Introduction

Groundwater is an important freshwater supply as it is typically more reliable than surface water, especially in semiarid and arid regions. Dependence on groundwater resources for agriculture is increasing (Siebert et al. 2010; Wada et al. 2010) and many groundwater basins globally are under considerable stress due to overdevelopment (Custodio et al. 2017; Gleeson et al. 2012; Steward et al. 2013; Werner et al. 2013). Interdisciplinary and participatory approaches are becoming increasingly recognised as critical to effective groundwater management (Barthel et al. 2017) and there is currently a strong movement from groundwater scientists worldwide towards addressing groundwater challenges and opportunities “…to ensure groundwater continues to benefit society today and in the future” (Global Groundwater Statement 2019). Effective management of groundwater systems under stress requires an understanding of the hydrogeology and water balance, as well as how humans interact with groundwater (e.g., extraction and use behaviour; Jakeman et al. 2016). Many components of a water balance, including evapotranspiration, recharge, and aquifer storage, are difficult to measure or determine with precision and must be estimated with field-based methods or models (Lowe et al. 2009b). Additionally, unmetered groundwater extractions are significant sources of uncertainty for water balances (Lowe et al. 2009a); which limits understanding of risks to the resource, particularly when long-term sustainability or recovery from stress is under threat.

Metering of groundwater extraction has become a self-evident best practice of integrated water resource management (Molle and Closas 2021) as it can provide high-quality volumetric extraction data with a measurement performance of ±5%, provided the meter is fitted and maintained in accordance with manufacturing standards (Molle and Closas 2021; DAWE 2021). However, groundwater metering is rare—metering is typically only required by government agencies where extraction must comply with licence or permit volume limits, for example for high volume use for irrigation in heavily stressed aquifers (Charalambous 2011; Molle and Closas 2021; Newman et al. 2018; Sinclair Knight Merz 2012; Torak and Painter 2011; Ursitti et al. 2018), as there are significant challenges with regard to the logistics, costs of installation and maintenance, acceptance by users, data collection, political will and enforcement (Molle and Closas 2021; Ursitti et al. 2018). Methods for estimating groundwater extraction that are more commonly utilised include simple water use coefficients (Lovelace 2009; Lowe et al. 2009b), or more complex agricultural/hydrological models that use satellite remote sensing to estimate crop or stock water demands (Cheema et al. 2014; Guerra et al. 2007; Singh et al. 2020; Kent et al. 2020). Other indirect approaches include inverse modelling to infer extraction rates from groundwater levels (Martinez-Santos et al. 2010; Shao et al. 2014), remote sensing to estimate changes in groundwater storage (Castano et al. 2010; Gemitzi and Lakshmi 2018; Liaqat et al. 2016), agent-based modelling to link extraction rates to factors affecting human decisions (Castilla-Rho et al. 2015) and statistical modelling to integrate different types of data to estimate extraction (Keir et al. 2019). Indirect metering, where groundwater extraction is estimated based on energy consumption (Dash et al. 1999; Frenzel 1985; Wang et al. 2020) or pumping hours (Arvin 1992) is another approach that has been used to estimate groundwater extraction but requires data on extraction rates to derive extraction estimates and is most useful for mains-electricity powered wells. The primary obstacle to improving estimates of groundwater use is the general lack of metered extraction data, which is crucial in validating estimation methods and reducing uncertainty (Levin and Zarriello 2013).

Groundwater extraction for stock and domestic use (stock and domestic extractions, also known as basic landholder rights in some Australian jurisdictions, are permitted to be used for drinking water for livestock and normal household purposes, including potable supply, watering a garden around the homestead, filling a swimming pool and general household purposes) is typically considered a small component of overall water use (Charalambous 2011; Larsen et al. 2014; Lowe et al. 2009b). However, studies increasingly show that stock and domestic users can have a significant impact on groundwater systems where they occur in large numbers. For example, in the Great Artesian Basin (GAB), Australia, cattle grazing is the main land use and groundwater extraction for stock and domestic use is estimated to represent 50% of the total water extraction (DAWE 2020). Further, as groundwater systems approach the limits of sustainable extraction and come under stress due to the effects of drought or new industries competing for water (e.g., the coal seam gas industry in the GAB’s Surat Basin), the need to account for all forms of extraction, including for stock and domestic purposes, becomes increasingly important for effective groundwater management (Bureau of Meteorology 2015; Larsen et al. 2014; Parsons Brinckerhoff 2011). Yet, bores (otherwise known as wells) for stock and domestic purposes are usually exempt from regulation and metering, and management relies on estimates of use (Larsen et al. 2014).

