Final Report Summary - COBWEB (Citizen Observatory Web)
Executive Summary:
COBWEB was a research project that succeeded in its goal of developing a generic crowdsourcing infrastructure platform and toolkit that could be used in multiple scenarios generating data of sufficient quality to be used by policymakers. At the end of the project, different components comprising the platform are at different Technology Readiness Levels (TRL); with some high TRL outputs being open sourced, eg, Fieldtrip Open.
Achieving an understanding of how citizen science can be harnessed in the service of societal goals is important. There is a veritable deluge of data being generated by citizens and some of that data is useful (or can be made useful) for environmental decision making. Perhaps more importantly, the technology is creating opportunities for greater citizen involvement in environmental decision making. For example, it has been observed that realising the UN’s Sustainable Development Goals will require massive citizen support. COBWEB made a contribution, most significantly in the following areas: standardisation, quality control, security/privacy and use of the UN’s World Network of Biosphere Reserves (WNBR).
A new Open Geospatial Consortium (OGC) Citizen Science Working Group has been established and looks set to become a key forum for advancing related global interoperability issues. COBWEB’s Sensor Web Enablement for Citizen Science (SWE4CS) initiative is a first step in the standardisation process and led the way in developing a harmonised information model for exchanging CS data and addressing semantic interoperability issues. The high TRL access management federation approach advocated by COBWEB for managing privacy and security has had impact on GEOSS and is being used by followon projects. Quality control continues to be a major concern for initiatives dealing with crowdsourced data; COBWEB pioneered a web services chaining approach that allows great flexibility in the processes applied in response to the multitude of different circumstances under which such data is collected. This too is being further developed by followon projects.
Using a unique co-design approach, COBWEB concentrated on mobilising citizens within Biosphere Reserves. We found the WNBR to be an excellent (and underused) resource for this kind of research project; where access to an already incentivised and organised pool of citizens enthusiastic about the sustainable development agenda is a definite advantage.
Having a government involved as a full partner in the project also proved to be an advantage; COBWEB benefited from direct communication channels with bodies responsible for setting and delivering policy relating to environmental data. This impact continues post project as Welsh Government is sustaining the COBWEB platform for a period in support of environmental decision making and policy.
Project Context and Objectives:
Project context:
Key context for COBWEB is the huge upsurge in the availability of mobile devices of increasing sophistication with concomitant increase in the volume of citizen sourced data. The big challenges were, and remain, how to make these data useful for better decision making and harnessing the citizen enthusiasm and technology in the service of societal goals. In some areas, COBWEB made some significant advances which are being leveraged post-project by various initiatives, including Citizen Observatory related followon projects funded under H2020.
COBWEB was predicated upon UNESCO’s World Network of Biosphere Reserves (WNBR) and the original proposal referenced the UN’s Millennium Development Goals. Over the period of the project, the Sustainable Development Goals (SDG) were published, including a clear statement referencing the importance of geographic information. The UN’s Man and Biosphere programme responded by aligning with the SDG’s as did multiple other public authorities at all levels.
The chief expected impact from COBWEB may be summarised as enabling greater citizen influence in environmental governance. The project set out to use a small number of Biospheres in Wales, Greece and Germany as representative testbeds for developing and validating the technology developed. In the process, we developed a unique co-design approach to respond to community priorities, mobilise citizens and help gauge the effectiveness of the WNBR concept. This aligned with a general shift towards co-design/co-creation and encouragement of greater citizen involvement in environmental decision making. Welsh Government was a full partner in the project and contributed substantially towards understanding the shifting context in respect of environmental policy development and delivery. They continue to support the deployment of the COBWEB platform in mobilising citizen’s associations post-project.
Standards were an important organising principle for COBWEB from the start and formed much of the context for the work undertaken. Over the period of the project, semantic web standards underpinning the open data movement continued to gain traction; representing a challenge in terms of compatibility with more traditional open geospatial interoperability standards. The toolkit COBWEB created and open sourced (Fieldtrip Open) can publish data in various formats depending upon what is required, eg, INSPIRE specific schema, linked data, GEOJson, or Sensor Web Enablement for Citizen Science. SWE4CS ‘Standardized Information Models to Optimize Exchange, Reusability and Comparability of Citizen Science Data’ is a first step on the standardisation ladder and was originally conceived as a standard for publishing citizen science data into GEOSS.
Linking with authoritative data from initiatives such as INSPIRE and GEOSS is part of a broader movement within the citizen sourced data community towards leveraging the wealth of related data from different sources for different purposes. COBWEB’s main contributions were in conflation/fusion with Spatial Data Infrastructure (SDI) initiatives like INSPIRE for display or analysis purposes, conflation with data from sensors, and Quality Assurance (QA). Note there is overlap between the latter, for example, SDI data can be used for QA, social media type data, eg, Twitter, Facebook, can be modelled as sensors.
An overarching problem in Citizen Science is that large quantities of data are being created but exist in silos. Useable standards either don’t exist, are neglected, poorly understood or tooling is unavailable. It is still early days and this is reflected in the different Technology Readiness Levels (TRLs) of outputs from research projects such as COBWEB. SWE4CS for example, is at an intermediate TRL. Similarly, the standards based web services chaining approach to QA pioneered by COBWEB allows great flexibility in the data brought to bear and processes applied in response to the huge variety of different Use Cases encountered in the Citizen Science space.
The access management federation approach advocated by COBWEB for managing privacy and security is high TRL and is likely to have continuing impact. This is because of the continued lack of a widely accepted, robust, practical way of securely sharing valuable data across administrative domains. The requirement to be able to securely share (with a high level of assurance) in a genuinely interoperable way controlled data such as the location of endangered species, and the associated need to maintain individual privacy, is a constant in the volunteered geographic information, crowdsourcing, citizen science context within which COBWEB was executed.
Project objectives:
COBWEB set out to research and develop a “generic crowdsourcing infrastructure platform” manifested as a toolkit which could be downloaded and used in multiple scenarios. The citizen observatory platform should be capable of integrating a wide variety of different information sources such as physical sensors embedded in the environment, cyber sensors leveraging online data, or social sensors leveraging social media streams.
One of our objectives was to explore the potential for integrating the platform into Biosphere information systems and evaluating the use of the WNBR as an integral part of projects like COBWEB, which require active cooperation of citizens and citizen groups.
Addressing the question: how can the business of government be improved using citizen science data, empowering citizens, enabling contributions towards better environmental governance in keeping with policy aims were overarching objectives for all the citizen observatory projects funded by the EU.
Meeting real world needs and addressing community priorities in a flexible way was a feature of COBWEB. To give the project focus, COBWEB committed to, and concentrated on, delivering outputs in three pilot case study areas:
1. Creation and validation of data products from Earth Observation data
2. Biological monitoring
3. Flooding
Policy objectives were addressed at a local, national, European or global level as appropriate. Helping realise the benefits of existing investment in SDI initiatives such as INSPIRE and GEOSS, and improving interoperability through adherence to open standards and by using the processes of the standards defining organisations was an objective of COBWEB throughout. This informed the design and development of the platform, especially in respect of the science and technology approaches used to developing solutions for security, privacy, quality assurance and conflation.
Project Results:
The following description of the main Science and Technology results from the COBWEB project draws extensively from the scientific paper ‘Citizen OBservatory WEB (COBWEB): A Generic Infrastructure Platform to Facilitate the Collection of Citizen Science data for Environmental Monitoring’ (http://ijsdir.jrc.ec.europa.eu/index.php/ijsdir/article/view/406/393).
The COBWEB project provides open source tooling which can be downloaded , configured and modified if necessary to enable citizen observatory’s – infrastructure which enables citizens to use their own mobile devices to make observations across a wide range of different citizen science scenarios. The platform developed and described below is generic, extensible and powerful enough to accommodate sophisticated requirements arising from a broad range of stakeholders across different sectors, yet sufficiently flexible to enable non-experts to create and use mobile applications which meet their needs.
It should be noted that COBWEB was a research project. As such, the various components described below are at different levels of technical maturity ranging from Technology Readiness Level (TRL) 4 (validated in laboratory) to TRL 9 (actual system proven in operational environment). The TRL scale is used as a tool for decision making on research and development investments at the EU level and has been developed to enable assessment and comparison of technologies in respect of maturity.
COBWEB provides high TRL tooling for collecting new data and, where possible, the software developed leverages and complements existing well established high TRL open source projects such as those in the OSGeo suite. COBWEB has also conducted research (low TRL) into the use of other crowdsourced data, for example, from social media streams and sensors, for both quality assuring and enriching observations. The latter is an example of the benefits of leveraging SDI as standards are critical for relating and combining (the terms fusion and conflation are also used) spatial data from various sources.
System design through stakeholder engagement and co-design:
COBWEB commenced with a process of requirements gathering for a system that would enable citizens in the Dyfi Biosphere Reserve area in Wales to collect data in the 3 pilot case study areas as introduced above. Requirements were initially gathered through a process of desk study reviewing existing systems and structured interviews. The latter started with citizen groups located within the Biosphere Reserve and were widened and further refined through broader stakeholder engagement; in particular with groups associated with environmental governance and stewardship.
In parallel, a process of rapid prototyping software development was entered into with the intention of creating software that demonstrated the feasibility of meeting identified requirements and surfacing issues and challenges. Once we had reached the requisite level of software maturity, to further assess the viability of the proposed solution and better understand the needs of the citizen, COBWEB then engaged in a period of structured co-design activity.
Co-Design:
Co-design happens where “participants are invited to cooperate with designers, researchers and developers during an innovation process”. Seven citizen groups active in the Dyfi Biosphere Reserve in mid-Wales were commissioned through an open tendering process to help gather requirements and validate our concept of a citizen observatory by mobilising citizens (usually volunteers) to go out in the field to collect data.
