Groundwater Monitoring through Citizen Science: A Review of Project Designs and Results

Citizen science is increasingly prominent in the field of freshwater monitoring. Although there is a large body of literature related to surface water monitoring, less experience exists regarding citizen‐based monitoring of groundwater resources. This calls for a better understanding of the actual experiences of citizen science in the field of groundwater monitoring, including specific project designs and results. Based on a systematic review of 33 citizen‐based groundwater monitoring activities, we analyze (1) the design of citizen science projects, including their general project characteristics, institutional characteristics, citizen characteristics, and forms of interactions, as well as (2) their results, including their data outputs, outcomes on citizens, and impacts on problem‐solving. Results show that projects mainly have positive results on data quantity and quality and partly influence the citizens themselves and their contributions to groundwater‐related problem‐solving. Information on project design characteristics is scarce and mostly hints at the relevance of specific process mechanisms such as training and feedback. Based on these results, we suggest groundwater‐related projects to consider involving citizen scientists in monitoring activities in order to benefit research and citizens alike. Such involvement needs, however, careful design including relevant training to unfold its full potential for both sides. Researchers are advised here to rigorously report on both design and results to further improve citizen science practice in the future.


Introduction
Water demand doubles globally every 20 years due to a growing population and accelerating economy (Wada et al. 2011). Consequently, water scarcity has become a familiar problem in many parts of the world (Oki and Kanae 2006;Liu et al. 2017). Groundwater has often been utilized to relieve this water stress (Yang and Zehnder 2002). Almost a quarter of the world's current population relies on groundwater as their primary source of water (Alley et al. 2002) and a half or more of the water used for irrigation is supplied from underground (Siebert et al. 2010). Over time, overexploitation of groundwater becomes a growing problem, leading to adverse environmental effects such as the decline of groundwater levels, deepening of cones of depressions, salt water intrusion in the coastal regions, and even ecosystem deterioration (Mousavi et al. 2001;Konikow and Kendy 2005). Groundwater also gets polluted by nitrates and pesticides used in NGWA.org Vol. 61, No. 4-Groundwater-July-August 2023 (pages 481-493) modern agriculture, chlorinated or mineral hydrocarbons, and other soluble chemicals from industries and urban areas (Kinzelbach et al. 2003;Babiker et al. 2007). In addition to these, climate variability is also an important factor that directly impacts the quality and quantity of groundwater (Rivera et al. 2004;Joshi et al. 2020).
To effectively mitigate the issues mentioned above, integrated groundwater resource management is needed. The implementation of integrated groundwater management approaches relies on detailed data on water levels and water quality parameters. Regarding the latter, the most common are physical (pH, temperature, turbidity, electrical conductivity), chemical (total dissolved solids, total hardness, nitrates, sulfates, heavy metals), and biological (fecal coliform, total coliform) parameters (Adekunle et al. 2007;Longe and Balogun 2010;Sarala and Babu 2012). However, data on groundwater quantity and quality are rather sparse in many world regions, as a recent overview by the World Water Quality Alliance (WWQA 2021) has shown. In addition, the installation and management of monitoring wells, automated water level loggers, and sustaining these extensive monitoring programs for long periods are immensely resourceintensive (Manda et al. 2021).
Against this background, research increasingly calls for new monitoring approaches, including citizen science monitoring (Baalbaki et al. 2019;Manda et al. 2021;WWQA 2021). Citizen science is often defined as "(t)he collection and analysis of data relating to the natural world by members of the general public, typically as part of a collaborative project with professional scientists" (Oxford Dictionary 2020). The main idea is that citizens' involvement in research helps address the groundwater data shortage while also stimulating other outcomes, such as improving citizens' knowledge and awareness of groundwater problems. One pioneering example is the GET WET! (Groundwater Education through Water Evaluation and Testing) program which has recruited many different volunteers to create a database for groundwater quality in five New England states of the United States, serving both educational purposes and future water resource management (Thornton and Leahy 2012;Peckenham and Peckenham 2014).
But while the literature on citizen science in water monitoring is increasing exponentially, reported experiences mainly focus on surface waters, specifically rivers and lakes (San Llorente Capdevila et al. 2020;Fekete et al. 2021;Walker et al. 2021). If researchers report on their groundwater-related experience, they refer to single case studies only, which are highly sensitive to context (Thornton and Leahy 2012;Little et al. 2016;Baalbaki et al. 2019;Jamieson et al. 2020;Manda et al. 2021). If a broader perspective on groundwater is provided, research focuses on specific types of citizen science, such as crowdsensing and respective challenges (Speir et al. 2022). Consequently, there is, to date, no systematic understanding of groundwater-related citizen science projects, including their designs and results. Such a systematic understanding is, however, needed to judge citizen science's potential role in addressing the groundwater data shortage.