A review of the literature (summarised in Table 1) has shown that voluntary metering of a sample set of water users can be a cost-effective way to obtain groundwater extraction data in areas where metering is not compulsory. This can be very useful to inform the development of estimation methods and water conservation strategies but is relatively uncommon. Walker et al. (2021) offer a comprehensive review of around 550 papers on community participation in water science projects, and do not mention any programs involving voluntary metering of groundwater abstraction. Our own literature review, which included scientific journals and web searches, identified seven voluntary metering programs (Table 1). Only three of these programs related to voluntary groundwater extraction metering and none covered the prevalent groundwater extraction type in Australia—stock and domestic extractions.

Table 1 Summary of a literature review on voluntary metering programs
Full size table

This paper reports on a program of voluntary metering of groundwater extraction for stock and domestic use in the Surat Cumulative Management Area (Surat CMA), Queensland, Australia (the case study area). A review of the literature suggests this is the only program of its kind worldwide. The program was undertaken to improve understanding of the factors driving groundwater extraction for stock and domestic use and to support the development of lower cost estimation methods in the case study area. Previous attempts to estimate stock and domestic groundwater use in the area include the work by Singh et al. (2020), which used property grazing potential (based on remote sensing data) and the availability of other water supplies to estimate groundwater extraction for stock and domestic use, and the work by Keir et al. (2019), which used statistical modelling to integrate different types of data to estimate extraction. Both studies highlighted the need for a targeted metering program to collect groundwater extraction data for a representative subset of users to validate estimations and reduce uncertainties (Keir et al. 2019; Singh et al. 2020), emphasising the importance of the voluntary metering program reported here. Roll out of the program and use of the data for validation of estimation methods was not straightforward and presented many challenges. The objective of the paper is therefore to contribute to knowledge about the challenges of undertaking voluntary metering of privately owned bores in a remote area and how to overcome these challenges, following the view (e.g., Andreassian et al. 2010) that giving prominence to analysing challenges is key to advancing hydrological sciences. Specific aims of the paper are to discuss the challenges of and solutions to (1) recruiting participants, (2) installing and maintaining a metering network, and (3) interpreting and using the data collected. The paper has been structured to integrate ‘methods’ and ‘results’ into a ‘case study’ section, rather than using a conventional ‘methods-results’ structure, because it is the process of developing the case study that will be discussed and concluded upon. Methods and results are inseparable in this case.

Case study

The information and data presented here are updated as of May 2021 when the most recent field visit took place.

Case study area

The case study area for the voluntary metering program is the semiarid Surat CMA (160,000 km2), which is the Queensland government’s area of interest for assessing and managing cumulative impacts on groundwater resources associated with expanding coal seam gas and coal mine operations (Fig. 1). The Surat CMA largely overlaps with the geologically defined Surat Basin (Fig. 1) and includes aquifers of the GAB, the world’s largest artesian groundwater basin, and a nationally significant water resource asset that supports economic activity, groundwater dependent ecosystems and cultural heritage (Ordens et al. 2020). The 57 bores that were metered during the program were spread as far north as Rolleston, as far east as Dalby, and as far south and west as the boundaries of the Surat CMA, to provide data representative of the range of climatic, hydrogeological, and hydrological environments in the study area. The climate, hydrogeology and hydrology of the Surat CMA are described in detail in OGIA (2016) and (2019).

Fig. 1
figure 1

a Case study area the Surat CMA, b groundwater supply bores color-coded by purpose (adapted from Keir et al. 2019)

Full size image

The dominant land use in the Surat CMA is cattle grazing (70%; Keir et al. 2019). Cattle grazing generally occurs on large properties (1,810 ha on average, but may be up to 50,000 ha; based on data sourced from Keir et al. 2019). Most properties are remote from townships and not connected to town water. After cattle grazing, the main land uses in the Surat CMA are large-scale irrigated cropping, forestry, nature conservation, and gas extraction (Keir et al. 2019). Other land uses include urban, industrial, mining (mainly coal) and intensive animal production (feedlots; OGIA 2019).

The agriculture sector is the dominant user of groundwater in the Surat CMA. There are an estimated 18,840 stock and domestic bores within the case study area (Fig. 1) with groundwater being the only reliable source of water for many graziers. However, only 15 stock and domestic bores in the area were metered (less than 0.1%) prior to our voluntary program because metering is not a government requirement. The results of this program are being used to inform the estimation of groundwater extraction for stock and domestic use, which is a key input to the numerical groundwater flow model that is used for impact assessment in the Surat CMA (OGIA 2019; Singh et al. 2020).

