Making the invisible visible: the impact of federating groundwater data in Victoria, Australia

Groundwater is a precious resource, however because it is hidden from view, the nature of groundwater can be misunderstood by non-scientists and is often the subject of myths (Price 1996). Globally, the expanding demand for groundwater to supply human consumption, energy and food production has led to groundwater resource overexploitation (Gorelick & Zheng 2015) with corresponding threats to environmental and ecological values (e.g. Nevill et al. 2010) and the sustainability of food production (e.g. Scanlon et al. 2012). As a result, groundwater exploitation in many countries is regulated by statutory requirements that increasingly consider the competing economic, social and environmental needs (e.g. Holman & Trawick 2011; AWA 2012; Fernandez et al. 2014; Gill et al. 2014).

Referred to as the New Digital Age (Schmidt & Cohen 2013), or era of Big Data (e.g. Boyd & Crawford 2012; Mayer-Schonberger & Cukier 2013) the present time period provides unprecedented opportunities for a deeper understanding and appreciation of our global environments, including hydrogeological environments. The volume of digital data on natural environments has grown exponentially, especially in the physical (e.g. Lynch 2008; Bell et al. 2009) and environmental sciences (e.g. Porter et al. 2012) where much of it is collected by sensors. The use of volunteered geographic information and citizen science is also rapidly expanding the volume of water and environmental data (e.g. Fienen & Lowry 2012; Werts et al. 2012; Sui et al. 2013; Little et al. 2015). In addition, data availability has vastly improved as governments in many countries adopt open data policies (Zuiderwijk & Janssen 2014). Yet paradoxically, despite this unprecedented access to data, limitations remain on how to use these data to best develop water management policy and further the public understanding of groundwater science (e.g. Loch et al. 2014).

Part of the problem is the sheer range of information sources and volume of data that is available. In Australia for example, information and data on groundwater are distributed via dozens of web-portals, web-based geographic information system (GIS) tools, password protected portals, cloud storage, portable storage devices; hardcopy maps, theses, reports, newsletters, documents, videos and podcasts. Outside of the research community, this impressive resource of data, information and knowledge is largely ignored simply because most people do not have the knowledge, capability or desire to deal with the data deluge. Many people feel increasingly time-poor and even though there is a plethora of data available, there is little opportunity or desire to undertake the research required to bring available information together in ways that best answer the questions that will guide future planning for sustainable and equitable groundwater use.

To partially address these issues spatial data infrastructure (SDI) has been developed and deployed to federate groundwater data from disparate database sources into a single web portal thereby making data more easily discoverable. Globally, the Canadian Groundwater Information Network (GIN) was the initial exemplar that was developed using open geospatial standards and technologies (Boisvert & Brodaric 2012). Other examples include the European Commission's INSPIRE network (Uslander 2005), the United States National Groundwater Monitoring Network Data Portal (NGWMN) (ACWI 2013), the New Zealand SMART system (Klug & Kmoch 2014) and the Australian National Groundwater Information System (NGIS) (Iwanaga et al. 2013; BOM 2015). In all cases these portals are managed by the government agencies with the statutory responsibility for groundwater management.

By contrast, the Visualising Victoria's Groundwater (VVG) portal (www.vvg.org.au) was developed outside of the government by Federation University Australia (FedUni), to federate all known groundwater data for the State of Victoria, Australia. The initial purpose of the VVG portal was to assemble datasets for university research as well as make legacy data, community-sourced groundwater information and government datasets visible to the public. The portal was launched on July 12, 2012, by the Centre for eResearch and Digital Innovation (CeRDI) at FedUni in collaboration with international and national research agencies, state government departments, regional water authorities and industry partners. It is innovative because it offers real-time access to information and data that are normally invisible to most of the community. The system seamlessly integrates data and information using international data exchange standards, federating all or parts of groundwater databases with disparate schemas and stored on disparate systems, subject to the custodians' consent. Tools for data querying and 3D visualisations were developed to assist decision making and community engagement.