The invitation to tender was issued via the Dyfi Biosphere Reserve and COBWEB websites and resulted in a strong response from the community within the Biosphere Reserve expressing a desire to work with an unexpectedly (from the project consortiums perspective) rich set of environmental phenomena. This generated a diverse set of requirements and may be interpreted as evidence of the Biosphere Reserve mechanisms successfully working to help meet the sustainable development agenda in the context of a research project which needs to mobilise citizens.
Each citizen group’s proposal was analysed and refined in respect of what it could offer the project, e.g. research potential, policy implications, technical challenges, ethical implications, etc. The result was a challenging set of requirements to guide system design, based on real user needs, and the opportunity to predicate further research on real world requirements.
The seven co-design sub-projects were programme managed by COBWEB partner Ecodyfi (the delivery arm of the Dyfi Biosphere Reserve partnership) with each sub-project allocated a ‘champion’ from within the project. A Steering Group with representation across the COBWEB consortium was established to coordinate and ensure consistency. Throughout the 2015 and 2016 field seasons the seven co-design partners listed above had, between them, over a thousand people collecting data during fieldtrips of varying duration and frequency that they organised. This is a significant number of people considering that the Dyfi Biosphere Reserve area is mostly rural.
Regular workshops, interviews, participation in, and feedback from, the fieldtrips facilitated understanding throughout. Concentrating on the experience gained by the wide variety of users using the software and access to these users, the consortium used the field data and the intelligence flowing back from the interactions with the citizens and co-design partners to refine and improve the concept and system design.
Some lessons learned: the difficulty of enabling communication between scientists and ‘ordinary’ citizens to facilitate dialogue on sometimes obscure scientific concepts should not be underestimated. Interviews with volunteers participating in a project that can claim to be working towards better environmental governance (having Welsh Government as a partner on the project was important) proved to be a strong motivator. A particular challenge of this kind of co-design activity is the need to continually manage expectations - it was emphasised repeatedly to the sub-project leads and volunteers that they were participating in a research project which was testing and validating our concept of a citizen observatory; the components of which are at various TRL’s with research and development continuing in parallel. Significant technical support had to be allocated to respond quickly to prevent disillusionment with failing software.
Participants responded with enthusiasm to software releases where they could see the issues they identified being addressed and their ideas implemented. One of the outcomes has been establishment of an enthusiastic partner network interested in further work. The role of the Biosphere Reserve concept in facilitating sustainable development related activity was widely appreciated, particularly by the seven co-design partner organisations.
COBWEB Workflow:
The concept of surveys is central to COBWEB. A survey represents the context for a group of citizens going out into the field with one or more tasks over a period of time in a certain area. It constitutes a set of forms, workflows (business processing rules applying to observations) and datasets, all of which may be adjusted as the needs of a particular survey dictate.
COBWEB has found that, depending upon exact circumstances and the amount of control required, several key roles may be identified (Principal Investigator, Project Coordinator, Administrator, Survey Manager, Quality Reviewer, Citizen, Consumer, Publisher, Semantics Expert) in order to enact surveys. These roles are of broad applicability, although, depending upon circumstances, they may overlap.
COBWEB set out with the aim of being generic; the same software can be used to enable observations on a wide variety of environmental parameters. Surveys are authored by the ‘Survey Manager’ in order to meet specific requirements. When doing so and if required, the ‘Quality Reviewer’ liaises with the sponsor to ensure that quality control processes are established that meet their needs and which generates sufficient metadata to enable ‘fit for purpose’ decisions to be made by consumers, ie, is the data being described by the metadata suitable, appropriate and of sufficient quality for the envisaged use.
Quality control may require diverse input. For example; it may be advantageous or necessary to have data from a variety of sensors; these in turn can be a combination of physical sensors embedded in the environment, cyber sensors leveraging online data, or social sensors leveraging social media streams. Similarly, it may desirable to configure access to authoritative data from SDI type initiatives such as GEOSS, INSPIRE network services and national SDI’s. This may be necessary for conflation purposes (relating and combining spatial data from various sources); as part of quality control or to assist with visualising the results.
Architecture:
The system architecture necessary to support the above was developed through a combination of rapid prototyping and new software builds informed by requirements derived through co-design activities and other methods of stakeholder engagement. The architecture consists of the following key components: portal website, generic application designer, apps, storage middleware, quality assurance and conflation, sensor networks, whilst also implementing access control and privacy, and open standards. It is a feature of COBWEB that the components we have developed embed within an architecture that uses existing open source projects.
Portal Websites:
This is the main point of entry to COBWEB instances, providing access to the functionality documented in the sections below. In the simplest case, this is where users login and request to join selected citizen. An example was integrated with the Dyfi Biosphere Reserve website for demonstration purposes. Once users have contributed observations via their mobile device, results are available for visualisation via the portal using Web Map Services (WMS), Web Feature Services (WFS) or Sensor Observation Services (SOS).
At the heart of the portal is the latest version (3.0) of GeoNetwork, extended under COBWEB to facilitate typical citizen science use cases. GeoNetwork is an open source catalogue application offering resource registration capabilities and the ability to expose those resources for discovery via a number of standards, most notably those of ISO/TC 211 and the Open Geospatial Consortium (OGC), e.g. the OGC Catalogue Services for the Web (CSW) standard. Metadata can be exported to other standards such as Dublin core, DCAT, and schema.org.
Survey discovery: After having been created, surveys can be discovered by any interested parties via the portal. Making surveys discoverable using accepted and widely used open standards helps avoid waste, duplication, assists with recruiting volunteers and facilitates research and development in the field.
Data discovery: Making citizen science data discoverable online, within communities like GEOSS and INSPIRE, is more challenging than project or survey discovery and there are strong overlaps with the work underway developing a harmonised common data model and assessing data quality. Questions addressed by COBWEB include:
• What level of aggregation is used to determine quality: dataset or feature level?
Many user groups are interested in datasets as a whole (‘what is the spatial distribution of that species’). Other users may be interested in selected features (meaning abstractions of real world phenomena – buildings, rivers, plants, animals) and want to put questions such as ‘when did that particular observation of the species of interest occur’. Quality in COBWEB is therefor determined at both the feature and dataset level.
• How is the quality of datasets and/or features represented?
Quality varies at the feature level, e.g. different observations in the same survey may be of different quality, it may be less certain that a particular species has been observed even though the observer is the same. Quality is therefor best represented as part of the data and we recommend storing quality elements as part of the data.
• Is dataset quality assessed by aggregating the quality level of each of the observations in the dataset?
Some quality elements evaluated at feature level can be aggregated at dataset level for rapid assessment. Some quality only makes sense at the feature level. The ability to visualise quality cartographically may be useful in many circumstances.
• What schema is being used?
For metadata the iso19139 and iso19157 schemas are used, but they can be exported on request to Dublin Core, DCAT (Data Catalog Vocabulary) or schema.org . For datasets a generic SWE4CS model is proposed, which allows extension to include any external ontologies.
Access control and privacy:
Citizens can participate in COBWEB surveys anonymously (for reasons of privacy and data sharing this can be desirable) or they can register by providing identity information to user management systems integrated with the portal. Whether they do so or not depends on the survey; three kinds of surveys are distinguished in COBWEB:
1. Public survey. Any user, anonymous or registered, can join and contribute.
2. Registered survey. Only registered users can see the survey in the portal listings and contribute. An example of where this might be required is where the identity of the user making the observation is being used as a proxy for the quality of the observation, e.g. where the observer is expert in the field of study and recognized as such by the project coordinator and/or the principal investigator.
3. Private survey. Invited registered users only.
The need for this level of control was identified at the beginning of the design process where initial stakeholder engagement revealed a desire to be able to make and share observations of protected species and consequently, a requirement to be able to control access to sensitive data; for example, species protected under the UK Wildlife and Countryside Act or listed in the Natural Resources Wales Sensitive Species List.
It is not desirable or permitted to make publically available over the web detailed information on the location of these species without any access control. Conversely, these are often the most valuable data for consideration in environmental monitoring, and exactly the kind of information most needed for management and policy purposes.
The need for a reusable, well understood, high TRL, solution to security interoperability (sharing restricted data using standards across organizational boundaries) is a common SDI requirement. Throughout the extensive stakeholder engagement conducted by COBWEB, we found this requirement recurring over and over again in citizen science scenarios. The corollary, existing solutions that do not take security (interoperable or not) into account is also commonplace. The private survey concept developed by COBWEB is a powerful feature of the platform.
In addition to data security, concomitant questions of privacy were also a requirement within COBWEB. It is essential to enable users to register using personal information so that decisions concerning what they are authorised to access and contribute towards can be made. In the current citizen science landscape, identity information is also frequently used for quality assurance purposes and verifying observations.
Based upon the OASIS Security Assertion Markup Language (SAML) standard, the key advantages of the access management federation approach are that it is a proven, high TRL, industry strength solution that allows Single Sign On (SSO) to protected web based resources across administrative domains. Users can login using their familiar organisational credentials (presented via a SAML Identity Provider (IdP)) and then access protected resources (typically OGC web services in SDI scenarios) presented as SAML Service Providers (SP). Users only need to login once (SSO) and can then access multiple protected resources in the federation, providing they are suitably authorised.
Fine grained access control is established based upon attributes of the users and the actions on resources they want to do. The OGC’s GeoXACML standard, which extends the OASIS eXtensible Access Control Markup Language (XACML) to enforce geo-specific constraints, is another mature standard with high TRL guarantee. In COBWEB different levels of access are enforced to ensure the privacy of citizens that collected information and also the wellbeing of the observed species.
