K. Tsanis, J. Hydrol., 309, 229
(2005);
https://doi.org/10.1016/j.jhydrol.2004.12.001
76. M. Akhlaq, A. Ellahi, R. Niaz, M. Khan, S.S. Sammen and M. Scholz,
Tellus, Ser. A, Dyn. Meterol. Oceanogr., 76, 177 (2024);
https://doi.org/10.16993/tellusa.4069
77. A. Elmotawakkil, N. Enneya, S. Bhagat, M. Ouda and V. Kumar, J.
Hydroinform., 27, 299 (2025);
https://doi.org/10.2166/hydro.2025.290
78. T. Chen and C. Guestrin, XGBoost: A Scalable Tree Boosting System;
In Proceedings of the 22nd ACM SIGKDD International Conference
on Knowledge Discovery and Data Mining Association for Computing
Machinery, pp. 785-794 (2016);
https://doi.org/10.1145/2939672.2939785
79. Y. LeCun, Y. Bengio and G. Hinton, Nature, 521, 436 (2015);
https://doi.org/10.1038/nature14539
80. S. Hochreiter and J. Schmidhuber, Neural Comput., 9, 1735 (1997);
https://doi.org/10.1162/neco.1997.9.8.1735
81. V. Sehgal, M.K. Tiwari and C. Chatterjee, Water Resour. Manage., 28,
2793 (2014);
https://doi.org/10.1007/s11269-014-0638-7
82. United States Environmental Protection Agency, Risk Assessment
Guidance for Superfund (RAGS): Volume I-Human Health Evaluation
Manual (Part E), U.S. Environmental Protection Agency: Washington,
DC, USA (2004).
83. United States Environmental Protection Agency, Risk A ssessment
Guidance for Superfund (RAGS): Volume I-Human Health Evaluation
Manual (Part F): Supplemental Guidance for Inhalation Risk
Assessment, U.S. Environmental Protection Agency: Washington, DC,
USA (2011).
84. A. Mohammadpour, E. Gharehchahi, M. Golaki, M . Gharaghani, F.
Ahmadian, S. Abolfathi, M. Samaei, M.G. Uddin, A. Olbert and A.
Mousavi Khaneghah, Results Eng., 27, 105421 (2025);
https://doi.org/10.1016/j.rineng.2025.105421
85. M.G. Uddin, M. Imran, A. Sajib, M. Hasan, M.T. Diganta, T. Dabrowski,
A. Olbert and M. Moniruzzaman, J. Contam. Hydrol., 261, 104307 (2024);
https://doi.org/10.1016/j.jconhyd.2024.104307
86. J. Chen, H. Gui, Y. Guo and J. Li, Int. J. Environ. Res. Public Health,
19, 12000 (2022);
https://doi.org/10.3390/ijerph191912000
87. J. Iqbal, C. Su, M. Wang, H. Abbas, M.Y.J. Baloch, J. Ghani, Z. Ullah
and M.E. Huq, Environ. Sci. Pollut. Res. Int., 30, 61606 (2023);
https://doi.org/10.1007/s11356-023-25958-x
88. A. Narsimha and V. Sudarshan, Appl. Water Sci., 7, 2717 (2016);
https://doi.org/10.1007/s13201-016-0489-x
89. A.B. Arrieta, N. Díaz-Rodríguez, J. Del Ser, A. Bennetot, S. Tabik, A.
Barbado, S. García, S. Gil-López, D. Molina, R. Benjamins, R. Chatila
and F. Herrera, Inf. Fusion, 58, 82 (2020);
https://doi.org/10.1016/j.inffus.2019.12.012
90. M.T. Ribeiro, S. Singh and C. Guestrin, Why should I trust you?:
Explaining the Predictions of any Classifier ; In: Proceedings of the
22nd ACM SIGKDD International Conference on Knowledge Discovery
and Data Mining, pp. 1135-1144 (2016);
https://doi.org/10.1145/2939672.2939778.
91. H. Brammer and P. Ravenscroft, Environ. Int., 35, 647 (2009);
https://doi.org/10.1016/j.envint.2008.10.004
92. D.J. Power, Decision Support Systems: Concepts and Resources for
Managers, Greenwood Publishing Group (2002).
93. J.P. Shim, M. Warkentin, J.F. Courtney, D.J. Power, R. Sharda and C.
Carlsson, Decis. Support Syst., 33, 111 (2002);
https://doi.org/10.1016/S0167-9236(01)00139-7
Vol. 38, No. 11 (2026) Emerging Soft Computing Methods for Groundwater Quality Prediction: A Comprehensive Review 2853
94. E. Turban, R. Sharda, D. Delen and D. King, Decision Support and
Business Intelligence Systems, Pearson, edn 9 (2011).
95. P.G.W. Keen and M.S. Scott Morton, Decision Support Systems: An
Organizational Perspective, Addison-Wesley (1978).
96. R.H. Sprague and E.D. Carlson, Building Effective Decision Support
Systems, Prentice Hall (1982).
97. A. Halevy, P. Norvig and F. Pereira, IEEE Intell. Syst., 24, 8 (2009);
https://doi.org/10.1109/MIS.2009.36
98. J. Malczewski, Int. J. Geogr. Inf. Sci., 20, 703 (2006);
https://doi.org/10.1080/13658810600661508
99. C. Batini, C. Cappiello, C. Francalanci and A. Maurino, ACM Comput.
Surv., 41, 1 (2009);
https://doi.org/10.1145/1541880.1541883
100. F. Tao, H. Zhang, A. Liu and A.Y.C. Nee, IEEE Trans. Ind. Inform.,
15, 2405 (2019);
https://doi.org/10.1109/TII.2018.2873186
101. A. Fuller, Z. Fan, C. Day and C. Barlow, IEEE Access, 8, 108952 (2020);
https://doi.org/10.1109/ACCESS.2020.2998358
102. B. McMahan, E. Moore, D. Ramage, S. Hampson and B.A. y Arc as,
Communication-Efficient Learning of Deep Networks from
Decentralized Data, Proceedings of the 20th International Conference
on Artificial Intelligence and Statistics (AISTATS), vol. 54, pp. 1273-
1282 (2017).
103. P. Kairouz, H.B. McMahan, B. Avent, A. Bellet, M. Bennis, A.N.
Bhagoji, K. Bonawitz, Z. Charles, G. Cormode, R. Cummings, R.G.L.
D’Oliveira, H. Eichner, S. El Rouayheb, D. Evans, J. Gardner, Z.
Garrett, A. Gascón, B. Ghazi, P.B. Gibbons, M. Gruteser, Z.
Harchaoui, C. He, L. He, Z. Huo, B. Hutch inson, J. Hsu, M. Jaggi, T.
Javidi, G. Joshi, M. Khodak, J. Konečný, A. Korolova, F. Koushanfar,
S. Koyejo, T. Lepoint, Y. Liu, P. Mittal, M. Mohri, R. Nock, A. Özgür,
R. Pagh, H. Qi, D. Ramage, R. Raskar, M. Raykova, D. Song, W. Song,
S. U. Stich, Z. Sun, A. T. Suresh, F. Tramèr, P. Vepakomma, J. Wang,
L. Xiong, Z. Xu, Q. Yang, F.X. Yu, H. Yu and S. Zhao, Found. Trends
Mach. Learn., 14, 1 (2021);
https://doi.org/10.1561/2200000083
论文原文与出版信息
Asian Journal of Chemistry