This review, therefore, aims to systematically summarizing the research project's experiences with citizendriven groundwater monitoring to identify key lessons learned and open questions. Our research focuses on three key questions: What are the basic characteristics of groundwater-related citizen science projects, including their location, duration, and parameters measured? How are citizen science projects further designed, including the characteristics of institutions and citizens involved as well as their forms of interactions? And, finally, what are the actual results of groundwater-related citizen science projects? To answer these questions, we conducted a systematic review of peer-reviewed scientific literature on groundwater-related citizen science projects, including 33 studies published between 2012 and 2022, and based on acknowledged categories in the literature.
The methods section provides a detailed overview of the methodological approach, including the identification and selection of scientific literature, the methods for data collection, and data analysis. The results section presents the results along with the three key questions, providing summarized information and examples from different studies. We then discuss the results and provide lessons learnt on the potential role of citizen science in groundwater monitoring, considering the specific requirements of the groundwater topic in contrast to other water resources such as rivers and lakes.

Methods
This paper systematically reviews citizen science projects in groundwater monitoring to understand project design and results. Our specific approach is a metaanalysis of cases, here defined as projects involving citizens in monitoring groundwater regardless of temporal or regional scales. As such, conceptual articles related to the potential role of citizen science in groundwater monitoring are excluded. This approach helps test the actual application of citizen science and its results regardless of any theoretical assumptions.
Citizen science projects were identified within peerreviewed scientific literature only to account for the scientific quality of citizen science project results. To identify relevant literature, a systematic search on the platforms SCOPUS and Web of Science was implemented on July 20, 2021 and cross-checked on the October 17, 2022. These platforms represent standard, but potentially complementing platforms for identifying scientific literature at the interface of the social and natural sciences. The search string included the keywords "groundwater" AND "citizen science" OR "voluntary monitoring" OR "communitybased monitoring," respectively. The search string was implemented within titles, keywords, and abstracts without limitations to years, regions, languages, and so on.
Based on this approach, we identified 103 articles, of which 54 appeared on SCOPUS and 49 on Web of Science between 2012 and 2022. The first screening of papers revealed 41 doublings between the results of the platforms or between articles referring to "citizen science," "voluntary monitoring," and community-based monitoring, resulting in a total of 62 papers after their exclusion. The second screening of titles and abstracts revealed that 35 articles did not employ citizen science in groundwater but referred to citizen science or groundwater monitoring more generally. Therefore, these articles have been excluded as well, resulting in a total of 27 papers for further analysis. The analysis of these papers and further snowballing has revealed another six articles related to the field of groundwater and citizen science, which have been included in the follow-up analysis, resulting in 33 papers in total for further analysis (Figure 1 and Table S1 for more detailed information).
Of these 33 short-listed papers (Table S2), 30 papers are journal articles and one paper is a book chapter, conference proceeding, or a project report, respectively. The papers are published in a wide variety of groundwater, hydrology, and environmental science journals, showing no specific trend to specific journals. The papers were published between 2012 and 2022, with an increasing interest in citizen science applications in groundwater since 2019 ( Figure 2). The first authors' affiliation shows that this interest has mainly been generated by North American and European institutions (16 from the United States, four from the UK, and one from Switzerland). In addition, there are four publications each from Africa and Asia, in collaboration with and funded by institutions from North America and Europe.
The 33 papers have been coded based on a multistep qualitative research process. In a first step, a coding scheme was developed, including acknowledged categories to describe citizen science projects, their design, and results, as identified in the field of water monitoring and beyond (also San Llorente Capdevila et al. 2020; Fekete et al. 2021;Kirschke et al. 2022). Main categories include (1) the general project descriptions including projects goals and scope, (2) citizen characteristics such as age, gender, or education, (3) institutional characteristics such as the type of organization and its motivation, (4) process mechanisms such as training and feedback cultures, as well as (5) project results including data outputs and further outcomes and impacts (Table 1).
The use of this coding scheme is reasonable as it builds on both literature-based and empirically substantiated categories in the field of citizen science, water, and environmental monitoring. More specifically, general project characteristics such as the location and duration of projects, the parameters measured, the regional scale, NGWA.org S. Nath and S. Kirschke Groundwater 61, no. 4: 481-493 Table 1 and the frequency of monitoring are basic information that researchers report upon by default in water monitoring projects and generally serve to contextualize the information on project results (San Llorente Capdevila et al. 2020). Institutional factors such as the institutional motivation laid down in project goals (e.g., research or awareness raising), the respective institutional responsibilities (e.g., academic or civil society organization), and the funding (e.g., amount and type of funding organization) may all influence the results of projects in terms of more research-oriented data outputs or more citizen-and problem-directed outcomes and impacts (MacPhail and Colla 2020; Kirschke et al. 2022). Similarly, research has often suggested that citizens and their characteristics, including their age, gender, motivation, and environmental awareness may differ in various context and, thus, also influence project results (Alender 2016;Jollymore et al. 2017;Mac Domhnaill et al. 2020). Process mechanisms such as recruitment mechanisms, training, tools for data collection, tools for data transfer, and communication and feedback are often considered key to project success, specifically with regards to the quantity and quality of data, but also for further engagement of citizens (Kosmala et al. 2016;Le Coz et al. 2016;Zhou et al. 2020).