For the COBWEB framework, this means that surveys can be configured which access both unprotected and protected data sources, e.g. for conflation and/or quality assurance purposes. Private surveys can be setup which enable citizens to collect and share protected data with public authorities in compliance with both data protection and wildlife protection legislation. Public authorities can leverage the full benefits of interoperability, for example, by sharing sensitive data across organisational boundaries with authorised individuals using protected OGC web services, and potentially accessing all citizen sourced data without recourse to non-interoperable, one-off, bespoke security point solutions or lossy mechanisms such as anonymisation, obfuscation, reduction of spatial resolution, etc.
Generic application designer:
For each community, there is at least one ‘Survey Manager’ whose privileges entitle them to setup and create surveys. COBWEB employs a hybrid App approach enabling survey managers to build custom data collection forms using the generic application designer component of the portal website. Forms designed at the portal can then be synchronised with the generic COBWEB application on individual user’s mobile devices. The generic and extensible application designer supports a wide variety of form elements to cater for a broad range of user requirements.
Similar framework approaches to creating citizen science projects are not uncommon as exemplified by initiatives such as Open Data Kit, Indicia and EpiConnect amongst others. Some of the aspects that help distinguish COBWEB are its flexible and powerful approach to automating quality control using multiple inputs, use of an interoperable security solution and adherence to open standards wherever possible.
Mobile applications (Apps):
The generic Application solution offers the citizen the ability to install onto their mobile device, and login using an appropriate identity provider. The citizen then has the ability to either contribute to a directly available survey (public or registered) or a private survey which they have been invited to by the Survey Manager. The citizen will then be presented with the form designed by the Survey Manager, allowing them to participate in data collection with or without network coverage.
In addition to this, the generic capability described above is complemented by functionality allowing the cacheing of high quality basemapping on individual handsets for use in areas of poor or no network coverage. The background mapping used by COBWEB in the UK has been created using a variety of open products (mostly OpenStreetMap and Ordnance Survey Open Data Products) cartographically optimised for zooming in and out on mobile devices in both urban and rural areas. Investigation using Greek data has illustrated the difficulty of replicating this rich open map stack across Europe.
To demonstrate the effective ‘separation of concerns’ in the architecture, and how the COBWEB framework can be used in scenarios where lower level access to inbuilt mobile device functionality is required, a native Application in the flooding thematic area case study area was also developed. This uses the same interface as the generic application for communicating with the ‘Storage Middleware’.
Storage middleware:
Storage Middleware receives the observations from the App. As long as Oauth v2 authorisation is supported, the Storage Middleware component provides a generic REST-based API accessed storage compatibility layer on top of a range of cloud based providers (Google Drive, Dropbox, etc.) or physical storage media where local storage is required.
Storage Middleware is a central component of the COBWEB architecture used for managing survey schemas and exporting geospatial observations to the desired encodings, e.g. KML (Keyhole Markup Language), Geopackage, Shapefile, GeoJSON, CSV, etc. By synchronising all stored information with a relational database (the pre-quality PostGIS database in Figure 2) export of data via OGC Web Services (WMS/WFS) is supported.
Quality assurance and conflation:
Since the emergence of citizen science as a means of supporting scientific research, data quality has been considered an important issue for data use. Quality assurance accordingly plays an important role for the analysis of data obtained from citizen science and for the set-up of citizen science projects. It is commonly agreed that unknown data quality makes citizen science data of limited use. COBWEB’s approach to quality assurance is twofold: first, a number of quality measures are determined based on the variety of data provided by crowdsourcing activities, observations from the co-design projects, sensor feeds and social media; secondly means to express those measures in observation metadata is provided. This also includes an approach to link and conflate observations with relevant external datasets on the Web.
To allow for a customizable and dynamic quality assurance, remaining independent from actual citizen science projects, COBWEB uses a standards based web service chaining approach, and thus enables survey designers to adapt and adjust the quality assurance process to their needs. This generic approach on QA is necessary, because the relevance of specific quality control processes is usually highly use case dependent. The solution extends a pre-existing typology of quality assessment types to seven categories (or pillars) covering a range of specific quality controls generating quality metadata elements.
The implementation of the quality assurance process is based on the Business Process Model and Notation (BPMN) and OGC Web Processing Service (WPS) standards. This combination allows for the definition and dynamic binding of atomic quality control processes, encapsulated by WPS interfaces, to be used in a workflow environment. Therefore, the JBPM suite (workflow editor and workflow engine) has been customized to work with OGC services and has been integrated as a component of the COBWEB portal. Thus, each survey manager has the authority to create quality assurance workflows for their particular survey.
Capabilities to link and conflate citizen science observations with external data are designed and implemented to 1) assist the quality assurance process and 2) infer spatial information from the identified spatial data relations. Since all conflation processes are also offered via the OGC WPS interface, seamless integration and use by the workflow engine is assured. Currently, identified links of an observation can be attached as additional attributes or stored as RDF (Resource Description Framework) triples in a Linked Data store. Whereas the first option is primarily used to enrich and validate observations, the latter option allows for the reasoning on spatial data relations by means of the Semantic Web, in particular by the application of SPARQL (SPARQL Protocol and RDF Query Language) queries.
Sensor networks:
Though data collection from mobile devices is fundamental to the generic COBWEB framework, a variety of sensor platforms, monitoring multiple environmental parameters, within the Dyfi Biosphere Reserve testbed area, have been deployed. These sensor networks have been developed mainly as a result of dialogue between research scientists working on COBWEB and the co-design partners.
There are two ways by which sensor data are incorporated into the storage middleware:
1. Automated sensing. The physical sensors embedded in the environment are augmented with telemetry capabilities that enable them to operate in an autonomous manner, without human intervention, in transmitting data to a base station.
2. Participatory sensing. Data collection by citizen participating in surveys when in close proximity to suitably equipped sensors. The survey participant’s mobile device connects directly to the sensor using short range communication and the citizen physically relays the data back to a position where the data can be uploaded to the COBWEB servers
Physical sensors: Both participatory and automated approaches were researched and developed within COBWEB. Wasp Motes were used for participatory sensing. Surveys are configured such that when a citizen comes within range of one of the Wasp Motes, it is discovered by the citizen’s mobile device. When an observation is made, the mobile communicates with the sensor via the Bluetooth LE standard and the environmental parameters of interest are simultaneously captured. Such data can be used in quality control or to otherwise supplement the metadata associated with that observation.
The In-Situ and Davis Weather station sensors are capturing data automatically and transmitting back to the COBWEB servers. Data from the In-Situ sensors are being used in association with habitat reversion at co-design partner RSPB’s Yns-Hir reserve. Data from the weather stations are being used in educational scenarios with co-design partner Ysgol Bro Hyddgen and to investigate pollinator scenarios with co-design partner Penparcau Community Forum. In the pollinator scenarios, citizens go out on a regular basis to record observations of butterflies and their larval foodplants around Penparcau in the Dyfi Biosphere Reserve. The data recorded includes plant/butterfly species, plant coverage, butterfly activity, and habitat information.
Weather data (temperature, wind speed etc.), from weather stations, sited in strategic locations, is used in a Twitter alert service to send tweets to citizens following a dedicated Twitter account to encourage participants to go out and collect data when the weather is suitable for butterfly activity. The weather data is also used post-data collection to validate and add value to the observations.
When collecting data within the field, sensors on board the user’s mobile device are harnessed to provide additional information. For example, accelerometers values are recorded when the user is taking photographs to enable the line of sight to be determined - useful for various quality assurance purposes. It should be noted that the generic COBWEB App accesses this functionality via a Cordova plug-in while the native flooding App obtains the data directly from the sensor.
Cyber sensing: The harvesting of online geotagged sensor data from a variety of sources through different web-based API’s. In COBWEB, the approach has been demonstrated using data accessed through the Shoothill (UK river levels) and Weather Underground (network of personal weather stations) API’s.
There are two ways in which sensor data are associated with surveys within COBWEB. In the case of participatory sensed data, sensor data are included as part of the survey data and exported to the pre-quality database via the storage middleware in the usual manner. Otherwise, for both cyber and physical sensors, the data are stored directly in the pre-quality database either by sensor nodes transmitting the data in the case of physical sensors or by the harvester in the case of cyber sensors.
Social sensors: Leveraging social media streams. In COBWEB, research and development has concentrated on the use of Twitter and Flickr API’s. Data from both social media platforms are acquired by two methods.
1. Bounding boxes for Biosphere Reserves are used to monitor and capture any social media postings within these areas.
2. Keywords are used to identify relevant postings worldwide.
Once social media postings have been captured they are stored in a database and are available for viewing at the portal or potentially for use within a Quality Assurance workflow.
Standards and Spatial Data Infrastructures:
COBWEB aimed to maximise technical interoperability by the use of standards wherever appropriate - the sections above make multiple references to standards from several standards defining organisations such as, OGC, ISO/TC 211, OASIS and W3C.
All the citizen observatory’s developed under the FP7 call which funded COBWEB were required to make data collected available within GEOSS without restriction.
Publishing data into SDI-like initiatives such as GEOSS was addressed using a cooperative approach based mainly on the processes of the OGC. For example: the project contributed to three OGC led GEOSS Architecture Implementation Pilots; five ad hoc Citizen Science meetings were organised by COBWEB during OGC Technical Committee meetings over the duration of the project, resulting in the formation of a fully constituted Citizen Science Domain Working Group.
One of the main results was the publication of the OGC Discussion Paper ‘Standardized Information Models to Optimize Exchange, Reusability and Comparability of Citizen Science Data’. These are the first steps on the standardisation ladder; broad community uptake of a harmonised common information model applicable to a wide range of citizen science scenarios would be a valuable contribution towards being able to share and future proof the value of data collected. Data compliant with this information model can be discovered and accessed through standardised web interfaces, e.g. OGC web services, and integrated with SDI’s as most, if not all, SDI’s are based on open geospatial interoperability standards.
