industrial, agricultural, and mining development, as well as urbanization and
groundwater-resource management (Chapman 1996; Schwarzenbach et al. 2010).
Among these activities, unsewered sanitation, wellhead contamination, accidental
spillage, agrochemicals, irrigation, and saline intrusion have serious impacts on the
quality of groundwater used for drinking purposes (Chapman 1996, Schwarzenbach
et al. 2010). Other activities responsible for groundwater contamination are land and
stream discharge of sewage and effluent, sewage oxidation of lagoons, sewer
leakage, landfilling, solid-waste disposal, highway drainage soak-away, process
water/effluent lakes, tank and pipeline leakage, well-disposal effluent, aerial fallout,
sludge and slurry disposal, wastewater irrigation, mine-drainage discharge, process
water/sludge lagoons, solid mine tailings, oilfield brine disposal, hydraulic disturbance, and recovering water levels (Chapman 1996; Schwarzenbach et al. 2010).
These activities can be triggered by a number of social, economic, environmental,
and political actions, as well as by variations in climate and socioeconomic level
(Singh 2017; Vörösmarty et al. 2000; Alcamo et al. 2007). It has been consistently
documented that anthropogenic activities are the major contributors to groundwater
pollution; for example, human-created compounds (such as Aldrin, Chlordane,
DDT, Dieldrin, Dioxins, Endrin, Furans, Heptachlor, Hexachlorobenzene, Mirex,
PCBs, and Toxaphene, popularly known as the “Dirty Dozen”) were particularly
targeted by the 2004 Stockholm Convention on Persistent Organic Pollutants
(Schweitzer and Noblet 2018). According to the World Health Organization
(WHO), nearly 11% of the global population still lacks a primary drinking-water
service, and approximately 29% of the worldwide population relies on unsafe
drinking water, which led to the loss of 1.2 million lives (WHO 2017; Ritchie and
Roser 2019). According to the International Agency for Research on Cancer
(IARC), both anthropogenically and naturally occurring chemicals in groundwater
are categorized according to their possible carcinogenic effects on humans into either
Group 1 (carcinogenic), Group 2A (probably carcinogenic), Group 2B (possibly
carcinogenic), Group 3 (not carcinogenic), or Group 4 (probably not carcinogenic)
(WHO 2017). Arsenic and fluoride are the two most common naturally occurring
chemicals. Nitrate (in both its NO 3
À and NO 2
À forms) is one of the agriculturally
produced compounds posing a significant threat to human health (WHO 2017).
Groundwater contamination is a global public health challenge, and the prediction
of groundwater contaminant presence would help in creating proactive mitigation
policies (Burri et al. 2019; Singh 2018). In the United States (U.S.) alone, more than
100,000 lifetime cancer cases may be linked to carcinogenic chemicals in drinking
water (Evans et al. 2019).
The use of various machine learning (ML), deep learning (DL), and artificial
intelligence (AI) techniques in developing prediction models based on
environmental data is highly trusted among scientists (Singh et al. 2018). However,
the application of these techniques to the development of groundwater contaminantprediction models is still in its rudimentary stages. Considering the great expense of
environmental data collection and laboratory analysis, ML, DL, and AI techniques
could be extremely useful for developing systems supportive of decision-making in
environmental-contamination monitoring. For example, it would cost nearly
72
S. K. Singh et al.
groundwater-resource management (Chapman 1996; Schwarzenbach et al. 2010).
Among these activities, unsewered sanitation, wellhead contamination, accidental
spillage, agrochemicals, irrigation, and saline intrusion have serious impacts on the
quality of groundwater used for drinking purposes (Chapman 1996, Schwarzenbach
et al. 2010). Other activities responsible for groundwater contamination are land and
stream discharge of sewage and effluent, sewage oxidation of lagoons, sewer
leakage, landfilling, solid-waste disposal, highway drainage soak-away, process
water/effluent lakes, tank and pipeline leakage, well-disposal effluent, aerial fallout,
sludge and slurry disposal, wastewater irrigation, mine-drainage discharge, process
water/sludge lagoons, solid mine tailings, oilfield brine disposal, hydraulic disturbance, and recovering water levels (Chapman 1996; Schwarzenbach et al. 2010).
These activities can be triggered by a number of social, economic, environmental,
and political actions, as well as by variations in climate and socioeconomic level
(Singh 2017; Vörösmarty et al. 2000; Alcamo et al. 2007). It has been consistently
documented that anthropogenic activities are the major contributors to groundwater
pollution; for example, human-created compounds (such as Aldrin, Chlordane,
DDT, Dieldrin, Dioxins, Endrin, Furans, Heptachlor, Hexachlorobenzene, Mirex,
PCBs, and Toxaphene, popularly known as the “Dirty Dozen”) were particularly
targeted by the 2004 Stockholm Convention on Persistent Organic Pollutants
(Schweitzer and Noblet 2018). According to the World Health Organization
(WHO), nearly 11% of the global population still lacks a primary drinking-water
service, and approximately 29% of the worldwide population relies on unsafe
drinking water, which led to the loss of 1.2 million lives (WHO 2017; Ritchie and
Roser 2019). According to the International Agency for Research on Cancer
(IARC), both anthropogenically and naturally occurring chemicals in groundwater
are categorized according to their possible carcinogenic effects on humans into either
Group 1 (carcinogenic), Group 2A (probably carcinogenic), Group 2B (possibly
carcinogenic), Group 3 (not carcinogenic), or Group 4 (probably not carcinogenic)
(WHO 2017). Arsenic and fluoride are the two most common naturally occurring
chemicals. Nitrate (in both its NO 3
À and NO 2
À forms) is one of the agriculturally
produced compounds posing a significant threat to human health (WHO 2017).
Groundwater contamination is a global public health challenge, and the prediction
of groundwater contaminant presence would help in creating proactive mitigation
policies (Burri et al. 2019; Singh 2018). In the United States (U.S.) alone, more than
100,000 lifetime cancer cases may be linked to carcinogenic chemicals in drinking
water (Evans et al. 2019).
The use of various machine learning (ML), deep learning (DL), and artificial
intelligence (AI) techniques in developing prediction models based on
environmental data is highly trusted among scientists (Singh et al. 2018). However,
the application of these techniques to the development of groundwater contaminantprediction models is still in its rudimentary stages. Considering the great expense of
environmental data collection and laboratory analysis, ML, DL, and AI techniques
could be extremely useful for developing systems supportive of decision-making in
environmental-contamination monitoring. For example, it would cost nearly
72
S. K. Singh et al.