Turning to results, data outputs including the quantity and quality of data are often a key motivation to involve citizens in research, but also a highly discussed topic (Lottig et al. 2014;Jollymore et al. 2017;Quinlivan et al. 2020). Outcomes in terms of citizen change is also often mentioned, including here as diverse aspects such as awareness raising regarding environmental issues, learning about science and nature, and even further engagement in research and politics (e.g., Chase and Levine 2018;Church et al. 2019;Peter et al. 2019). Also, the creation of further impacts such political advancement and problem solving is increasingly debated in the citizen science literature (Carlson and Cohen 2018;Pocock et al. 2019;Turbé et al. 2020).
In a second step, the 33 research articles were analyzed based on the coding scheme described above. This analysis included identifying text segments related to the mentioned codes using MAXQDA software-a licensed software for qualitative data analysis. The respective results have been cross-checked within the team of authors, resulting in slight adaptations of text segments in MAXQDA. As a result, we have identified 479 text segments related to 23 subcodes as a basis for further analysis.
The follow-up analysis of text segments included three consecutive steps. First, we assessed the quantitative appearance of codes to receive a general overview of the density of information related to citizen science design and results in the field of groundwater monitoring provided in the case studies. The appearance of codes was assessed based on the number of text segments per code and subcodes, counting the appearance once per paper only. Second, the qualitative information provided in the text segments was summarized and cross-checked by the team of authors. Respective results have been further outlined in text format, tables, and figures as represented in the "Results" section.

Quantitative Overview of Text Segments
Text segments were counted once per paper per code to get an overview of the information available for each subcode and main code. Figure 3 shows the number of text segments identified for each subcode; Figure 4 further summarizes the information provided in Figure 3 along the main codes. Together, these figures show that information on the design and the results of citizen science projects is distributed unequally between the main codes and subcodes. On the one side of the spectrum, information on citizen characteristics is particularly scarce, including basic characteristics such as the age of citizens and their gender, as well as their education, awareness, and motivational background. Likewise, information on the results of citizen science projects is relatively scarce, with a more considerable amount of information available on direct data outputs (quantity and quality) and far less information available on outcomes (effects of citizen science projects on citizens) and impacts (effects on actual problem-solving). Compared to these less represented categories, the codes of institutional characteristics and process mechanisms are slightly more prominent. In terms of institutional characteristics, information focuses on basic institutional aspects such as institutional responsibilities and funding. Information on process mechanisms relates to training  rather than to tools for data collection and data transfer or further communication and feedback. Finally, and on the other side of the spectrum, most information is available on basic project characteristics, including project goals and the project's location.

General Project Information
Information on project location is available in all 33 cases. The distribution of cases is quite varied, with the maximum number of cases being from North America (13, out of which 12 are from the United States) and from Africa (10, out of which seven are in Ethiopia). There are also four cases from Asia and two cases from Australia, Europe, and South America respectively. It is to be noted that out of the seven Ethiopian cases, five correspond to the same region of Northwest Ethiopia and research project (Walker et  Information on project duration is available in 26 out of 33 cases. The study period ranges strongly from a few days to 5 years. Typically, however, citizen science projects ran for a minimum period of 3 months up to 2 years. Information on parameters is available in 29 out of 33 cases. Citizen science projects measure both water levels (18 cases) and water quality parameters (14 cases). In terms of water quality, projects measure physical, chemical, and biological parameters. Among these water quality parameters, pH, electrical conductivity, and nitrate concentrations analyzed most often. The biological water quality parameters of total coliform and fecal coliform are additionally measured in two cases. Also, in three distinct cases in the United States, two studies were carried out to test the presence of heavy metals in groundwater (arsenic, cadmium, uranium, etc.) in mining areas (Babich et al. 2021;Webber et al. 2021) and one study was performed to analyze different PFAS compounds present in private wells near bio-solids application sites (Alam et al. 2022).
Information on the regional scale is available for 28 out of 33 cases. The distribution and total number of groundwater wells is typically rather low but varies a lot between the cases, from five monitoring wells (Walker et al. 2016;Gowing et al. 2020) to 313 sampling sites (Alther et al. 2021). The number of wells depends here upon the resources available and the goal and nature of the study.