Two approaches were explored within COBWEB: First, the definition of an application profile by specialising the Observations and Measurements (O&M), SensorML and SweCommon information models. Second, the definition of an O&M, SensorML and SweCommon encodings best practices to allow more efficient reuse of existing components.
Semantic interoperability is a major concern, especially considering the growth in the open data movement and their use of semantic web standards. Semantics is concerned with attaching unambiguous meaning to terms and concepts, typically, using ontologies (formal models describing semantics). In the citizen science domain, as in others, this is very important as understanding exactly what was observed and in which context is essential to making the data comprehensible; both for the immediate use, by others and in the future. The ability to share data, reuse, integrate and compare is often compromised by a lack of semantic interoperability. As would be expected in a research project whose main objective is developing a generic infrastructure that aims to satisfy requirements across as broad a range as possible of the multitude of potential citizen science scenarios, semantics was an area of intense research and development interest in COBWEB. The approach described below is documented in the OGC Discussion Paper referenced above.
In terms of the COBWEB workflow, semantics can be injected at various stages, but the best place to introduce semantics is when the survey context is defined. At this point, the goals of the project should be clear, and if not, then it is time well spent clarifying exactly what is expected to be observed and to what extent exact definitions can be incorporated. Ideally, the Survey Manager would liaise with the Principal Investigator and possibly also the Quality Reviewer and other experts as appropriate, to incorporate sufficient semantics to ensure that the information generated by the survey is of maximum usefulness to end Consumers.
During the Sept 2016 OGC Technical Committee meeting, a live demonstration of the ‘Interoperability Profiles (IP)’ approach developed under COBWEB was given. Using this method, the Survey Manager selects (or creates) an IP during survey design; the IP provides unambiguous references to authoritative definitions of concepts being used in the survey, in this case, it was a link to the relevant standards entry (biodiversity standard Darwin Core terms) defining the invasive species Japanese Knotweed. As a consequence of using this IP during survey creation, every time an observation of a Knotweed was made using this COBWEB survey, the data then contains references to the authoritative definition thereby ensuring semantic interoperability.
Note that this approach is independent of serialisation; the Citizen Science application profile enables encoding as XML (Extensible Markup Language), GeoJSON or linked data. The observations and raw data are of primary interest, various levels of processed or aggregated data may be required at different times depending upon circumstances. All types of observations are supported; including raw observations and those which are derived or result from subsequent quality assurance or conflation processing. A provenance model is included to provide information on the history of each observation.
Of course, widespread adoption and use of this approach presupposes that the relevant semantic resources have been made internet addressable using semantic web standards, that IP’s have been created and that the approach has received widespread community uptake. This is far from the case; however, the low TRL approach described above demonstrably works, can and is being used as a platform to stimulate further discussion, research and development.
Potential Impact:
COBWEB conducted research into the feasibility of creating a common generic framework for mobile device apps for use in citizen science for environmental monitoring, using the UNESCO World Network of Biosphere Reserves as a testbed. We have demonstrated that, in principle, it is possible and brings multiple benefits when starting from a position where the importance of interoperability and flexibility is paramount. This stands in contrast to the current proliferation of Citizen Science Apps generating silos of data.
Specifically, it has been shown that creating a generic solution to automating quality control and assurance, which is sufficiently flexible to address the huge range of potential scenarios, is beneficial to the reuse of citizen science data. Further development would result in the ability to make very large volumes of data useable, and is an area of active ongoing research.
Similarly, COBWEB has contributed to the understanding that further research and development is needed to address whether it is possible to create an easily useable framework which is sufficiently flexible to allow a broad range of different kinds of familiar semantic resources to be employed in designing surveys before citizens go into the field. Without this, despite post-processing server-side, continued problems associated with a lack of semantic interoperability may be anticipated.
Despite perceived complexity and proliferation, the use of open interoperability standards still presents the most realistic chance of preventing the waste of resources and reuse opportunities inherent in creating silos of data locked into proprietary solutions. In the citizen science domain, as in other domains that leverage SDI, the lack of a widely adopted solution to security interoperability is again apparent as a major barrier. At the technical level, COBWEB has shown that access management federations work; progressing this relies upon action at the organisational, legal and political level – outwith the scope of COBWEB.
However, standardisation efforts should continue and adherence be required to help realise investment in SDI type initiatives such as GEOSS.
COBWEB has had significant potential impact in highlighting that, if agreement can be reached on a harmonised common information model, with sufficient community support, most, if not all, crowdsourced, citizen science type data could be made compliant with the developing standardisation efforts initiated under COBWEB. The immediate prize would be a boost to the usefulness of these data by reducing integration costs and enabling the myriad of potential consumers of such data to exploit existing standards based tooling and develop new standards based solutions on top. COBWEB has also contributed to the understanding that the long term curation value of holding citizen sourced data at the observation level of granularity compliant with a well thought out, agreed and documented information model is to maximise potential future access and re-use.
The amount of citizen sourced environmental data will continue to increase dramatically and clearly has the potential to be of use for policy formation and delivery in respect of environmental decision making. At the global scale, towards the end of 2015, the UN adopted the 2030 Agenda for Sustainable Development, this included a set of 17 Sustainable Development Goals (SDG) to end poverty, fight inequality and injustice, and tackle climate change by 2030. The importance of geospatial information in monitoring progress with SDG targets is explicitly recognised in clause 76 of the official resolution (United Nations, 2015). It may be argued that citizen sourced geospatial information has a potentially important contribution towards realising the SDG’s worthy of significant further research.
In this respect, based on experiences in COBWEB, we would recommend increased focus on the UNESCO World Network of Biosphere Reserves (WNBR) as a testbed for sustainability science related research and development. This is particularly appropriate as at the 2016 Fourth World Congress of Biosphere Reserves, the UN’s Man and Biosphere Programme resolutely aligned the WNBR with the SDG’s in the resulting Lima Action Plan .
Using co-design processes and consortia such as those employed in COBWEB, WNBR mechanisms can be leveraged to more easily mobilise sufficient numbers of enthusiastic citizens in order to help realise meaningful outcomes based on real world needs.
The COBWEB framework solution described above is complex and significant further work is required to present only the required level of complexity, depending upon circumstances, when required. Integrating all the components (at different TRL’s) to create an actual system proven in an operational environment (TRL 9) was beyond the scope of this research project.
With the direct involvement of OGCE on COBWEB as a full partner, the use of interoperability standards has always been central to the project and will be key to exploitation of project results. Towards the end of this four year project, in June 2016, COBWEB sponsored the OGC Technical Committee (TC) meeting in Dublin. The project was highly visible and was represented multiple times during the week, including a half day ‘COBWEB Summit’ entirely dedicated to those aspects of the project within the scope of a standards defining organisation.
An ad hoc Citizen Science meeting was also held during the TC week; this was the fourth of five ad hoc meetings organised by COBWEB at TC’s held over the course of the project at various locations around the globe: Barcelona, Sydney, Washington, Dublin and Orlando.
The culmination of this discussion and community consultation came at the Sept 2016 Orlando TC when the charter for the new Citizen Science Domain Working Group (DWG) was voted on and approved, as was the release of the OGC Discussion Paper ‘Standardized Information Models to Optimize Exchange, Reusability and Comparability of Citizen Science Data’ (published March 2017). COBWEB led on all the above and is explicitly acknowledged as doing so, project representatives in Orlando concluded by giving a live proof-of-concept demonstration of the documented approach in action.
At the time of writing, the work described above has already had impact and is a milestone in the standardisation of citizen sourced data. Several recently funded Horizon 2020 projects referenced the work in their proposals and are exploiting COBWEB results; this includes NextGEOSS and all four projects funded under H2020 SC5-17-2015: Demonstrating the concept of 'Citizen Observatories' - the latter was evident at the Brussels cluster meeting attended by the author in Nov 2016. Indications are that the OGC Citizen Science DWG will continue to be an important global forum for discussing common approaches to interoperability challenges in the citizen sourced data domain.
Longer term, the work may lead to the formation of a Standards Working Group and eventually a full international standard with broad uptake. As noted above, the potential impact of this is great; vast amounts of potentially valuable citizen sourced data are currently being locked in silos and opportunities are being lost now and will continue to be in the future because adherence to a well thought out data model is not happening. Whether an accepted standard emerges via the OGC route or not is dependent upon multiple different factors now outside the influence of COBWEB, although the COBWEB coordinator hopes to positively influence this as the current European co-chair of the new Citizen Science DWG.
Various other aspects of COBWEB have also filtered through into new projects. One of the H2020 SC5-17-2015 projects – LandSense, is reusing elements of the COBWEB approach to Quality Assurance (QA) and security.
As noted above, a major criticism of citizen sourced data is that it is often of insufficient quality to be of use for the kinds of scenarios typical of policy formation and delivery. The situation is compounded by the current lack of automated QA methods and bottlenecks caused by reliance upon experts visually checking the quality of submitted observations. Raising the TRL of the highly flexible standards based web services chaining approach, developed under COBWEB, to the point where it could be used in production to automate QA and add appropriate metrics to observation metadata would be valuable indeed.
The situation with the access management federation approach used for security and privacy in COBWEB is different. This is already proven to be high TRL, key components are, in some cases, eg, SAML based access management federations in the academic sector, already in use by literally millions round the globe. The barriers to widespread uptake are not technical; they are political, organisational and financial. Long term impact will depend upon whether other solutions to sharing data across administrative domains using interoperability standards emerge, or whether the lack of a widely accepted approach increases uptake. Over the lifetime of the project, COBWEB used the regular GEOSS Architecture Implementation Pilots to disseminate our knowledge of this area - current indications are that the lack of a widely adopted interoperable security solution continues to be an issue.