Arguably the most novel aspect of the VVG research project has been in evaluating the impacts of the web portal over its initial two years of operation, through a multidisciplinary collaboration between hydrogeologists, information technologists and social scientists. This team used a combination of tools including survey instruments, expert reference groups, and internet analytics to explore the following research questions:

1. How has the VVG project and web portal impacted at the industry and community level since the program commenced?

2. To what extent has the web portal supported decision making at the industry and community level?

3. In what way has the provision of current groundwater data been improved since the establishment of the VVG portal?

4. To what extent has the VVG portal been used to assist groundwater and catchment managers?

5. To what extent have there been increased productivity gains for industry and users of the web portal?

These questions are designed to test the value of the investment required to federate groundwater data from authoritative and trusted sources and then build the tools that allow groundwater information to be visualised. The value is assessed by the end users' adoption of the project and the practice change it creates.

Established as a Colony in 1851, the State of Victoria occupies 227,416 km² of the southeast Australian mainland (roughly comparable in size to Laos, Romania or the UK). Records of drilling by the Victorian Government commenced in 1884 and were published in a series of annual reports, generally referred to as the Boring Records, until 1965 (e.g. Langtree 1885; GSV 1965). The first comprehensive groundwater database was assembled by the Geological Survey of Victoria (GSV) in the late 1960s with the introduction of the Groundwater Act 1969, the first groundwater legislation for the State. From the mid-1980s onwards the hardcopy records were progressively transferred to a digital database, and included private wells licensed as groundwater bores, as well as groundwater investigation or observation bores drilled by other government agencies such as the State Rivers and Water Supply Commission (SRWSC) and the Soil Conservation Authority (SCA) and subsequent equivalents (although these agencies also kept their own bore databases).

Machinery of Government changes in mid-1988 saw the State bore database duplicated as statutory functions were divided between various departments. One copy was merged with several rural water authority databases to become the Victorian Groundwater Data Base (VGDB), which subsequently became the Groundwater Management System (GMS) and finally the Water Measurement Information System (WMIS) currently under the custodianship of the Department of Environment, Land, Water and Planning (DELWP). The other copy remained with the GSV and was developed into the Geological Exploration and Development Information System (GEDIS), which included the mineral, stone and hydrocarbon exploration bores, currently under the management of the Department of Economic Development, Jobs, Transport and Resources (DEDJTR). Although data exchange was attempted for a few years following the split, the databases ultimately grew into quite separate entities.

In the late 1970s the SCA developed a separate bore database for monitoring groundwater levels in observation bores that were constructed for salinity investigations, now under the custodianship of DEDJTR. In addition, several other bore databases that were developed by former public utility agencies (e.g. the State Electricity Commission, SRWSC, Country Roads Board, Victorian Railways, Ports and Harbours, etc.) have now been privatised. Although much of the historic groundwater data have been captured on the WMIS, a vast amount of hydrogeological, geotechnical and lithological information has been archived.

The current situation is that groundwater data in Victoria are divided across several government departments, water agencies, research organisations, public archives and private industries.

In collaboration with the project partners and stakeholders, the VVG portal was designed to include the following features:

• user requests will be fulfilled via real-time access to remote databases by integrating the interoperable web services they each provide;

• the data resides with the data managers (ensuring currency and validity);

• it has a spatial map function that is intuitive to use (similar to Google Maps);

• all forms of data are included – vector, raster, text and multimedia;

• data downloads are allowed (subject to data custodian's consent);

• spatial data entities link to the original source documents and images;

• it is capable of dynamically synthesising the data;

• interactive 3D visualisations can be created for user-selected scenes;

• users can add, edit or update data (subject to quality assurance and quality control);

• the spatial data and models are credible to the user.

These features of the VVG are also reflected in allied spatial information systems built by CeRDI for broader research applications (e.g. Dahlhaus et al. 2011, 2012; Milne et al. 2014; Thompson et al. 2014), which share key principles including:

• use open-source and standards-compliant software wherever possible;

• build upon existing collaborative software initiatives and contribute enhancements/tools back to the research community;

• ensure the flexibility of the developed system to consume data from a variety of sources so as not to interfere with existing provider work practices;

• ensure end-user tools and applications are fast, intuitive and easy-to-use;

• software is cloud-based so there is no end-user requirement for software, updates, computation power or plug-ins.