Information on the frequency of monitoring is available in 19 out of 33 cases. In most cases, data was collected on a weekly basis (Manda and  data was collected daily (Ferede et al. 2020;Gowing et al. 2020), and there are examples of data collection once in 2 days (Walker et al. 2016), monthly (Prajapati et al. 2021a;Prajapati et al. 2021b), and once in 2 months (Little et al. 2016). In three cases, the frequency of data collection is not mentioned explicitly, but the total number of readings taken is mentioned (Penningroth et al. 2013;Dawson et al. 2019;Jamieson et al. 2020;Babich et al. 2021;Webber et al. 2021). It is to be noted here that the frequency of data collection depends upon which parameters are being measured and whether they are being measured automatically or manually. Water level data can be checked daily, but this makes less sense for water quality data as the variation might be significantly less. Also, collecting data manually on a daily basis might be deterring for the citizen scientists.

Institutional Characteristics
Information on project goals is available in all 33 cases. The projects' goals are quite varied, but most of them can be categorized into mainly two groups. The first group of cases is rather data-oriented and is thus concerned with using citizen science data to validate or better understand certain models or phenomena where the conventionally available data is not enough or decreasing, for example, in data-scarce regions. Such data orientation may also help address a pertinent groundwater (quality) problem. Examples are the validation of groundwater models (Khadim et al. 2020), examining the potential extent of chloride concentrations in groundwater (Pieper et al. 2018), understanding the implication of variations in rainfall and land use on groundwater level fluctuations (Prajapati et al. 2021a), or assessing the extent of groundwater and marine inundation in response to future sea-level rise scenarios (Manda and Allen 2016). The second group of cases focuses more on citizen science, thus (1) generally promotes monitoring of groundwater through citizen science (Peckenham and Peckenham 2014;Little et al. 2016;Eastoe and Clark 2018;Ferede et al. 2020;Bernedo Del Carpio et al. 2021;Goldin et al. 2021;Gómez et al. 2021) or (2) aims at addressing issues of trustworthiness of citizen science data and at increasing the benefits and use of citizen-generated groundwater data (Thornton and Leahy 2012;Peckenham and Peckenham 2014;Walker et al. 2016;Khadim et al. 2020;Alther et al. 2021;Rigler et al. 2022). Sometimes, these two goals of addressing data scarcity and citizen involvement are explicitly present.
Information on institutional responsibilities was available in 30 out of 33 cases. In most cases (22 cases), there was one institutional actor responsible, whereas, in a smaller number of cases (8 cases), multiple institutions were responsible. Taken together, the institutions involved in the organization of citizen science activities are from research (27 cases in total), mostly representing universities (20 cases) and, partly, also nonuniversity research entities (9 cases). In comparison, nongovernmental organizations (NGOs) and public authorities are represented less (3 cases for governments, 5 cases for NGOs). In one case, there was no institution involved as the research activity was organized and published by a citizen scientist (Dawson et al. 2019).
Information on funding was available in 28 out of 33 cases. Funding is mostly provided by research entities through government research grants or by governmental agencies directly (27 cases). In some instances, there is also funding through NGOs (3 cases), the citizens themselves (1 case), or financial institutions (1 case).

Citizen Characteristics
Information on citizens involved is available in 27 out of 33 cases. The total number of citizens involved in these cases varies strongly and mostly depends upon the study area under consideration. However, most of the cases had up to 50 volunteers involved, with five exceptions only which are more large-scale and, thus, also demonstrate higher participation rates (e.g., Jamieson et al. 2020;Bernedo Del Carpio et al. 2021). In 9 out of 20 cases, high school or university students were involved. Other cases involved members of the community including households (Ferede et al. 2020;Bernedo Del Carpio et al. 2021;Alam et al. 2022), farmers (Bhatti et al. 2017;Goldin et al. 2021), local well managers (Alther et al. 2021), and even tourists, bushwalkers, and indigenous ranger groups (Davis et al. 2021) along with individually trained citizen scientists in two cases (Eastoe and Clark 2018;Dawson et al. 2019). Also, the number of citizens involved dropped by the project completion time (Little et al. 2016;Baalbaki et al. 2019;Manda et al. 2021;Prajapati et al. 2021b). Manda et al. 2021, for instance, mention a dropout rate of 25% (from 28 participants to 21 who finalized the project); Little et al. 2016 present similar results with a dropout rate of 20% (from 50 to 40 participants). According to Little et al. 2016 andPrajapati et al. 2021b, the reason for dropouts can be a loss of interest, people's moving, and changes in lifestyles, among others. However, Little et al. 2016 also highlight that many individuals were interested in continuing the monitoring for the long term and even after the completion of the project period.
Information on age is available in 9 out of 33 cases only. In these cases, projects typically involve younger but also elderly citizens, including school children of different grades, university students, middle-agers, and retirees (7 cases). Two projects include school children only.
Information on education is available in 12 out of 33 cases only. In these cases, projects involve rather welleducated citizens at the high school or university level (6 cases) or a mix of different educational levels from primary and middle school to relatively high educational levels such as high school or university level (6 cases).