Participant recruitment

Funding was available to meter 43 bores and so the recruitment process aimed to identify up to 43 properties spread across the Surat CMA with varying characteristics (e.g., in terms of property size, source aquifer, and agricultural use, focussing on stock and domestic use) to form a representative subset of groundwater users. A range of approaches were employed to identify bore owners willing to participate in the metering program. Consultation with relevant landholder and industry peak bodies and organisations within the study area was undertaken (including AgForce and Gasfields Commission Queensland), involving presenting at industry body meetings and community forums, and writing articles for newsletters to publicise the program. Queensland state government agencies, local government organisations and coal seam gas companies were contacted to request recommendations on suitable contacts. Prominent members of the community with an interest in water resource management, as well as local hydrogeologists and drillers, were also contacted to recommend contacts. Letters were sent to landholders seeking participants for the program, and local community events were attended to network with landholders. Towards the end of the recruitment process, cold calling of landholders was undertaken to improve spatial coverage of metering across the study area and ensure a representative subset of users were metered.

During the initial stages of the recruitment process, concerns were raised by bore owners regarding the confidentiality of data gathered. Participants were therefore offered the option that participation be confidential, meaning a participant’s identity, location and raw data are kept confidential and not disclosed to any third party outside the research team, while allowing the data collected to be used in an aggregated or unidentified form. A copy of the consent forms and information sheets provided to bore owners are provided in the electronic supplementary material (ESM) sections A and B, respectively. Participants were also offered complimentary groundwater quality analysis and access to the bore usage data collected. An example of a bore usage report sent to bore owners is provided in section C of the ESM. Interactions with bore owners during the recruitment process were guided by The University of Queensland ethics approval and a communications plan developed with an experienced hydrogeologist that had worked in the local area for many years.

The recruitment process identified 26 potential participants with approximately 135 bores on 46 properties. The participants included large pastoral companies, gas companies and family-owned businesses that operate multiple cattle properties in the area, as well as smaller family-owned businesses. The contribution from each recruitment method is summarised in Table 2.

Table 2 Participant recruitment methods
Full size table

The most successful recruitment method was referral, with referral by a local hydrogeologist most successful, followed by referral by a member of the community with an interest in water resource management and contacts provided by a local driller. Government agencies and organisations were unable to provide referrals due to confidentiality concerns.

Cold calling led to the recruitment of nearly half the participants, although this was a relatively time-consuming process as background work was required to identify properties with bores in areas of interest and to obtain owner contact details. Written materials, including a letter distributed to landholders in the study area and an article in an industry peak body newsletter, were unsuccessful at recruiting participants. Three participants expressed interest in being involved after hearing about the program via word of mouth.

Bore selection

Bores were selected for metering to represent a range of agricultural uses (focusing on the stock and domestic class) and source aquifers and provide spatial coverage of the study area. The suitability of bore infrastructure for installation of a metering device was a key consideration and was assessed during a property visit where possible. Due to the remoteness of many properties, the suitability of bore infrastructure was often assessed through phone discussions with landholders and a review of photos of the bore and associated pipework. The accessibility of the bore for ongoing site visits, including the condition of access tracks, was also considered.

The metering network design was reviewed in 2017 and changes were made to the bores selected for metering. The revised network focused entirely on bores used for stock and domestic purposes and relocated some meters to properties where only a portion of all operational bores were metered, allowing for the total groundwater extraction to be metered at the property-level (the principal level at which water was managed). Bores that were difficult to access or were rarely used were removed from the metering program.

Meter selection

Options for metering technology, including direct and indirect metering options, were reviewed and assessed during an initial feasibility study (Keir et al. 2014). Direct metering was deemed most suitable for the study area as indirect metering options were complicated by the range of mostly unmetered energy sources used to power pumps in the study area (including solar, diesel, wind, and electricity) and the need to undertake additional field experiments to provide the basis for conversion from energy consumption to groundwater abstraction.

Groundwater bore pumping mechanisms and pipework vary greatly from bore to bore, and therefore the choice of flow meter is an important consideration. Noninvasiveness of flow metering technology was a primary concern for landholders participating in the program, and hence conventional in-line flow meters such as mechanical and electromagnetic meters (Sinclair Knight Merz 2012) were considered unsuitable. Ultrasonic time transit flow meters, which clamp to the outside of the bore discharge pipe, were selected, with a tipping bucket flow meter (Chow 1976) installed at one bore that was unsuitable for ultrasonic metering (as its discharge pipe did not flow full). Ultrasonic time transit flow meters are typically used for high accuracy contactless measurement of liquid and gas flow in industrial and resource applications but can also be used to measure bore discharge by measuring the velocity of the fluid in the discharge pipe (Sanderson and Yeung 2002). These meters have some desirable properties for groundwater extraction monitoring, including suitability for a range of pipe characteristics (diameter, material, wall thickness), semipermanent installation, ability to measure flow from both artesian and subartesian bores, high range of measurement, as well as the ability to record high resolution time series using digital data loggers. However, ultrasonic time transit flow meters have some shortcomings. They are generally more expensive than conventional in-line flow meters, and the reliability of the data collected is affected by the flow profile, which may be disturbed near pumps, elbows, or valves in the pipework (Sanderson and Yeung 2002). Data quality from these meters is also influenced by variations in transducer alignment, spacing and connection to the pipe (Sanderson and Yeung 2002). The configuration of the bore pipework, and the ability to protect transducers from interference, were therefore important considerations for bore selection and meter installation.