Another key dissemination channel used for summarising and communicating project results was the scientific paper ‘Citizen OBservatory WEB (COBWEB): A Generic Infrastructure Platform to Facilitate the Collection of Citizen Science data for Environmental Monitoring’ published in a special Citizen Science edition of the Joint Research Centres (JRC) International Journal of Spatial Data Infrastructures Research. Writing this paper in the final year of the project coincided with the special edition and JRC hosting a two day workshop (data and service infrastructures for Citizen Science) during Jan 2016 of a global grouping centred around the European Citizen Science Association focussed on interoperability of Citizen Science data. JRC subsequently agreed to be charter members of the new OGC Citizen Science DWG.
Other core dissemination activities communicating results towards the end of the project were a COBWEB article intended for a more general audience in the widely distributed Impact magazine and, with greater novelty value, the comic book: Crowd Power: the COBWEB Guide to Citizen Science, available off the COBWEB website . Here you can also find a short documentary video comprised of interviews with key project personnel recorded during the final consortium meeting in Sept 2016 .
From the outset, COBWEB committed to open standards, open data and open source software. In the last quarter of the project, effort was expended in open sourcing those high TRL outputs that the consortium considered to have the greatest exploitation potential. This resulted in the creation of Fieldtrip Open as a separate entity with an existence beyond the end of COBWEB.
Fieldtrip Open is comprised mainly of the generic COBWEB App, the survey designer and the middleware. It is fully documented and extensible allowing the incorporation of quality control processes and additional controls as required. Observations can be published in a variety of different formats/schemas and made available via OGC web services. The intention is that the software can be used with other well known open source geospatial projects, eg, GeoNetwork, that it forms the basis for multiple initiatives and that those benefiting contribute back to the codebase to create a viable sustained open source community.
In March 2017, the prospect of the latter happening received a boost through the Welsh Government’s awarding the year long Welsh Citizen Observatory Pilot (WCOP) contract to a subset of the COBWEB consortium. WCOP was created to sustain the existence of the COBWEB infrastructure as deployed in the Dyfi Biosphere Reserve to allow further opportunities to influence citizen science projects to adopt more mature data collection and management strategies and to demonstrate the benefits of standards based technology and open data models for government. The focus for WCOP will be on projects identified by Welsh Government as having the greatest impact in terms of environmental decision making and policy. Because the software is generic, flexible and does not require software engineering expertise to create surveys, the expectation is that it will also continue to be used by a variety of community groups, including those involved in co-design during the execution of COBWEB.
The Commission’s investment in COBWEB has clearly had an impact, this is particularly evident in how citizen involvement using new technology in environmental governance is viewed in Wales. This may be taken as additional evidence of the benefit of having direct involvement of government bodies in large collaborative research projects that aim to have policy impact. As Welsh Government is a devolved administration, exploitation of COBWEB results by the public sector to realise policy objectives is having impact, and will continue to have impact, not only in Wales, but across the UK through national fora such as the UK Environmental Observation Framework.
List of Websites:
https://cobwebproject.eu/
COBWEB was a research project that succeeded in its goal of developing a generic crowdsourcing infrastructure platform and toolkit that could be used in multiple scenarios generating data of sufficient quality to be used by policymakers. At the end of the project, different components comprising the platform are at different Technology Readiness Levels (TRL); with some high TRL outputs being open sourced, eg, Fieldtrip Open.
Achieving an understanding of how citizen science can be harnessed in the service of societal goals is important. There is a veritable deluge of data being generated by citizens and some of that data is useful (or can be made useful) for environmental decision making. Perhaps more importantly, the technology is creating opportunities for greater citizen involvement in environmental decision making. For example, it has been observed that realising the UN’s Sustainable Development Goals will require massive citizen support. COBWEB made a contribution, most significantly in the following areas: standardisation, quality control, security/privacy and use of the UN’s World Network of Biosphere Reserves (WNBR).
A new Open Geospatial Consortium (OGC) Citizen Science Working Group has been established and looks set to become a key forum for advancing related global interoperability issues. COBWEB’s Sensor Web Enablement for Citizen Science (SWE4CS) initiative is a first step in the standardisation process and led the way in developing a harmonised information model for exchanging CS data and addressing semantic interoperability issues. The high TRL access management federation approach advocated by COBWEB for managing privacy and security has had impact on GEOSS and is being used by followon projects. Quality control continues to be a major concern for initiatives dealing with crowdsourced data; COBWEB pioneered a web services chaining approach that allows great flexibility in the processes applied in response to the multitude of different circumstances under which such data is collected. This too is being further developed by followon projects.
Using a unique co-design approach, COBWEB concentrated on mobilising citizens within Biosphere Reserves. We found the WNBR to be an excellent (and underused) resource for this kind of research project; where access to an already incentivised and organised pool of citizens enthusiastic about the sustainable development agenda is a definite advantage.
Having a government involved as a full partner in the project also proved to be an advantage; COBWEB benefited from direct communication channels with bodies responsible for setting and delivering policy relating to environmental data. This impact continues post project as Welsh Government is sustaining the COBWEB platform for a period in support of environmental decision making and policy.
Project Context and Objectives:
Project context:
Key context for COBWEB is the huge upsurge in the availability of mobile devices of increasing sophistication with concomitant increase in the volume of citizen sourced data. The big challenges were, and remain, how to make these data useful for better decision making and harnessing the citizen enthusiasm and technology in the service of societal goals. In some areas, COBWEB made some significant advances which are being leveraged post-project by various initiatives, including Citizen Observatory related followon projects funded under H2020.
COBWEB was predicated upon UNESCO’s World Network of Biosphere Reserves (WNBR) and the original proposal referenced the UN’s Millennium Development Goals. Over the period of the project, the Sustainable Development Goals (SDG) were published, including a clear statement referencing the importance of geographic information. The UN’s Man and Biosphere programme responded by aligning with the SDG’s as did multiple other public authorities at all levels.
The chief expected impact from COBWEB may be summarised as enabling greater citizen influence in environmental governance. The project set out to use a small number of Biospheres in Wales, Greece and Germany as representative testbeds for developing and validating the technology developed. In the process, we developed a unique co-design approach to respond to community priorities, mobilise citizens and help gauge the effectiveness of the WNBR concept. This aligned with a general shift towards co-design/co-creation and encouragement of greater citizen involvement in environmental decision making. Welsh Government was a full partner in the project and contributed substantially towards understanding the shifting context in respect of environmental policy development and delivery. They continue to support the deployment of the COBWEB platform in mobilising citizen’s associations post-project.
Standards were an important organising principle for COBWEB from the start and formed much of the context for the work undertaken. Over the period of the project, semantic web standards underpinning the open data movement continued to gain traction; representing a challenge in terms of compatibility with more traditional open geospatial interoperability standards. The toolkit COBWEB created and open sourced (Fieldtrip Open) can publish data in various formats depending upon what is required, eg, INSPIRE specific schema, linked data, GEOJson, or Sensor Web Enablement for Citizen Science. SWE4CS ‘Standardized Information Models to Optimize Exchange, Reusability and Comparability of Citizen Science Data’ is a first step on the standardisation ladder and was originally conceived as a standard for publishing citizen science data into GEOSS.
Linking with authoritative data from initiatives such as INSPIRE and GEOSS is part of a broader movement within the citizen sourced data community towards leveraging the wealth of related data from different sources for different purposes. COBWEB’s main contributions were in conflation/fusion with Spatial Data Infrastructure (SDI) initiatives like INSPIRE for display or analysis purposes, conflation with data from sensors, and Quality Assurance (QA). Note there is overlap between the latter, for example, SDI data can be used for QA, social media type data, eg, Twitter, Facebook, can be modelled as sensors.
An overarching problem in Citizen Science is that large quantities of data are being created but exist in silos. Useable standards either don’t exist, are neglected, poorly understood or tooling is unavailable. It is still early days and this is reflected in the different Technology Readiness Levels (TRLs) of outputs from research projects such as COBWEB. SWE4CS for example, is at an intermediate TRL. Similarly, the standards based web services chaining approach to QA pioneered by COBWEB allows great flexibility in the data brought to bear and processes applied in response to the huge variety of different Use Cases encountered in the Citizen Science space.
The access management federation approach advocated by COBWEB for managing privacy and security is high TRL and is likely to have continuing impact. This is because of the continued lack of a widely accepted, robust, practical way of securely sharing valuable data across administrative domains. The requirement to be able to securely share (with a high level of assurance) in a genuinely interoperable way controlled data such as the location of endangered species, and the associated need to maintain individual privacy, is a constant in the volunteered geographic information, crowdsourcing, citizen science context within which COBWEB was executed.
Project objectives:
COBWEB set out to research and develop a “generic crowdsourcing infrastructure platform” manifested as a toolkit which could be downloaded and used in multiple scenarios. The citizen observatory platform should be capable of integrating a wide variety of different information sources such as physical sensors embedded in the environment, cyber sensors leveraging online data, or social sensors leveraging social media streams.
One of our objectives was to explore the potential for integrating the platform into Biosphere information systems and evaluating the use of the WNBR as an integral part of projects like COBWEB, which require active cooperation of citizens and citizen groups.
Addressing the question: how can the business of government be improved using citizen science data, empowering citizens, enabling contributions towards better environmental governance in keeping with policy aims were overarching objectives for all the citizen observatory projects funded by the EU.
Meeting real world needs and addressing community priorities in a flexible way was a feature of COBWEB. To give the project focus, COBWEB committed to, and concentrated on, delivering outputs in three pilot case study areas:
1. Creation and validation of data products from Earth Observation data
2. Biological monitoring
3. Flooding
Policy objectives were addressed at a local, national, European or global level as appropriate. Helping realise the benefits of existing investment in SDI initiatives such as INSPIRE and GEOSS, and improving interoperability through adherence to open standards and by using the processes of the standards defining organisations was an objective of COBWEB throughout. This informed the design and development of the platform, especially in respect of the science and technology approaches used to developing solutions for security, privacy, quality assurance and conflation.