Information on gender is available in 5 out of 33 cases only. In these three cases, both men and women were represented. However, in one case, there is a strong bias toward men (Prajapati et al. 2021b) and in another case, there was a strong bias toward women (Baalbaki et al. 2019). The bias on women is ascribed to the composition of the female research team and to the fact that the recruitment process was mostly organized through women's organizations, among others. In addition, Baalbaki et al. (2019) make an interesting point about the role of gender in citizen science monitoring, with male participants showing high interest in data collection and female participants showing strong interest in data analysis, which the authors explain with the local culture of indoor and outdoor working in the respective case.
Information on motivation is available in 11 out of 33 cases only. In most cases (7 cases), citizen scientists were motivated to learn more about water quality to improve water management in general or to respond to a specific or potential local problem. Particular examples of concerns regard the 'implications of groundwater on stormwater flooding and potable water supply' (Manda et al. 2021) and the potential contamination of private wells due to shale gas extraction (Penningroth et al. 2013) or poor sanitary conditions (Gómez et al. 2021). In four cases, citizens also received tokens of appreciation, such as an award for monitoring (Jamieson et al. 2020) or monetary compensation Bernedo Del Carpio et al. 2021;Rigler et al. 2022). Interestingly, two cases also highlighted the motivation of networking for increasing job opportunities through citizen science (Goldin et al. 2021;Rigler et al. 2022).
Information on awareness was available in 7 out of 33 cases only. In these five cases, different types of awareness have been reported, namely (1) (Wanda et al. 2017). Reported levels of awareness were different, with partly high awareness regarding the need for water level monitoring or slightly lower awareness when it comes to water quality problems or scientific methods.

Process Mechanisms
Information on recruitment is available in 23 out of 33 cases. The researchers adopted three main techniques for recruitment. The first approach was using their vast network and already established relationships to identify potential participants (Manda and Allen 2016;Manda et al. 2021;Prajapati et al. 2021b). The second approach was conducting informative workshops and sessions in which researchers described their project objectives and desired outcomes (Baalbaki et al. 2019;Haile et al. 2019;Goldin et al. 2021;Manda et al. 2021). The third approach was to reach out to broader and more diverse audiences through the use of local media such as newspapers and television (Pieper et al. 2018;Grace-McCaskey et al. 2019), advertisements and posters (Little et al. 2016;Manda et al. 2021), and social media platforms (Webber et al. 2021), or even by directly contacting the potential volunteers through telephone or e-mail (Thornton and Leahy 2012). Sometimes, it was essential to evaluate the volunteer's skills before the recruitment based on interviews or questionnaires (Wanda et al. 2017;Baalbaki et al. 2019;Ferede et al. 2020) and based on specific criteria set by the organizing institutions (Rigler et al. 2022).
Information on training is available in 25 out of 33 cases. Typically, projects conducted workshops and seminars to instruct the participating citizen scientists about (1) generally collecting water samples from wells (Babich et al. 2021;Webber et al. 2021;Alam et al. 2022 Information on tools for data collection is available in 22 out of 33 cases. The program organizers generally provided the citizen scientists with various tools and kits for monitoring. However, the specific tools provided depended upon the respective monitoring task. If the goal was to measure groundwater levels, citizen scientists were provided with data sheets (Manda and Allen 2016;Haile et al. 2019;Manda et al. 2021), electronic water level meters (Manda and Allen 2016;Eastoe and Clark 2018;Gowing et al. 2020;Manda et al. 2021;Rigler et al. 2022) or tape measures (Goldin et al. 2021;Prajapati et al. 2021a). If the goal was to collect water samples, citizen scientists received respective sampling kits (Wanda et al. 2017;Pieper et al. 2018;Alther et al. 2021;Davis et al. 2021;Webber et al. 2021;Alam et al. 2022). In some cases, citizen scientists were to monitor water quality parameters and were, thus, provided with respective starter kits or chemistry testing kits (Thornton and Leahy 2012;Peckenham and Peckenham 2014;Little et al. 2016;Jamieson et al. 2020). In a special case (Bhatti et al. 2017), citizen scientists were also provided with digital GPS enabled cameras to take photographs of a borehole that was to be monitored at their location.