Meter installation

Groundwater extraction meters were installed on 43 bores between December 2015 and May 2016. Following a review of the metering network in 2017, 14 meters were moved to alternative bores in October 2017 and April 2018 and one meter was held as spare. All costs associated with the purchase, installation and ongoing maintenance and download of the meters were borne by the research funders with no cost to participants.

A typical unit as shown in Fig. 2 was comprised of: (1) a Flexim Fluxus F501/ADM 5107 flow meter (transmitter and data logger) with a pair of clamp-on sensors (transducers) and mounting saddles; (2) a power supply, including a 12V 55Ah deep cycle battery, 12V 50W solar panel and regulator; and (3) a mounting system with a lockable control box, and a base piece to hold the battery box and secure the unit against wind. All external equipment was specified to Ingress Protection ratings of IP67, with cable runs enclosed in plastic corrugated conduit to avoid bird and animal damage and bird spikes installed on solar panels to reduce perching and thus accumulation of faeces. For some installations, portable welded steel ‘cattle panels’ were used to enclose the control box and/or the clamp-on sensor location. Following installation of the meters, quality checks were undertaken by operating the bore pump and checking signal strength, sound speed, and that the meter was recording sensible, stable flow values. The meters were set to record average flow rates over intervals ranging from 1 min to 1 h on digital loggers.

Fig. 2
figure 2

a A typical clamp-on ultrasonic flow metering unit, b clamp-on sensors fastened to a 63-mm polyethylene (PE) pipe using mounting saddles, and c portable ‘cattle panels’ enclosing the flow metering unit

Full size image

A tipping bucket flow meter was used to monitor a bore that was windmill-driven at the time of installation (it is now solar-equipped) discharging into a small, raised dam. The meter was designed with a tip volume sufficient for accurate measurement of this bore with a typical flow rate of 0.2 L/s and an expected maximum flow rate of 1 L/s. This was one of the few monitoring sites with cellular reception and was equipped with a telemetry device to transmit the timestamp of all recorded tips every hour to a remote FTP server. The system was powered by a 12V 7Ah battery and 12V 50W solar panel. A simple mounting system which could be adjusted to allow level mounting of the bucket underneath the discharge pipe outlet was used. The installed system is shown in Fig. 3.

Fig. 3
figure 3

Tipping bucket flow meter measuring windmill-powered groundwater extraction (now solar-powered)

Full size image

Metering network and characteristics of sampled properties

The initial and current metering networks are summarised in Table 3. A total of 57 bores have been metered over the course of the metering program. Property-scale extraction data, where all groundwater extraction on the property is metered, are available for 20 properties.

Table 3 Metering network
Full size table

The current metering network comprises 42 meters on 29 beef cattle grazing properties spread throughout the Surat CMA. The 42 metered bores are all used for livestock watering and 17 are also used for domestic purposes (which may include watering a garden around the homestead). The bores access groundwater from subartesian (depth to groundwater ranging from 1 to 90 m below ground level) and artesian aquifers of the Surat CMA and are equipped with a variety of pump types (positive displacement and centrifugal) and power sources (mains electricity, diesel engines, solar photovoltaic panels, and windmills). Two of the bores are shared bores and distribute groundwater to more than one property.

The properties that currently have metered bores (excluding share bores) range from 130 to 10,600 ha (Fig. 4a). The sampled properties represent 98% of the range of cattle grazing property sizes in the Surat CMA (Fig. 4b) but overrepresent medium-scale properties (in the range 2,500–10,000 ha).

Fig. 4
figure 4

a Area of properties with metered stock and domestic bores (excluding share bores). b Area of properties used for cattle grazing in the Surat CMA. Properties less than 5 ha or with less than 10 head of cattle (based on cattle number data sourced from Keir et al. 2019) are considered periurban or ‘lifestyle’ (hobby farm) blocks and have not been included. The figure excludes 75 properties ranging in size between 15,000 and 52,700 ha.

Full size image

The currently metered properties include 21 breeding properties, 4 backgrounding properties (growing animals prior to entry to the feedlot), and 2 properties which house bulls, with some breeding properties also used for backgrounding when pasture condition allows. Head of cattle accessing bore water on metered properties ranged between 15 and 3,590 adult equivalent—AE, a standard measure of cattle grazing loads in northern Australia developed by McLean and Blakely (2014)—during the monitoring period. The stocking rate varied between 0.08 and 1.7 AE/ha and was 0.33 AE/ha on average.

