Priority contaminants for monitoring include salinity (usually monitored as electrical conductivity, EC), acidity (pH), major ions, nitrate, microbiological pollutants, some contaminants of emerging concern (CECs) and geogenic contaminants, including those mentioned previously (IAH 2017). However, a key question is: what groundwater quality parameters can be easily upscaled to a global assessment, given that many groundwater quality issues are local and considerable variations in hydrogeology exist across aquifer systems?
Challenges for assessing groundwater quality arise from the four-dimensional nature of flow systems. Groundwater systems are often highly heterogeneous, meaning that samples from wells in close proximity may produce very different results, especially if taken from different depths. Well construction may also affect groundwater quality data—for example, two wells of identical depth may produce contrasting results if one has a grouted upper well casing and the other does not. Furthermore, long transport times in many groundwater flow systems mean that groundwater pollution and rehabilitation may involve considerably longer timescales relative to surface-water contamination—for example, nitrate stored in the unsaturated zone may cause contamination of underlying aquifers for many decades (Ascott et al. 2017).
Aside from poor sampling or analysis procedures, including a lack of suitable analytical equipment in many regions, unrepresentative groundwater-quality data can result from sampling boreholes originally designed for monitoring water levels and whose location and construction may be unsuitable for monitoring groundwater quality. Boreholes for monitoring groundwater quality must be sited carefully and constructed to allow sample collection at different depths (Misstear et al. 2017). Construction materials and procedures for sampling and handling must be designed to avoid false positives or negatives. This is especially important when the contaminants are redox-sensitive, volatile, or present in trace concentrations such as CECs, which include pharmaceuticals, personal care products, per- and polyfluoroalkyl substances (PFAS), wastewater treatment products, nanoparticles and plastics, and remain largely unmonitored in groundwater (Lapworth et al. 2012).
Groundwater quality data are relatively scarce due to the often-limited public accessibility of data from national groundwater quality monitoring networks, which themselves are lacking in many countries. With exceptions such as the European Union (EU) Water Framework Directive (European Commission 2000), many national agencies are not required to make groundwater quality data available. Even if data are publicly available, questions arise about their reliability, representability and quality, unless an internationally recognized quality assurance process is employed.
For interpretation, the context of monitoring data needs to be considered, including the locations of boreholes or springs, borehole depths, and sampling protocols as well as laboratory analyses. Often this information is not available or is kept at different institutions and not at a central location. Public accessibility of groundwater quality data is further hampered by national restrictions that make data available only for research or multilateral reporting and assessment purposes; however, international norms exist that offer guidance on improving access, for example, the Aarhus Convention (Mason 2010).
There are opportunities to engage new approaches and existing or emerging tools in assessing groundwater quality, including citizen science, earth observation data, geophysical methods and improved sensors. Only the first two will be briefly described here.
Citizen science (CS), an innovative approach to monitoring groundwater quality, is based on direct collaboration between the general public and scientists. The objective is to use smart, yet cost-effective tools to generate water-related data to support existing monitoring systems; however, several attributes of the conceptualization of CS activities can affect its success, and results need to be validated independently (San Llorente Capdevila et al. 2020). The vast majority of CS applications for gathering groundwater quality data currently takes place in North America and Europe, though a growing number of citizen science groundwater-quality-monitoring programmes are being deployed elsewhere.
Satellite-based earth observations (EO) can produce proxies for groundwater contamination processes. Poulin et al. (2020) showed that EO on population density, road density, precipitation, temperature and landcover in Uganda and Bangladesh were strongly correlated with microbial contamination levels in shallow groundwater. Earth observations can provide variables for use in prediction modelling—for example, EO-derived information on anthropogenic activities such as landcover or population density can be used to predict nitrate, herbicide concentrations and salinity in groundwater (Anning et al. 2012; Stackelberg et al. 2012). Numerous studies have employed EO products with machine learning (ML) to develop models of geogenic groundwater contamination of arsenic (Podgorski et al. 2020) and fluoride; however, these and other ML-based approaches (e.g., deep learning) would require further development before effective use in large-scale prediction of anthropogenic-related contaminants such as salinity, chloride, microbes, other trace elements and trace organic compounds.