Information on tools for data transfer was available in 19 out of 33 cases. Citizen scientists typically submitted their monitoring data and information to the researchers through websites and online portals (Thornton and Leahy 2012;Little et al. 2016;Bhatti et al. 2017 into a central online database acting as a repository (Cartwright et al. 2020;Prajapati et al. 2021a). Data were also reported through fax and e-mail (Little et al. 2016;Jamieson et al. 2020;Rigler et al. 2022), phone calls (Little et al. 2016;Gómez et al. 2021) and sometimes also through specific smartphone apps (Bernedo Del Carpio et al. 2021;Goldin et al. 2021). If samples were collected or if hard copy data sheets existed, these were collected through the postal service (Pieper et al. 2018;Jamieson et al. 2020;Alther et al. 2021;Alam et al. 2022). In one specific case (Little et al. 2016), there was a change in the transfer of data, from a reporting by telephone, fax, and e-mails at the beginning to a web-based portal at a later stage to increase efficiency. Information on communication and feedback was available in 19 out of 33 cases. In most cases, researchers communicated with the participants to keep them informed about the project's progress and to present the interim results (Thornton and Leahy 2012;Eastoe and Clark 2018;Baalbaki et al. 2019;Haile et al. 2019;Gowing et al. 2020). This communication helped to ensure that citizen scientists were continuously reminded of the importance of their activities and increased their engagement and interest in the project activities. Actual communication was based on various communication tools such as in-person discussions like seminars and focus group discussions (Eastoe and Clark 2018;Baalbaki et al. 2019;Haile et al. 2019;Gowing et al. 2020), telephone and e-mail communication (Manda and Allen 2016;Grace-McCaskey et al. 2019;Manda et al. 2021), social media platforms (Bernedo Del Carpio et al. 2021;Davis et al. 2021;Prajapati et al. 2021b), newsletters (Little et al. 2016), and SMS reminders (Jamieson et al. 2020;Prajapati et al. 2021b). In some cases, citizen scientists could also provide feedback through interviews or surveys (Thornton and Leahy 2012;Pieper et al. 2018;Ferede et al. 2020;Webber et al. 2021), which was beneficial to understanding if citizens faced any problems throughout the monitoring process.

Project Results
Information on data quantity is available in most cases (21 out of 33 cases). In these cases, researchers judge the amount of data collected by the citizen scientists rather positively since their contributions support data collection at larger spatial or temporal scales. Jamieson et al. (2020), for example, highlight that "the spatial reach of groundwater monitoring and its frequency have increased." The data collected mostly relate to water levels but also include groundwater quality data. An example is Alther et al. (2021), where citizen science data have revealed 'a previously undocumented groundwater fauna' (Alther et al. 2021, 1). Another example is Penningroth et al. (2013), where we find that almost two thousand wells in New York 'have been analyzed for at least one chemical' since the 1990s.
Information on the quality of data is available in most cases (21 out of 33 cases). Research describes both methods for quality control and the actual quality of data.
In terms of methods, we find a variety of forms, including the comparison of citizen science data with 'automated water level loggers' (Manda and Allen 2016;Manda et al. 2021), with data collected randomly by the scientists (Manda and Allen 2016;Manda et al. 2021), with previous monitoring data (Little et al. 2016;Jamieson et al. 2020;Prajapati et al. 2021b), with photographic records (Goldin et al. 2021;Prajapati et al. 2021b), or through 'phone call verification of methods' (Gómez et al. 2021). In exceptional cases, there was no 'validation against formal sources' possible as no official data existed (Walker et al. 2016). In one case (Dennis and Dennis 2019), data is collected through a mobile application that "introduces a type of block chain approach where all data is accepted but marked as low confidence until verified by a trusted user." In terms of quality, research finds that citizen science data are typically of high quality, with a number of small errors only (e.g., Peckenham and Peckenham 2014;Little et al. 2016;Manda and Allen 2016;Gómez et al. 2021;Webber et al. 2021;Rigler et al. 2022). However, there is also some variety in data quality depending on the parameter measured. Whereas chemical and physical data were rather of high quality and comparable to data collected by scientists (e.g., Peckenham and Peckenham 2014;Baalbaki et al. 2019), biological parameters seemed to be more prone to errors in one case (Baalbaki et al. 2019). In some cases, the high quality of data has been ascribed to a number of design factors such as training on how to collect data or standardized monitoring protocols (Cartwright et al. 2020;Goldin et al. 2021;Manda et al. 2021).
Information on outcomes is available in about half of the cases (18 out of 33 cases). The outcome that was mentioned most often is the actual learning of citizens in terms of groundwater systems (e.g., Manda and Allen 2016;Walker et al. 2016;Grace-McCaskey et al. 2019;Jamieson et al. 2020;Gómez et al. 2021), nature in general (Manda and Allen 2016;Haile et al. 2019), water quality (e.g., Baalbaki et al. 2019;Alam et al. 2022), or scientific issues and methods (Dawson et al. 2019;Jamieson et al. 2020;Webber et al. 2021;Rigler et al. 2022). This learning was partly extended beyond the actual participants in the project through the building of an educational website (Little et al. 2016). In some cases, particular outcomes such as a change in personal behavior (stop drinking water for safety reasons) (Pieper et al. 2018), community empowerment (Khadim et al. 2020), a 'sense of ownership of their resources' (Walker et al. 2016), a "sense of being a research partner as opposed to a subject" (Walker et al. 2016), or further outcomes such as political engagement (Baalbaki et al. 2019;Haile et al. 2019) were highlighted. In terms of political engagement, Baalbaki et al. (2019), for instance, shows that citizens "voiced their opinion and concerns to the local authorities during the public session." The same case further highlights sustainable monitoring and institution building, given the "formation of a locally elected water committee which will be responsible for continuous monitoring of the groundwater resources" (Baalbaki et al. 2019).