Project Results:
The following description of the main Science and Technology results from the COBWEB project draws extensively from the scientific paper ‘Citizen OBservatory WEB (COBWEB): A Generic Infrastructure Platform to Facilitate the Collection of Citizen Science data for Environmental Monitoring’ (http://ijsdir.jrc.ec.europa.eu/index.php/ijsdir/article/view/406/393).
The COBWEB project provides open source tooling which can be downloaded , configured and modified if necessary to enable citizen observatory’s – infrastructure which enables citizens to use their own mobile devices to make observations across a wide range of different citizen science scenarios. The platform developed and described below is generic, extensible and powerful enough to accommodate sophisticated requirements arising from a broad range of stakeholders across different sectors, yet sufficiently flexible to enable non-experts to create and use mobile applications which meet their needs.
It should be noted that COBWEB was a research project. As such, the various components described below are at different levels of technical maturity ranging from Technology Readiness Level (TRL) 4 (validated in laboratory) to TRL 9 (actual system proven in operational environment). The TRL scale is used as a tool for decision making on research and development investments at the EU level and has been developed to enable assessment and comparison of technologies in respect of maturity.
COBWEB provides high TRL tooling for collecting new data and, where possible, the software developed leverages and complements existing well established high TRL open source projects such as those in the OSGeo suite. COBWEB has also conducted research (low TRL) into the use of other crowdsourced data, for example, from social media streams and sensors, for both quality assuring and enriching observations. The latter is an example of the benefits of leveraging SDI as standards are critical for relating and combining (the terms fusion and conflation are also used) spatial data from various sources.
System design through stakeholder engagement and co-design:
COBWEB commenced with a process of requirements gathering for a system that would enable citizens in the Dyfi Biosphere Reserve area in Wales to collect data in the 3 pilot case study areas as introduced above. Requirements were initially gathered through a process of desk study reviewing existing systems and structured interviews. The latter started with citizen groups located within the Biosphere Reserve and were widened and further refined through broader stakeholder engagement; in particular with groups associated with environmental governance and stewardship.
In parallel, a process of rapid prototyping software development was entered into with the intention of creating software that demonstrated the feasibility of meeting identified requirements and surfacing issues and challenges. Once we had reached the requisite level of software maturity, to further assess the viability of the proposed solution and better understand the needs of the citizen, COBWEB then engaged in a period of structured co-design activity.
Co-Design:
Co-design happens where “participants are invited to cooperate with designers, researchers and developers during an innovation process”. Seven citizen groups active in the Dyfi Biosphere Reserve in mid-Wales were commissioned through an open tendering process to help gather requirements and validate our concept of a citizen observatory by mobilising citizens (usually volunteers) to go out in the field to collect data.
The invitation to tender was issued via the Dyfi Biosphere Reserve and COBWEB websites and resulted in a strong response from the community within the Biosphere Reserve expressing a desire to work with an unexpectedly (from the project consortiums perspective) rich set of environmental phenomena. This generated a diverse set of requirements and may be interpreted as evidence of the Biosphere Reserve mechanisms successfully working to help meet the sustainable development agenda in the context of a research project which needs to mobilise citizens.
Each citizen group’s proposal was analysed and refined in respect of what it could offer the project, e.g. research potential, policy implications, technical challenges, ethical implications, etc. The result was a challenging set of requirements to guide system design, based on real user needs, and the opportunity to predicate further research on real world requirements.
The seven co-design sub-projects were programme managed by COBWEB partner Ecodyfi (the delivery arm of the Dyfi Biosphere Reserve partnership) with each sub-project allocated a ‘champion’ from within the project. A Steering Group with representation across the COBWEB consortium was established to coordinate and ensure consistency. Throughout the 2015 and 2016 field seasons the seven co-design partners listed above had, between them, over a thousand people collecting data during fieldtrips of varying duration and frequency that they organised. This is a significant number of people considering that the Dyfi Biosphere Reserve area is mostly rural.
Regular workshops, interviews, participation in, and feedback from, the fieldtrips facilitated understanding throughout. Concentrating on the experience gained by the wide variety of users using the software and access to these users, the consortium used the field data and the intelligence flowing back from the interactions with the citizens and co-design partners to refine and improve the concept and system design.
Some lessons learned: the difficulty of enabling communication between scientists and ‘ordinary’ citizens to facilitate dialogue on sometimes obscure scientific concepts should not be underestimated. Interviews with volunteers participating in a project that can claim to be working towards better environmental governance (having Welsh Government as a partner on the project was important) proved to be a strong motivator. A particular challenge of this kind of co-design activity is the need to continually manage expectations - it was emphasised repeatedly to the sub-project leads and volunteers that they were participating in a research project which was testing and validating our concept of a citizen observatory; the components of which are at various TRL’s with research and development continuing in parallel. Significant technical support had to be allocated to respond quickly to prevent disillusionment with failing software.
Participants responded with enthusiasm to software releases where they could see the issues they identified being addressed and their ideas implemented. One of the outcomes has been establishment of an enthusiastic partner network interested in further work. The role of the Biosphere Reserve concept in facilitating sustainable development related activity was widely appreciated, particularly by the seven co-design partner organisations.
COBWEB Workflow:
The concept of surveys is central to COBWEB. A survey represents the context for a group of citizens going out into the field with one or more tasks over a period of time in a certain area. It constitutes a set of forms, workflows (business processing rules applying to observations) and datasets, all of which may be adjusted as the needs of a particular survey dictate.
COBWEB has found that, depending upon exact circumstances and the amount of control required, several key roles may be identified (Principal Investigator, Project Coordinator, Administrator, Survey Manager, Quality Reviewer, Citizen, Consumer, Publisher, Semantics Expert) in order to enact surveys. These roles are of broad applicability, although, depending upon circumstances, they may overlap.
COBWEB set out with the aim of being generic; the same software can be used to enable observations on a wide variety of environmental parameters. Surveys are authored by the ‘Survey Manager’ in order to meet specific requirements. When doing so and if required, the ‘Quality Reviewer’ liaises with the sponsor to ensure that quality control processes are established that meet their needs and which generates sufficient metadata to enable ‘fit for purpose’ decisions to be made by consumers, ie, is the data being described by the metadata suitable, appropriate and of sufficient quality for the envisaged use.
Quality control may require diverse input. For example; it may be advantageous or necessary to have data from a variety of sensors; these in turn can be a combination of physical sensors embedded in the environment, cyber sensors leveraging online data, or social sensors leveraging social media streams. Similarly, it may desirable to configure access to authoritative data from SDI type initiatives such as GEOSS, INSPIRE network services and national SDI’s. This may be necessary for conflation purposes (relating and combining spatial data from various sources); as part of quality control or to assist with visualising the results.
Architecture:
The system architecture necessary to support the above was developed through a combination of rapid prototyping and new software builds informed by requirements derived through co-design activities and other methods of stakeholder engagement. The architecture consists of the following key components: portal website, generic application designer, apps, storage middleware, quality assurance and conflation, sensor networks, whilst also implementing access control and privacy, and open standards. It is a feature of COBWEB that the components we have developed embed within an architecture that uses existing open source projects.
Portal Websites:
This is the main point of entry to COBWEB instances, providing access to the functionality documented in the sections below. In the simplest case, this is where users login and request to join selected citizen. An example was integrated with the Dyfi Biosphere Reserve website for demonstration purposes. Once users have contributed observations via their mobile device, results are available for visualisation via the portal using Web Map Services (WMS), Web Feature Services (WFS) or Sensor Observation Services (SOS).
At the heart of the portal is the latest version (3.0) of GeoNetwork, extended under COBWEB to facilitate typical citizen science use cases. GeoNetwork is an open source catalogue application offering resource registration capabilities and the ability to expose those resources for discovery via a number of standards, most notably those of ISO/TC 211 and the Open Geospatial Consortium (OGC), e.g. the OGC Catalogue Services for the Web (CSW) standard. Metadata can be exported to other standards such as Dublin core, DCAT, and schema.org.
Survey discovery: After having been created, surveys can be discovered by any interested parties via the portal. Making surveys discoverable using accepted and widely used open standards helps avoid waste, duplication, assists with recruiting volunteers and facilitates research and development in the field.
Data discovery: Making citizen science data discoverable online, within communities like GEOSS and INSPIRE, is more challenging than project or survey discovery and there are strong overlaps with the work underway developing a harmonised common data model and assessing data quality. Questions addressed by COBWEB include:
• What level of aggregation is used to determine quality: dataset or feature level?
Many user groups are interested in datasets as a whole (‘what is the spatial distribution of that species’). Other users may be interested in selected features (meaning abstractions of real world phenomena – buildings, rivers, plants, animals) and want to put questions such as ‘when did that particular observation of the species of interest occur’. Quality in COBWEB is therefor determined at both the feature and dataset level.
• How is the quality of datasets and/or features represented?
Quality varies at the feature level, e.g. different observations in the same survey may be of different quality, it may be less certain that a particular species has been observed even though the observer is the same. Quality is therefor best represented as part of the data and we recommend storing quality elements as part of the data.
• Is dataset quality assessed by aggregating the quality level of each of the observations in the dataset?
Some quality elements evaluated at feature level can be aggregated at dataset level for rapid assessment. Some quality only makes sense at the feature level. The ability to visualise quality cartographically may be useful in many circumstances.
• What schema is being used?
For metadata the iso19139 and iso19157 schemas are used, but they can be exported on request to Dublin Core, DCAT (Data Catalog Vocabulary) or schema.org . For datasets a generic SWE4CS model is proposed, which allows extension to include any external ontologies.