Due to confidentiality provisions associated with bore owner participation in the study, the precise locations of most monitoring sites cannot be disclosed. Landholders from seven properties with 12 metered bores have consented to full disclosure of data to date.

Data download

An initial field visit was conducted approximately 6 months following installation of the meters to download data from the meters and conduct maintenance checks and repairs. Field visits were subsequently conducted every 4–18 months. Field visits were generally conducted in the drier months from April to October as tracks to many of the bores are inaccessible during wet conditions. Typical maintenance checks included (1) checking sensors were securely fastened on the bore discharge pipe, clean, in alignment and the correct distance apart; (2) checking power supply to the battery and that solar panels were clean and not obstructed by vegetation; and (3) checking the logger box was upright and intact, and wires between the logger box and sensors were not damaged.

During the field visits undertaken in April and May 2019, August 2020, and May 2021, interviews were conducted with bore owners or property managers. These interviews included semistructured and closed questions to collect qualitative and quantitative data on water use practices, following comparable studies on characterising irrigation practices in other countries (O’Keeffe et al. 2016). This data included details of bore operation and use, head of stock watered, and property area reticulated by the bore and will be used to inform analysis and interpretation of the metered use data collected.

To protect the identity of participants and data collected, all data gathered for the project were stored in a secure data management system, which can only be accessed by researchers identified on the ethics approval and subject to program confidentiality requirements. Figure 5 summarises the available groundwater extraction data for the 57 bores metered during the program. The reasons for gaps in the data are discussed below.

Fig. 5
figure 5

Duration of groundwater extraction metering showing gaps in the available data. Property IDs are denoted by numbers and bore IDs are denoted by letters

Full size image

Data processing

Raw data downloaded from the meters were reviewed to assess data quality. Where anomalous readings were observed, the field notes and interviews were reviewed to identify the likely cause. An algorithm was used to remove sections of data where the meter could not accurately record flow (e.g., due to misalignment or disconnection of sensors from the pipe), filter out anomalous readings (e.g., due to convective movement of water in the distribution pipe such as shown in Fig. 6), and statistically impute missing values for flow events which were realistic in duration and timing but had isolated portions with unreasonable flow values (e.g., due to cavitation at the beginning and end of pumping periods).

Fig. 6
figure 6

a Raw and b postprocessed data for a meter on a diesel bore used for stock and domestic purposes. Postprocessing has removed short duration flows, which are understood to be due to convective movements of water in the distribution pipe that are not associated with pumping

Full size image

There are gaps in the groundwater extraction data for 30 of the 57 bores (Fig. 5). Misalignment or disconnection of sensors was the most common reason (occurring 30 times during the monitoring period), and mostly occurred due to interference with sensors by cattle or other animals, storm damage, or during maintenance or improvement works on the bore and associated pipework. Metering equipment malfunction occurred 16 times and included logger failure, logger error and battery failure. The flow meter control box was knocked over three times during the monitoring period by animals or inclement weather.

Data were also lost due to errors made by the research team, particularly at the start of the program, including not restarting the logger after downloading data (4 bores), setting the logger to record flow too frequently, resulting in the logger running out of memory (19 bores), not downloading data from the telemetry system before it expired (1 bore) and logger download software incompatibility with a particular laptop used for field work (2 bores).

To create a dataset useful to support research on groundwater extraction in the case study area, the postprocessed subdaily metered extraction data for each bore were aggregated to an annual (water year, 1 July to 30 June) time step. Where groundwater extraction data were missing for more than one-third of days in a year, the data were excluded from analysis. Where groundwater extraction data were available for at least two-thirds of days, average daily groundwater extraction for that water year was used to estimate annual groundwater extraction. The uncertainty associated with infilling data gaps was assessed as described section D of the ESM.

Groundwater extraction data were collected at 57 bores over 205 annual (water year) periods during the metering program. After postprocessing to remove spurious data and infill data gaps, extraction data were available for 54 bores over 171 annual (water year) periods. This corresponds to a 16.6% loss of annual extraction data due to issues with operation of the meters.

Groundwater extraction data

Figure 7a shows the distribution of average annual groundwater extraction from metered stock and domestic bores during the monitoring period. Extraction from metered bores was 5.8 ML/year on average and ranged from 0.31 to 22.5 ML/year, indicating significant variation in the annual volume of groundwater extracted at different bores.