Information on the potential impacts of citizen science on water management processes was available in 14 out of 33 cases only. (Potential) impacts mainly regard better management of groundwater resources based on better data on groundwater quantity and quality, leading to more informed decision-making Walker et al. 2019;Cartwright et al. 2020;Ferede et al. 2020;Jamieson et al. 2020;Bernedo Del Carpio et al. 2021;Gómez et al. 2021;Prajapati et al. 2021b;Rigler et al. 2022). In one case, potential impacts on food security have been highlighted (Walker et al. 2016). Contrary to these potential positive impacts, one case highlights the lack of impacts on policies due to a lack of trust in data generated by students (Peckenham and Peckenham 2014).

Discussion
This study aimed at a systematic understanding of groundwater-related citizen science projects, including their designs and results, to assess citizen science's potential role in addressing the groundwater data shortage and improving groundwater management. The review should help identify key lessons learned and open questions.
Our systematic review has shown that citizen science is, in fact, a new topic in the field of groundwater, with only a small number of examples as compared with the rapidly increasing number of citizen science studies in freshwater monitoring more generally (Fekete et al. 2021). However, just as for the general water-related citizen science literature, citizen science seems to be a trend that has significantly increased in recent years, and Northern American and European research entities mostly drive that.
Furthermore, an analysis of general project design factors shows that citizen science projects are mostly located in the United States and in Africa, take place for a period of about 3 months to 2 years, including both water quantity and water quality parameters, in a relatively low number of wells, and at roughly a weekly basis. While this result shows many similarities with the broader citizen science literature in freshwater monitoring (San Llorente Capdevila et al. 2020), we find that African cases are somewhat overrepresented in the field of groundwater. It is to be noted though, that most of the Ethiopian cases correspond to the same region of Northwest Ethiopia and research project (Walker et al. 2016;Haile et al. 2019;Walker et al. 2019;Ferede et al. 2020;Gowing et al. 2020). This latter point may be ascribed to the particularly prolific publication by individual scientists involved and, thus, may not indicate a trend.
Turning to institutions' goals, we find that institutions mainly pursue two types of goals: data collection in datascarce regions and generally promoting citizen science, for example, by checking the quality of citizen science data.
These goals are pursued mostly by scientists, with a small number of public authorities or NGOs involved in this area of citizen science. In line with the focus on scientists, citizen science monitoring activities are mainly funded by government research grants or governmental agencies. This institutional focus seems slightly different from the general literature in citizen science and freshwater monitoring regarding the data focus and comparably low involvement of nongovernmental organizations (e.g., San Llorente Capdevila et al. 2020; Kirschke et al. 2022). One potential explanation is that this review considered scientific literature only, which is, arguably, rather the terrain of academic institutions than of governmental agencies and nongovernmental organizations which may use different forms of outlets such as blogs or briefs. Another potential explanation may be that groundwater monitoring seems more complicated as compared with surface water monitoring. Consequently, the lead of scientists may be more relevant in groundwater monitoring. In addition, groundwater monitoring may be less suited for raising public awareness than larger monitoring campaigns, such as in the field of plastics (Syberg et al. 2020).
In terms of the citizens involved, information was rather sparse, but if information was provided, it confirmed general trends in citizen science monitoring. Groups of citizens are relatively small and include all age groups, both genders, and rather well-educated citizens. Citizens are typically aware of water management problems and are eager to learn more about nature and science, which, however, does not prevent dropping numbers of participation over time.
Considering process mechanisms, we find that several means for recruitment have been used, that training is typically offered for citizen scientists, and that there exist a variety of tools for data collection and data transfer. This, again, resonates well with the diversity of approaches used in citizen science (San Llorente Capdevila et al. 2020).
Turning to the results of citizen science, we find that groundwater-related citizen science projects indeed emphasize an increase in data through citizen scientists both on temporal and spatial scales. In addition, quality tests implemented in most projects reveal that citizen science data are mostly of high quality, with a small number of errors compared to automated or researcher's data collection. In addition to these data outputs, citizen science activities have contributed to citizen change, which we generally understand as awareness raising regarding environmental issues, learning about science and nature, and even further engagement in research and politics (e.g., Chase and Levine 2018;Church et al. 2019;Peter et al. 2019). In our case of groundwater, citizen change most explicitly referred to learning about groundwater systems and science. In some cases, these learning outcomes have turned into more political engagement and a potential change in groundwater management strategies, for example, to address groundwater scarcity or poor groundwater quality.