The SDI for the VVG project builds upon software projects fostered and supported by the Open Source Geospatial Foundation (www.osgeo.org). Delivery is primarily via a web-browser, the portal interface having been built to bespoke requirements upon the foundations of the OpenLayers (openlayers.org) javascript library. Other Javascript libraries like jQuery, jQueryUI and DHTMLX Tree have been leveraged to provide additional user-interface components and functionality.

While most data are consumed via interoperable services, there are a number of datasets hosted and delivered by VVG. Spatial data engines Mapserver (www.mapserver.org) and Geoserver (geoserver.org) are used for the Geospatial processing and service delivery using Open Geospatial Consortium (www.opengeospatial.org) standards. Vector data are commonly stored within a MySQL or PostGIS database and raster data are dynamically processed from its native format. To deliver complex web feature services (WFS) such as GroundwaterML (Boisvert & Brodaric 2007, 2008, 2012), the Geoserver app-schema extension has been deployed. Geonetwork (geonetwork-opensource.org) is used as the public-facing metadata catalogue for the portal.

At present the VVG portal seamlessly federates 79 datasets on Victorian groundwater from six disparate custodians. These are as follows:

• Bore data from: (1) WMIS, containing groundwater bore data for wells drilled under the statutory licences, managed by DELWP; (2) GEDIS, containing mineral, stone and hydrocarbon exploration bores, managed by DEDJTR; (3) the salinity observation bore database managed by DEDJTR; and (4) the research bore database managed by FedUni. The latter also contains data on existing bores that are not recorded elsewhere (termed ‘orphaned bores’). In total, data on over 400,000 bores have been federated.

• Groundwater spring data from the Victorian Mineral Springs Database managed by an individual researcher (Dr Andrew Shugg).

• Sites where groundwater contamination may have occurred and have either been issued with a Certificate and/or Statement of Environmental Audit; been declared a Groundwater Quality Restricted Use Zone; or been listed on the Priority Sites Register for environmental clean-up or pollution abatement notice. These sites are managed by EPA Victoria, the State environment protection authority.

• Groundwater surfaces interpolated for the Victorian Aquifer Framework (VAF) including the predicted depth to water table, predicted groundwater salinity, elevation of the natural surface, elevation of the geological basement, and the structure surfaces of the 17 aquifers and confining beds that make up the state-wide groundwater systems framework (SKM 2012). Additional modelling was undertaken in the VVG project to derive the depth to, thickness, and bottom elevation from these structure surfaces, resulting in 68 surfaces in total.

• Seamless geology map supplied as a web service from the GSV (DEDJTR).

• Boring records in digital form from the State Library of Victoria. Where the information can be matched, the boring record is linked to the data for that individual bore shown on the map portal.

In some bore databases, such as the FedUni groundwater research database and the WMIS, additional materials such as images, documents and sketch maps may be linked to the bore data. The intention is to always provide the source documents where possible.

The website includes background information about the project, the project partners, the data sources, historical context for the data, an extensive user guide, answers to frequently asked questions, news and newsletters, a documentary video about the project, a research blog and contact details. The web portal includes a Plain English disclaimer to alert the user to the facts that the data and information may not be accurate, current or complete; is subject to change without notice; is continually being validated, enhanced and updated; and is subject to the usual uncertainties of scientific research. Hence data quality and data provenance (metadata) are issues for the data custodians. It is their data and they set the rules of service.

Tools in the map portal include the ability to select between different base layers (supplied by Google), re-order the data layers, adjust the transparency of the layers, vary the query radius for bore data, export data as Excel spreadsheets, search for a street address and search for a bore identifier. Drawing tools are also included to allow users to mark on the map and send a comment via email. This function has been used by researchers and the general public to indicate where data may be incorrect, such as the location of a bore, and thus allow the data custodians to improve their data veracity.