Fig. 7
figure 7

a Average annual groundwater extraction for metered stock and domestic bores in the Surat CMA during the monitoring period 2015 to 2021. b Average annual groundwater extraction from properties with metered stock and domestic bores in the Surat CMA during the monitoring period 2015 to 2021. Results are for properties where all groundwater extraction on the property was metered

Full size image

Figure 7b shows the distribution of the same extractions after they were aggregated to the property scale (where all groundwater extraction at the property was metered). Extraction from metered properties was 8.3 ML/year on average and ranged from 0.31 to 25.9 ML/year, indicating significant variation between properties.

Discussion

The voluntary metering network described in the previous sections has been operating since 2015 and has gathered a unique data set on stock and domestic extractions in Australia. The metering program was established with the intent of obtaining groundwater extraction data for a representative sample of water users in the Surat CMA to improve understanding of the factors driving groundwater extraction for stock and domestic use and support the development of lower cost estimation methods. However, considering it is one of the few reported, ongoing voluntary extraction metering programs internationally, the findings will potentially have value across a range of settings globally where metering is not required. The principal value of this paper is to discuss the challenges encountered in the metering program and how they were resolved. This follows the view (e.g., Andreassian et al. 2010) that giving prominence to analysing challenges is key to advancing hydrological science, with the intent in this case of assisting the development of comparable studies internationally. Ultimately, once other relevant experiences are published and synthesised, it may be possible to develop more comprehensive guidance. Given the pressures on groundwater resources often involving diffuse, unmetered extractions, such guidance seems likely to be of value. The rest of this section focuses on expanding on the challenges encountered, lessons learnt to resolve them, how these relate to others’ experiences, and their potential broader relevance.

Challenges recruiting participants

Voluntary groundwater extraction metering programs inherently require public participation; however, there are a number of sensitivities associated with metering privately owned bores, which presented challenges in the initial stages of participant recruitment and throughout program implementation. A concern raised during the participant recruitment phase was that data gathered would be used by government to place limits on extraction and charge for water use. A similar concern was raised by participants in the study by Fishman et al. (2016) and is a typical issue for metering programs (Molle and Closas 2021). Another concern raised was regarding privacy of data collected, as also noted in Waite (2014).

Considering these concerns, the success in recruiting participants can be attributed partly to the program being run by a university rather than a government agency, which agrees with the findings of Gillespie et al. (2016) that the Surat CMA communities have a higher level of trust in universities than in other institutions including government agencies and gas companies. The research team focused on building trust during the recruitment phase by being open and transparent about the purpose of the study, how data would be used and funding sources. Additionally, participants were offered default terms that participation would be confidential, where their identity, location and extraction data are kept confidential and cannot be disclosed to a third party outside the research team, and any changes to this required the explicit permission and agreement of the participant. Establishing formal terms of agreement between participants and the research organisation was a successful strategy in the Fishman et al. (2016) study. Our experience demonstrates the importance of listening to and understanding community concerns and tailoring research program design to address issues raised when implementing participatory monitoring programs.

A critical success factor to recruit and retain participants was engaging local champions (trusted community informal leaders) who could influence others to participate (or otherwise). This is recognised in the literature for successful participatory models of groundwater management (Beal et al. 2018; Fielding et al. 2013; Shalsi et al. 2019) and citizen science (Jamieson et al. 2020; Little et al. 2016). These approaches can assist in creating a trust-based network between hydrogeologists and groundwater users, therefore strengthening connection and knowledge transfer, facilitating science outreach and communication, and contributing to participatory groundwater monitoring (Re 2015).

Cold calling was necessary to improve spatial coverage of metering across the study area and ensure a representative subset of users were metered, although overall the recruitment method led to an overrepresentation of medium-scale properties and no representation of properties greater than 10,600 ha. It is speculated that cold calling arouses suspicions of a sales or regulatory-driven exercise, which can make recruiting via this method challenging. Where cold calling was used, it was made easier by following guidance in a detailed communications plan, which had input from an experienced professional with knowledge of the local area, and by conducting cold calling during the second half of the recruitment phase, which meant research team members were able to draw on their experience engaging with bore owners during earlier stages of the recruitment process. The offer of complimentary groundwater quality analysis and the opportunity to have access to detailed records of bore usage were attractive to many bore owners.

Challenges metering bores using ultrasonic meters

The reliability of readings from the ultrasonic flow meters used in our program is influenced by a range of factors including how and where transducers are installed, pipework dimensions and properties, and fluid properties (Sanderson and Yeung 2002). The quality of data collected by the meters also has the potential to be affected by interference with metering equipment by humans, animals, and environmental factors (Gomani et al. 2010). This presented a number of challenges for the installation and ongoing operation and maintenance of the metering network.