This focus on data collection combined with potential outcomes on citizens and impacts on water NGWA.org S. Nath and S. Kirschke Groundwater 61, no. 4: 481-493 management resonates well with the broader literature on citizen science in freshwater monitoring (e.g., San Llorente Capdevila et al. 2020; Kirschke et al. 2022). However, the literature review has also revealed some significant differences between groundwater and surface water monitoring, with special hindrances such as "accidental contamination of the wells" in the course of water level monitoring (Little et al. 2016), the correct application of sophisticated groundwater monitoring techniques (Eastoe and Clark 2018), the converting of feet to meters, or from feet to decimal feet in groundwater level monitoring (Manda and Allen 2016), the "variables used to characterize the water level" which were in part counterintuitive to some citizens (Manda and Allen 2016), or "difficulties with accessing the groundwater monitoring wells because some of the locks got stuck in the locked position" (Manda and Allen 2016). This again adds to the literature focusing on the challenges of involving citizen scientists in groundwater monitoring (Speir et al. 2022).

Lessons Learnt
Our review has shown that citizen science in groundwater is an emerging topic only. Still, results indicate that citizen scientists can contribute to groundwater quantity and quality monitoring, although at lower scales and under the condition of certain design factors. What does this mean for scientists and practitioners interested in groundwater monitoring through citizens? Which lessons learnt can be derived?
First and foremost, the examples demonstrated in this literature review indicate that involving citizens in groundwater monitoring can be beneficial both for the scientists in terms of data outputs and for the citizens themselves, considering learning outcomes and potential subsequent changes in groundwater management. This should, in general, encourage scientists and practitioners to involve citizens in monitoring activities, lifting citizen science from a rather unrecognized concept in the field of groundwater monitoring to a well-tried, recognized, and more established tool for knowledge generation and exchange.
Second, our review shows that these potential benefits of citizen science for scientists and citizens are widespread and can include both countries from the Global North and the Global South, water quantity and water quality monitoring, including different types of institutions and citizens. If institutions are interested in initiating citizen science projects but bemoan if the capacity on site is appropriate, they should thus feel encouraged by best practice examples found in the literature in different monitoring settings.
Third, this is not to say that citizen science can be considered as a "magic" tool, solving governmental agencies' and scientist's concerns about lacking data. Citizen science in the field of groundwater is often limited to a small number of monitoring spots, a limited timeframe, and low frequencies, reflecting also the rather small number of actors involved. In the field of groundwater, it may thus be advised to formulate reasonable ambitions for citizen science projects, which go in line with the citizen's interests and time.
Fourth, institutions initiating citizen science projects should well consider comprehensive design needs adapted to their regional context, goals, and potential barriers. Institutions should schedule here enough time for the preparation of respective projects, the recruitment of citizens, trainings, and interaction with citizens. While hinting to time may sound mundane, it is key to prevent negative outcomes of citizen scientists in terms of data, learning, and motivation (Walker et al. 2021). Fifth, and considering these possibilities and design challenges of citizen science in the field of groundwater, we advise scientists to team up with public engagement officers of their institutions to prevent citizen science projects from any unforeseen pitfalls and, thus, to make citizen science projects a valuable experience for all those involved-be it the scientist or the citizens. Sixth, the above has shown that citizen science does not come for free, meaning that funding organizations should also provide the means to implement such projects. We note here that the types of resources may shift slightly from expensive technical equipment to human resources allowing for a well-designed citizen science project, including enough time for meaningful interaction among citizens and scientists, as well as tokens of appreciation for those involved. Funders may be encouraged here by both additional scientific outputs and resonant science-society interactions, potentially leading to citizen change toward learning and engagement in addressing local problems.
Seventh, we advise future citizen science projects to report more rigorously upon the characteristics of citizens, institutions, and the respective progress mechanisms to deepen learning on the effects of these design factors on project results in groundwater monitoring. In particular, data on citizen characteristics is very sparse and should be reported upon in citizen science publications related to groundwater. In addition, adverse outcomes of citizen science projects should be published just as rigorously as positive outcomes to enable future projects to learn from negative experiences and challenges as, for instance, reported in Walker et al. (2021) or in some of the literature we identified for this review (Little et al. 2016;Manda and Allen 2016;Eastoe and Clark 2018).
Such reporting and continuous monitoring may add to our key open question resulting from our literature review, which is about the potential impacts of such citizen science projects on advancing political decisions toward improved groundwater management. Has citizen science the potential to improve problem-solving through better data and citizen change? And which mechanisms are needed to come from better outputs and outcomes of science-society interactions to sustainable impacts? this topic during their study project on groundwaterrelated citizen science which has inspired us for this substantial review of the topic. The authors also wish to acknowledge the feedback provided at the European Citizen Science Association (ECSA) Conference 2022 and wish to thank the anonymous readers for their constructive comments on early versions of this manuscript.

Authors' Note
The authors do not have any conflicts of interest or financial disclosures to report.

Supporting Information
Additional supporting information may be found online in the Supporting Information section at the end of the article. Supporting Information is generally not peer reviewed.