Four query modes are available: a bore query that finds all the bores within a user-selected radius; an EPA data query that is used to discover the information behind a contaminated site; a query which returns the data related to a polygon on the geological map; and the ability to query the predicted depth to water table, water quality and hydrostratigraphy at any selected point. This last query effectively provides a virtual borehole log at any selected location in Victoria, based on the layers provided in the VAF.

For the impact assessment, a mixture of qualitative and quantitative methods was used, viz:

1. Surveys: An on-line benchmark survey of the stakeholders, collaborators and participants in the VVG project was undertaken at the commencement of the project in July to December 2012. Two years after the implementation of the site, two further surveys were undertaken: (1) a 1-minute ‘snapshot survey’ as an on-line pop up invitation available to those accessing the portal and (2) a 10-minute on-line survey, by invitation to key end users. Data across these three surveys was reviewed for compatibility, analysed and cross referenced, as appropriate.

2. Individual interviews: Structured interviews were held with targeted individuals to gain subjective insights into user perceptions of service delivery, levels of satisfaction and issues of concern. This provided data that supplement and validate the other data collected for the research.

3. Document analysis: Program documentation (internal) was thematically analysed to identify the synergies between the aims and the implementation of the VVG. In addition external documents, particularly audit reports posted on-line by EPA Victoria, were analysed to gain significant insights into the role of the VVG portal as a decision making tool.

4. E-mail feedback: Feedback was collected on an ongoing basis from the inception of the VVG portal. This provided longitudinal data for the study and enabled feedback on a variety of issues (e.g. technical enquiries, service improvement, service gaps, data queries, etc.) to be gathered across the life of the project. Both email requests and responses to requests were analysed as part of this data collection method.

5. Website analytics: Time-series statistics on portal usage were reviewed through the mechanism of Google Analytics. Data validity was ensured through a filtering process that verified only valid portal visits were included in the analysis.

Ethical approval appropriate for research involving individuals was gained from the Human Research Ethics Committee of FedUni and principles guiding the data collection and analysis process were observed to serve the best interests of all participants at each stage of the research process (Ethics Approval Number A14-015).

Data collection for the study was contemporaneous with the VVG portal implementation as there was no capacity to undertake pre-testing due to the funded research program timelines. The benchmark survey was intended to provide the initial gauge for impact measurement, although much of the interview and feedback data also provided insights into working with groundwater data prior to availability of the VVG. Consequently, while the absence of a structured pre-testing process (resulting in no comparative data from before the VVG portal implementation) and the absence of a control/comparison group slightly reduces the rigour, the scope of data gathered in situ provides for high levels of data validity.

In particular, the mix of methods allowed cross-referencing of findings and established a process whereby issues that were not addressed or identified through one data collection method could be picked up through alternative methods. Importantly, the use of a multi method approach allows for triangulation of the methods (interview, survey, feedback, analytics and document analysis), type (qualitative cross sectional and statistical across an extended timeframe), and data source (e.g. government employees, members of community and industry and other researchers). The resulting approach overcame many of the issues often raised in relation to data validity of qualitative case studies in research and conform to recommended strategies in data collection for effective case study research (Yin 2014).

The statistics for the impact assessment methods are listed in Table 1.

The approach used in the data analysis was to apply a hierarchy of impact to each pool of data. The impact hierarchy has three levels, loosely based on the concepts developed by Kirkpatrick & Kirkpatrick (2005, 2006, 2007) that relate to levels of learning. Their work measured learning in terms of Reaction (initial reaction to the training), Learning (a measurable increase in knowledge about an issue), Behaviour (a change in behaviour that reflects an application of what was learnt) and Results (changes in outcomes because of the learning). The hierarchy applied to the data in this study modifies this conceptualisation to measure impact in terms of the following:

• Level 1: The Primary Impact based on website usage statistics and feedback on the VVG portal, to measure its initial impact and value as a data resource. This level provides baseline data into the extent to which there has been an impact on practice within industry and community.