Challenges installing the metering network

Ultrasonic flow meters are heavily influenced by changes in flow profile caused by flow disturbances near pumps, elbows or valves in the pipework (Sanderson and Yeung 2002). To minimise the effects of these disturbances, the meter supplier recommended that transducers be located at a distance 10 times the pipe diameter downstream and upstream of features that cause flow disturbance wherever possible. It became apparent during meter installation that the pipework configurations of many of the bores meant this condition could not be met. Due to the remoteness of many properties, preliminary site inspections had not always been undertaken and the suitability of bore infrastructure had mostly been assessed through discussions with bore owners over the phone and photographs. While most bore owners were amenable to reconfiguring their bore distribution pipework to meet our requirements, on two occasions it was considered too difficult in which case the site was deselected. It also became apparent during meter installation that access to some bores was difficult due to the poor condition of access roads. This leads to the conclusion that assessing the suitability of bore infrastructure and access through discussions with landholders, photographs and publicly available maps is not always reliable. Where conducting a preliminary site inspection is considered too costly, it would be beneficial to confirm that the potential participant is open to the possibility of reconfiguring bore distribution infrastructure.

Another key consideration regarding ultrasonic flow meter reliability is that sediment or gas bubbles in the fluid can interfere with the transducer signal and result in measurement errors. Groundwater in parts of the Surat CMA has a natural gas content, which may affect meter readings, and gas may also be entrained in the pipe at the beginning and end of pumping periods due to cavitation. A key learning during our monitoring program was to position transducers on the side of distribution pipes to minimise anomalous readings as gas bubbles will occur at the top of the pipe.

Challenges in operating and maintaining the metering network

Ultrasonic flow meters are sensitive to variations in transducer alignment, spacing and connection to the pipe (Sanderson and Yeung 2002). To minimise potential changes to the initial installation geometry of the transducers, transducers were placed within cattle panels or farm infrastructure (including old windmill frames and bore enclosures) wherever possible. Nevertheless, movement of sensors due to animal or environmental effects was the most common reason for data completeness and quality issues; going forward the program will investigate the use of protective rails to secure transducers to pipes. Other lessons learnt during the program include the need for cable covers to protect wires from animal damage, fastening of cables to pipework (or other stable features) to reduce movement from trip hazards, regular cleaning of transducers to remove dust and increase signal strength, regular cleaning of solar panels to remove dust and bird faeces, and pruning of trees away from solar panels to increase power supply to the battery.

Owner-initiated changes to bore infrastructure occurred for a number of the metered bores during the program and included changes to the power supply, reconfiguration of pipework infrastructure, installation of new pipework infrastructure, installation of new pumps and lowering of pumps. Within the study area, there is a trend for bore power supply to be changed from windmill or diesel to solar, and while this generally will not affect the meter installation, the change in flow pattern is an important consideration in the postprocessing of data. Changes to pipework configuration or infrastructure will require transducers to be reattached, and if pipework dimensions and materials are changed, the logger will also need to be reprogrammed so that representative velocity measurements are recorded. Bore owners may attempt to reattach transducers on new or reconfigured pipework; however, without proper training, it can be difficult to achieve the requirements for transducer location, alignment, spacing and connection to the pipe (Sanderson and Yeung 2002) and results may therefore be compromised. This demonstrates the importance of regular field visits and clear communications with participants to understand when changes to bore infrastructure occurred. To reduce the risk of data loss, there may be benefits to expanding the scope of the voluntary metering program to reflect more of a citizen science approach where bore owners are trained to replace sensors on pipework, maintain meters and download data. Otherwise, near-real time data transmission through a more widespread telemetry network could assist in quickly flagging meter disturbances. Although challenging due to remoteness, this option is currently being considered. It would offer additional advantages such as reducing the costs of annual visits to download data and more rapid provision of data to landowners.

Another common issue encountered during the field visits was metering equipment malfunction. Equipment malfunction, which includes logger hardware failure, logger error and battery failure, has occurred during every field visit since 2017. This demonstrates the importance of maintaining the availability of spare hardware and ensuring a line of communication to equipment suppliers. The participation of the regional equipment supplier in the project has helped the efficiency of resolving failure and data errors.

Challenges interpreting data

One of the main challenges of groundwater extraction metering programs is interpreting the raw data collected. Decisions about postprocessing readings have the potential to introduce uncertainties into the metering results. An algorithm was developed for our program to provide a consistent approach to postprocessing data; however, the development of this algorithm proved challenging as the nature of anomalies varied between data sets and sources of anomalies were unclear in many cases. The algorithm needed to be tailored for individual bores in many instances.

Anomalous readings, at the beginning and end of pumping events, for example, were generally understood to be related to cavitation in the pipe; therefore, postprocessing consistently adjusted these readings to reflect the expected flow rate for the bore. However, other anomalies in the data could not always be explained and the approach to postprocessing needed to be based on information provided by bore owners during interviews or assumptions—for example, the postprocessing of data for a diesel bore, as shown in Fig. 5, involved removing low flows of short duration because the bore owner indicated the bore was operated at around 5.7 L/s and would not be pumped for less than 24 h. It was assumed the low flows in the data were due to the convective movement of water in the pipe not associated with pumping. The usefulness of the interview data for postprocessing demonstrates the importance of regular communications with bore owners regarding bore operation and changes to bore infrastructure.