Access control and privacy:
Citizens can participate in COBWEB surveys anonymously (for reasons of privacy and data sharing this can be desirable) or they can register by providing identity information to user management systems integrated with the portal. Whether they do so or not depends on the survey; three kinds of surveys are distinguished in COBWEB:
1. Public survey. Any user, anonymous or registered, can join and contribute.
2. Registered survey. Only registered users can see the survey in the portal listings and contribute. An example of where this might be required is where the identity of the user making the observation is being used as a proxy for the quality of the observation, e.g. where the observer is expert in the field of study and recognized as such by the project coordinator and/or the principal investigator.
3. Private survey. Invited registered users only.
The need for this level of control was identified at the beginning of the design process where initial stakeholder engagement revealed a desire to be able to make and share observations of protected species and consequently, a requirement to be able to control access to sensitive data; for example, species protected under the UK Wildlife and Countryside Act or listed in the Natural Resources Wales Sensitive Species List.
It is not desirable or permitted to make publically available over the web detailed information on the location of these species without any access control. Conversely, these are often the most valuable data for consideration in environmental monitoring, and exactly the kind of information most needed for management and policy purposes.
The need for a reusable, well understood, high TRL, solution to security interoperability (sharing restricted data using standards across organizational boundaries) is a common SDI requirement. Throughout the extensive stakeholder engagement conducted by COBWEB, we found this requirement recurring over and over again in citizen science scenarios. The corollary, existing solutions that do not take security (interoperable or not) into account is also commonplace. The private survey concept developed by COBWEB is a powerful feature of the platform.
In addition to data security, concomitant questions of privacy were also a requirement within COBWEB. It is essential to enable users to register using personal information so that decisions concerning what they are authorised to access and contribute towards can be made. In the current citizen science landscape, identity information is also frequently used for quality assurance purposes and verifying observations.
Based upon the OASIS Security Assertion Markup Language (SAML) standard, the key advantages of the access management federation approach are that it is a proven, high TRL, industry strength solution that allows Single Sign On (SSO) to protected web based resources across administrative domains. Users can login using their familiar organisational credentials (presented via a SAML Identity Provider (IdP)) and then access protected resources (typically OGC web services in SDI scenarios) presented as SAML Service Providers (SP). Users only need to login once (SSO) and can then access multiple protected resources in the federation, providing they are suitably authorised.
Fine grained access control is established based upon attributes of the users and the actions on resources they want to do. The OGC’s GeoXACML standard, which extends the OASIS eXtensible Access Control Markup Language (XACML) to enforce geo-specific constraints, is another mature standard with high TRL guarantee. In COBWEB different levels of access are enforced to ensure the privacy of citizens that collected information and also the wellbeing of the observed species.
For the COBWEB framework, this means that surveys can be configured which access both unprotected and protected data sources, e.g. for conflation and/or quality assurance purposes. Private surveys can be setup which enable citizens to collect and share protected data with public authorities in compliance with both data protection and wildlife protection legislation. Public authorities can leverage the full benefits of interoperability, for example, by sharing sensitive data across organisational boundaries with authorised individuals using protected OGC web services, and potentially accessing all citizen sourced data without recourse to non-interoperable, one-off, bespoke security point solutions or lossy mechanisms such as anonymisation, obfuscation, reduction of spatial resolution, etc.
Generic application designer:
For each community, there is at least one ‘Survey Manager’ whose privileges entitle them to setup and create surveys. COBWEB employs a hybrid App approach enabling survey managers to build custom data collection forms using the generic application designer component of the portal website. Forms designed at the portal can then be synchronised with the generic COBWEB application on individual user’s mobile devices. The generic and extensible application designer supports a wide variety of form elements to cater for a broad range of user requirements.
Similar framework approaches to creating citizen science projects are not uncommon as exemplified by initiatives such as Open Data Kit, Indicia and EpiConnect amongst others. Some of the aspects that help distinguish COBWEB are its flexible and powerful approach to automating quality control using multiple inputs, use of an interoperable security solution and adherence to open standards wherever possible.
Mobile applications (Apps):
The generic Application solution offers the citizen the ability to install onto their mobile device, and login using an appropriate identity provider. The citizen then has the ability to either contribute to a directly available survey (public or registered) or a private survey which they have been invited to by the Survey Manager. The citizen will then be presented with the form designed by the Survey Manager, allowing them to participate in data collection with or without network coverage.
In addition to this, the generic capability described above is complemented by functionality allowing the cacheing of high quality basemapping on individual handsets for use in areas of poor or no network coverage. The background mapping used by COBWEB in the UK has been created using a variety of open products (mostly OpenStreetMap and Ordnance Survey Open Data Products) cartographically optimised for zooming in and out on mobile devices in both urban and rural areas. Investigation using Greek data has illustrated the difficulty of replicating this rich open map stack across Europe.
To demonstrate the effective ‘separation of concerns’ in the architecture, and how the COBWEB framework can be used in scenarios where lower level access to inbuilt mobile device functionality is required, a native Application in the flooding thematic area case study area was also developed. This uses the same interface as the generic application for communicating with the ‘Storage Middleware’.
Storage middleware:
Storage Middleware receives the observations from the App. As long as Oauth v2 authorisation is supported, the Storage Middleware component provides a generic REST-based API accessed storage compatibility layer on top of a range of cloud based providers (Google Drive, Dropbox, etc.) or physical storage media where local storage is required.
Storage Middleware is a central component of the COBWEB architecture used for managing survey schemas and exporting geospatial observations to the desired encodings, e.g. KML (Keyhole Markup Language), Geopackage, Shapefile, GeoJSON, CSV, etc. By synchronising all stored information with a relational database (the pre-quality PostGIS database in Figure 2) export of data via OGC Web Services (WMS/WFS) is supported.
Quality assurance and conflation:
Since the emergence of citizen science as a means of supporting scientific research, data quality has been considered an important issue for data use. Quality assurance accordingly plays an important role for the analysis of data obtained from citizen science and for the set-up of citizen science projects. It is commonly agreed that unknown data quality makes citizen science data of limited use. COBWEB’s approach to quality assurance is twofold: first, a number of quality measures are determined based on the variety of data provided by crowdsourcing activities, observations from the co-design projects, sensor feeds and social media; secondly means to express those measures in observation metadata is provided. This also includes an approach to link and conflate observations with relevant external datasets on the Web.
To allow for a customizable and dynamic quality assurance, remaining independent from actual citizen science projects, COBWEB uses a standards based web service chaining approach, and thus enables survey designers to adapt and adjust the quality assurance process to their needs. This generic approach on QA is necessary, because the relevance of specific quality control processes is usually highly use case dependent. The solution extends a pre-existing typology of quality assessment types to seven categories (or pillars) covering a range of specific quality controls generating quality metadata elements.
The implementation of the quality assurance process is based on the Business Process Model and Notation (BPMN) and OGC Web Processing Service (WPS) standards. This combination allows for the definition and dynamic binding of atomic quality control processes, encapsulated by WPS interfaces, to be used in a workflow environment. Therefore, the JBPM suite (workflow editor and workflow engine) has been customized to work with OGC services and has been integrated as a component of the COBWEB portal. Thus, each survey manager has the authority to create quality assurance workflows for their particular survey.
Capabilities to link and conflate citizen science observations with external data are designed and implemented to 1) assist the quality assurance process and 2) infer spatial information from the identified spatial data relations. Since all conflation processes are also offered via the OGC WPS interface, seamless integration and use by the workflow engine is assured. Currently, identified links of an observation can be attached as additional attributes or stored as RDF (Resource Description Framework) triples in a Linked Data store. Whereas the first option is primarily used to enrich and validate observations, the latter option allows for the reasoning on spatial data relations by means of the Semantic Web, in particular by the application of SPARQL (SPARQL Protocol and RDF Query Language) queries.
Sensor networks:
Though data collection from mobile devices is fundamental to the generic COBWEB framework, a variety of sensor platforms, monitoring multiple environmental parameters, within the Dyfi Biosphere Reserve testbed area, have been deployed. These sensor networks have been developed mainly as a result of dialogue between research scientists working on COBWEB and the co-design partners.
There are two ways by which sensor data are incorporated into the storage middleware:
1. Automated sensing. The physical sensors embedded in the environment are augmented with telemetry capabilities that enable them to operate in an autonomous manner, without human intervention, in transmitting data to a base station.
2. Participatory sensing. Data collection by citizen participating in surveys when in close proximity to suitably equipped sensors. The survey participant’s mobile device connects directly to the sensor using short range communication and the citizen physically relays the data back to a position where the data can be uploaded to the COBWEB servers
Physical sensors: Both participatory and automated approaches were researched and developed within COBWEB. Wasp Motes were used for participatory sensing. Surveys are configured such that when a citizen comes within range of one of the Wasp Motes, it is discovered by the citizen’s mobile device. When an observation is made, the mobile communicates with the sensor via the Bluetooth LE standard and the environmental parameters of interest are simultaneously captured. Such data can be used in quality control or to otherwise supplement the metadata associated with that observation.
The In-Situ and Davis Weather station sensors are capturing data automatically and transmitting back to the COBWEB servers. Data from the In-Situ sensors are being used in association with habitat reversion at co-design partner RSPB’s Yns-Hir reserve. Data from the weather stations are being used in educational scenarios with co-design partner Ysgol Bro Hyddgen and to investigate pollinator scenarios with co-design partner Penparcau Community Forum. In the pollinator scenarios, citizens go out on a regular basis to record observations of butterflies and their larval foodplants around Penparcau in the Dyfi Biosphere Reserve. The data recorded includes plant/butterfly species, plant coverage, butterfly activity, and habitat information.