• Level 2: The Practice Impact based on the extent to which the VVG portal is being utilised within the workplace and/or community to aid in decision making. It also considers whether the end users have modified their work and/decision making practices, made productivity savings and integrate the portal into their individual practice decisions in industry and in community.

• Level 3: The Sector Impact based on the extent to which the VVG portal is becoming embedded in the activities undertaken and decisions made by the groundwater sector (i.e. regulators, practitioners, researchers and the community). At this level the impact of the VVG portal is considered in terms of its integration as a tool that the sector views as part of planning and decision making on groundwater use, environmental planning and research innovation.

All data were assigned a level (1–3) dependent on impact as assessed on the basis of findings across the various data collection methods.

Triangulation of survey results with the interview feedback and written qualitative data found a high level of correlation in regard to these same variables. The following statements provide a representative sample of end user views on issues of ease of use, information availability and trustworthiness of the data:

Where the user understood that the VVG dynamically accessed updated information from source data bases, the rating was high, e.g. Interoperability standards … you're really leading the way in that kind of thing [end user – government]. However where the notion that remote databases were accessed in real-time was not readily understood, there were greater concerns around data reliability, e.g. The difficulty with the VVG is … because [it doesn't have] direct access to the information, it's knowing whether it's got the latest readings [end user – sector unknown]. While this type of feedback was limited to less than 1% of the qualitative data collected from the participant pool, it nevertheless indicates a need for ongoing end user education on how dynamic access to remote databases ensures that the data are always current.

The impact data analysis also highlighted that some end users remain confused around data ownership and the integrity of the datasets available through the VVG portal. There were 8% of respondents who assessed that the quality of data not useful and 14% who assessed that the data could not be trusted. Qualitative insights highlighted that some end-users misunderstood that data accuracy remains the responsibility of the data custodians rather than the VVG web portal managers, for example:

However, the data show that the vast majority of end users understand that the VVG portal provided data that were as accurate and as current as possible and was drawn from existing and known data sources. The findings align with the published literature that identifies a positive association between ease of access and increased frequency of use (e.g. Shanahan 2009) and the importance of high quality data as a fundamental component of innovation in technology (Haug et al. 2011; Anstiss & Marjanovic 2012; Li et al. 2012; Horn et al. 2013).

One of the most definitive trends to emerge from the written and verbal qualitative data for this study was the consistency with which participants identified that the VVG portal had increased user capacity to manage the knowledge themselves. The single point of data access was consistently identified as a significant attraction:

The Primary Impact identifies clear alignments between the VVG study findings on access, quality and usage, with those of previous research (Table 2), particularly in terms of:

• building community capacity through facilitating access to data that was previously not directly available;

• enhancing the capacity for decision making by providing a single point of access for facilitating sharing and ownership of information; and

• ease of information access for increased participation and education.

It should be noted that the data provided in Figure 7 is cross sectional (taking and analysing data at a single point in time) rather than longitudinal (data involving the same individual over a set timeframe), hence it provides indicative rather than definitive evidence of change. Nevertheless, these data show an emerging shift in the type of input and information being sought by end users of the web portal. This can be further explored by the qualitative feedback, such as:

The steady increase in the number of requests made via VVG portal feedback for the raw data of the publicly available data indicate that they remain invisible or difficult to access for a number of end users. As a result, the VVG portal is seen as a de facto information provider and in some cases, has responded by providing web services to make the government information provision easier. The scope of these requests, and the regularity with which they are received, highlight that:

• since establishment, the VVG has become a central resource in terms of knowledge management and the provision of trusted advice on groundwater issues;

• ease of access and an open access policy plays an important role in building end user confidence in seeking information relevant to their needs;

• the VVG, through the provision of support, advice, and web services is addressing a service need not currently addressed in more traditional areas of groundwater information provision.