It is recognised that there may be inaccuracies in the information provided by bore owners—for example, at one location the bore owner indicated the bore would not be pumped for less than 6 h; however, review of the raw data shows clear pumping sequences without anomalous values of shorter durations. Therefore, there is a benefit to reviewing raw and postprocessed data with bore owners during field visits.

Challenges using results

Results of the metering program (Fig. 7) indicate significant variation in the annual volume of groundwater extracted for stock and domestic use at the bore and property scale. This presents challenges for how the data can be upscaled to estimate extraction at a regional scale. The variation in groundwater extraction is potentially related to the variation in bore and property characteristics (discussed in section ‘Metering network and characteristics of sampled properties’) including bore purpose (stock vs stock and domestic), bore pressure (subartesian vs artesian), property size, type of cattle operation (breeding vs backgrounding), and head of cattle accessing bore water. Multiple other factors affecting variation in groundwater extraction and use have been proposed in the literature including pasture quality, climatic conditions, and the availability of other water sources (Keir et al. 2019; Kent et al. 2020; Parsons Brinckerhoff 2011; Queensland Water Commission 2012; Singh et al. 2020). This metering program provides the first significant data set on stock and domestic extractions in Australia, and possibly the world, to test these hypotheses and will be further explored in a subsequent paper.

Most of the participants opted to participate in our program on a confidential basis due to data privacy concerns. While this does not reduce the scientific value of the data for the research team who have access to all data collected, it limits the wider utility of the data both for research and for groundwater model applications, as the data can only be used in an aggregated or unidentified form. The Singh et al. (2020) study, which used remote sensing and other data to estimate groundwater extraction for stock and domestic use in the Surat CMA, used data from only seven of the metered bores to assess the reliability of their method. If more participants consented to full disclosure of their metered data, this could support more accurate regional water use estimates, and hence contribute to improving the groundwater impact assessment for the Surat CMA (OGIA 2019). Initiatives by the Queensland government to engage with groundwater users in the Surat region towards improved groundwater information systems (Jamieson et al. 2020) have been successful, and have created new opportunities for recruitment under modified confidentiality agreements. Engagement with program participants on proposed modified confidentiality agreements, which give wider access to the data collected (including select Queensland government departments), are now being investigated by the research team.

Future program directions

While at this stage regular field visits are planned, our review has identified the potential for the program to take more of a citizen science approach. Currently, the program involves participants providing access to private properties and infrastructure so researchers can collect data, and contributing local knowledge, which is typical for many participatory monitoring programs (Gomani et al. 2010; Jamieson et al. 2020). However, programs that fall into the category of citizen science involve participants collecting and reporting data following training (Gomani et al. 2010; Jadeja et al. 2018; Jamieson et al. 2020; Little et al. 2016; Regmi et al. 2019). Participants in our program could potentially be trained to check metering equipment, replace, or realign sensors on pipework, download data, or upload information through a dedicated app. However, the willingness of participants to undertake these activities would need to be assessed initially before a training program and materials were developed. Other approaches under consideration for the program include the use of telemetry to provide near real-time information on pump performance and alert the project team or bore owner to the need for maintenance checks. Indirect metering, using a device that records bore pumping hours, is also under consideration as a cost-effective way to expand the network as the existing meters could be used to develop conversion factors for different energy sources.

Conclusions

Voluntary metering was used to measure extraction rates from a total of 57 privately owned groundwater bores used predominantly for stock and domestic purposes in remote areas of the Surat Basin, Queensland, Australia. Participation in the metering program was voluntary and involvement was encouraged by offering bore owners the option of various levels of confidentiality, with the default condition being full confidentiality. Stock and domestic extractions are not typically metered, and this program will provide a nationally, most probably globally, unique dataset on this type of water use. In addition, the learnings from the program can be used to develop comparable voluntary metering programs to improve understanding of water extraction in unmetered systems across a range of hydrological environments and agricultural settings. By managing the challenges discussed in this paper, the research team was able to produce a dataset that can be used to provide insights into the factors driving groundwater extraction. Understanding spatial and temporal variation in groundwater extraction, and the factors influencing this variation, is critical for developing and validating lower-cost estimation methods and demonstrates the importance of metering for water resource management. Of the numerous outcomes, perhaps the most important is that there is an appetite among many rural landowners to participate in voluntary groundwater extraction metering programs, which can be successfully implemented through careful, well-informed planning to tailor program design to participant needs.