For the sub-set of 49 interview and survey participants, 40 (82%) assessed the depth of data and 41 (84%) assessed that the volume of information available through the VVG was important for informing and supporting decision making. This was confirmed by the qualitative data, such as:

The value of the various datasets also varies by industry. For example, although the EPA Victoria sites (contaminated sites) rated lower in the on-line survey (Figure 10), an analysis of the contaminated site audit reports for 2012–2015 show that these data are of great value to the environmental consultancy industry. Since the launch of the VVG portal, a total of 321 audit and consulting reports have been added to the EPA Victoria file management system. Of these, 51 (16%) have included data that were identified as sourced from the VVG web portal, providing evidence of the influence of the VVG portal in that industry.

While not formalised through economic modelling, the analysis does provide a consistent view that the VVG portal reduces time spent in the preparation and sourcing of data, and in responding to requests for information. Again this was reinforced by the qualitative data:

Finally, when looking at the shift from Level 2 (Practice Impact) to Level 3 (Sector Impact), two impact measures were explored: the first is the opportunity for the VVG to meet diverse information needs and the second is the ability of the VVG to improve data accuracy and research potential.

While data were collected on the use of individual datasets (Figure 10), there was no measure within this survey data of the value end users found in the ability to draw on multiple sets of data at a single point of time. For this benefit, insights were drawn from analysis of the qualitative data which found that there were 172 instances (from the total participant pool of 278) in which feedback was provided specific to the value of the capacity to browse multiple datasets when making an inquiry about or investigating a groundwater issue.

In addition, 53 participants had specifically referred to the value of the multiple datasets in supporting accurate and timely responses, e.g.

A total of 46 participants made reference to the value of the spatial features of the VVG, for example:

Before the implementation of the VVG portal, groundwater data were only brought together by individuals on an as needs basis, so there was less potential to systematically identify data shortfalls or duplications. However, being able to bring together a range of datasets that otherwise are not normally viewed together has provided end users with the ability to rapidly see data duplication, data shortfalls, data gaps and/or data inaccuracies in existing groundwater information. This ability was assessed as an important development by the groundwater data custodians, the industry and the community as it facilitates improved groundwater data management and enhanced data accuracy for the sector.

Participant feedback also provided a reinforcement of the fact that the VVG portal is a means to enhance data, as recognised by those who work closely with the datasets available through the web portal. This awareness is captured in statements such as:

The analysis of the impact of implementing the VVG portal found that its growing use is facilitated by the ability to federate a diverse range of data, its ease of accessibility and the quality of the data. The frequency of use and repeat visitation rates testify to its value as a viable and sustainable mechanism for information access. These characteristics of the web portal are instrumental in breaking down knowledge silos and are shifting the paradigm from one of knowledge controlled by government and statutory authorities to one of knowledge ownership and control by end users.

Multiple data sets and functionality web portals such as the VVG can enhance capacity across the industry and broader community and in terms of the provision of:

• timely, informed and accurate responses to those seeking information/answers to queries;

• improved mechanisms for monitoring – both in relation to issues of compliance and in terms of maximising the potential for good outcomes and positive developments across a field of study; and,

• increased potential for economic savings as a result of productivity through time efficiencies.

The provision of multiple datasets from disparate sources within a single portal through the facility of interoperability, establishes a unique opportunity to collate, cross reference and consolidate data that has historically been hidden. This shift establishes a new foundation in accessing research-ready datasets and a new capacity for achieving research discoveries.

The VVG portal is changing practice in the Victorian groundwater sector. Positive experiences for users in accessing such a web portal is ultimately increasing end user interaction and participation in the process of collaborative data improvement, enhancing knowledge within the sector and empowering society with the value of Big Data. As the provision of diverse and complex information through the addition of multiple datasets grows, the relevance and applicability of these data provides end users with a resource to guide future planning for sustainable and equitable groundwater use.

The authors thank the VVG portal collaborators, stakeholders, participants and end-users for their contribution to this research. The VVG project was funded through a Victorian Government Broadband Enabled Innovation Program grant (2011–2013). Thanks to Kelsey McDonald and Jennifer Corbett who assisted with the data gathering and analysis. The authors acknowledge the improvements to the manuscript suggested by the four reviewers.