Weather data (temperature, wind speed etc.), from weather stations, sited in strategic locations, is used in a Twitter alert service to send tweets to citizens following a dedicated Twitter account to encourage participants to go out and collect data when the weather is suitable for butterfly activity. The weather data is also used post-data collection to validate and add value to the observations.
When collecting data within the field, sensors on board the user’s mobile device are harnessed to provide additional information. For example, accelerometers values are recorded when the user is taking photographs to enable the line of sight to be determined - useful for various quality assurance purposes. It should be noted that the generic COBWEB App accesses this functionality via a Cordova plug-in while the native flooding App obtains the data directly from the sensor.
Cyber sensing: The harvesting of online geotagged sensor data from a variety of sources through different web-based API’s. In COBWEB, the approach has been demonstrated using data accessed through the Shoothill (UK river levels) and Weather Underground (network of personal weather stations) API’s.
There are two ways in which sensor data are associated with surveys within COBWEB. In the case of participatory sensed data, sensor data are included as part of the survey data and exported to the pre-quality database via the storage middleware in the usual manner. Otherwise, for both cyber and physical sensors, the data are stored directly in the pre-quality database either by sensor nodes transmitting the data in the case of physical sensors or by the harvester in the case of cyber sensors.
Social sensors: Leveraging social media streams. In COBWEB, research and development has concentrated on the use of Twitter and Flickr API’s. Data from both social media platforms are acquired by two methods.
1. Bounding boxes for Biosphere Reserves are used to monitor and capture any social media postings within these areas.
2. Keywords are used to identify relevant postings worldwide.
Once social media postings have been captured they are stored in a database and are available for viewing at the portal or potentially for use within a Quality Assurance workflow.
Standards and Spatial Data Infrastructures:
COBWEB aimed to maximise technical interoperability by the use of standards wherever appropriate - the sections above make multiple references to standards from several standards defining organisations such as, OGC, ISO/TC 211, OASIS and W3C.
All the citizen observatory’s developed under the FP7 call which funded COBWEB were required to make data collected available within GEOSS without restriction.
Publishing data into SDI-like initiatives such as GEOSS was addressed using a cooperative approach based mainly on the processes of the OGC. For example: the project contributed to three OGC led GEOSS Architecture Implementation Pilots; five ad hoc Citizen Science meetings were organised by COBWEB during OGC Technical Committee meetings over the duration of the project, resulting in the formation of a fully constituted Citizen Science Domain Working Group.
One of the main results was the publication of the OGC Discussion Paper ‘Standardized Information Models to Optimize Exchange, Reusability and Comparability of Citizen Science Data’. These are the first steps on the standardisation ladder; broad community uptake of a harmonised common information model applicable to a wide range of citizen science scenarios would be a valuable contribution towards being able to share and future proof the value of data collected. Data compliant with this information model can be discovered and accessed through standardised web interfaces, e.g. OGC web services, and integrated with SDI’s as most, if not all, SDI’s are based on open geospatial interoperability standards.
Two approaches were explored within COBWEB: First, the definition of an application profile by specialising the Observations and Measurements (O&M), SensorML and SweCommon information models. Second, the definition of an O&M, SensorML and SweCommon encodings best practices to allow more efficient reuse of existing components.
Semantic interoperability is a major concern, especially considering the growth in the open data movement and their use of semantic web standards. Semantics is concerned with attaching unambiguous meaning to terms and concepts, typically, using ontologies (formal models describing semantics). In the citizen science domain, as in others, this is very important as understanding exactly what was observed and in which context is essential to making the data comprehensible; both for the immediate use, by others and in the future. The ability to share data, reuse, integrate and compare is often compromised by a lack of semantic interoperability. As would be expected in a research project whose main objective is developing a generic infrastructure that aims to satisfy requirements across as broad a range as possible of the multitude of potential citizen science scenarios, semantics was an area of intense research and development interest in COBWEB. The approach described below is documented in the OGC Discussion Paper referenced above.
In terms of the COBWEB workflow, semantics can be injected at various stages, but the best place to introduce semantics is when the survey context is defined. At this point, the goals of the project should be clear, and if not, then it is time well spent clarifying exactly what is expected to be observed and to what extent exact definitions can be incorporated. Ideally, the Survey Manager would liaise with the Principal Investigator and possibly also the Quality Reviewer and other experts as appropriate, to incorporate sufficient semantics to ensure that the information generated by the survey is of maximum usefulness to end Consumers.
During the Sept 2016 OGC Technical Committee meeting, a live demonstration of the ‘Interoperability Profiles (IP)’ approach developed under COBWEB was given. Using this method, the Survey Manager selects (or creates) an IP during survey design; the IP provides unambiguous references to authoritative definitions of concepts being used in the survey, in this case, it was a link to the relevant standards entry (biodiversity standard Darwin Core terms) defining the invasive species Japanese Knotweed. As a consequence of using this IP during survey creation, every time an observation of a Knotweed was made using this COBWEB survey, the data then contains references to the authoritative definition thereby ensuring semantic interoperability.
Note that this approach is independent of serialisation; the Citizen Science application profile enables encoding as XML (Extensible Markup Language), GeoJSON or linked data. The observations and raw data are of primary interest, various levels of processed or aggregated data may be required at different times depending upon circumstances. All types of observations are supported; including raw observations and those which are derived or result from subsequent quality assurance or conflation processing. A provenance model is included to provide information on the history of each observation.
Of course, widespread adoption and use of this approach presupposes that the relevant semantic resources have been made internet addressable using semantic web standards, that IP’s have been created and that the approach has received widespread community uptake. This is far from the case; however, the low TRL approach described above demonstrably works, can and is being used as a platform to stimulate further discussion, research and development.
Potential Impact:
COBWEB conducted research into the feasibility of creating a common generic framework for mobile device apps for use in citizen science for environmental monitoring, using the UNESCO World Network of Biosphere Reserves as a testbed. We have demonstrated that, in principle, it is possible and brings multiple benefits when starting from a position where the importance of interoperability and flexibility is paramount. This stands in contrast to the current proliferation of Citizen Science Apps generating silos of data.
Specifically, it has been shown that creating a generic solution to automating quality control and assurance, which is sufficiently flexible to address the huge range of potential scenarios, is beneficial to the reuse of citizen science data. Further development would result in the ability to make very large volumes of data useable, and is an area of active ongoing research.
Similarly, COBWEB has contributed to the understanding that further research and development is needed to address whether it is possible to create an easily useable framework which is sufficiently flexible to allow a broad range of different kinds of familiar semantic resources to be employed in designing surveys before citizens go into the field. Without this, despite post-processing server-side, continued problems associated with a lack of semantic interoperability may be anticipated.
Despite perceived complexity and proliferation, the use of open interoperability standards still presents the most realistic chance of preventing the waste of resources and reuse opportunities inherent in creating silos of data locked into proprietary solutions. In the citizen science domain, as in other domains that leverage SDI, the lack of a widely adopted solution to security interoperability is again apparent as a major barrier. At the technical level, COBWEB has shown that access management federations work; progressing this relies upon action at the organisational, legal and political level – outwith the scope of COBWEB.
However, standardisation efforts should continue and adherence be required to help realise investment in SDI type initiatives such as GEOSS.
COBWEB has had significant potential impact in highlighting that, if agreement can be reached on a harmonised common information model, with sufficient community support, most, if not all, crowdsourced, citizen science type data could be made compliant with the developing standardisation efforts initiated under COBWEB. The immediate prize would be a boost to the usefulness of these data by reducing integration costs and enabling the myriad of potential consumers of such data to exploit existing standards based tooling and develop new standards based solutions on top. COBWEB has also contributed to the understanding that the long term curation value of holding citizen sourced data at the observation level of granularity compliant with a well thought out, agreed and documented information model is to maximise potential future access and re-use.
The amount of citizen sourced environmental data will continue to increase dramatically and clearly has the potential to be of use for policy formation and delivery in respect of environmental decision making. At the global scale, towards the end of 2015, the UN adopted the 2030 Agenda for Sustainable Development, this included a set of 17 Sustainable Development Goals (SDG) to end poverty, fight inequality and injustice, and tackle climate change by 2030. The importance of geospatial information in monitoring progress with SDG targets is explicitly recognised in clause 76 of the official resolution (United Nations, 2015). It may be argued that citizen sourced geospatial information has a potentially important contribution towards realising the SDG’s worthy of significant further research.
In this respect, based on experiences in COBWEB, we would recommend increased focus on the UNESCO World Network of Biosphere Reserves (WNBR) as a testbed for sustainability science related research and development. This is particularly appropriate as at the 2016 Fourth World Congress of Biosphere Reserves, the UN’s Man and Biosphere Programme resolutely aligned the WNBR with the SDG’s in the resulting Lima Action Plan .
Using co-design processes and consortia such as those employed in COBWEB, WNBR mechanisms can be leveraged to more easily mobilise sufficient numbers of enthusiastic citizens in order to help realise meaningful outcomes based on real world needs.
The COBWEB framework solution described above is complex and significant further work is required to present only the required level of complexity, depending upon circumstances, when required. Integrating all the components (at different TRL’s) to create an actual system proven in an operational environment (TRL 9) was beyond the scope of this research project.
With the direct involvement of OGCE on COBWEB as a full partner, the use of interoperability standards has always been central to the project and will be key to exploitation of project results. Towards the end of this four year project, in June 2016, COBWEB sponsored the OGC Technical Committee (TC) meeting in Dublin. The project was highly visible and was represented multiple times during the week, including a half day ‘COBWEB Summit’ entirely dedicated to those aspects of the project within the scope of a standards defining organisation.
An ad hoc Citizen Science meeting was also held during the TC week; this was the fourth of five ad hoc meetings organised by COBWEB at TC’s held over the course of the project at various locations around the globe: Barcelona, Sydney, Washington, Dublin and Orlando.
The culmination of this discussion and community consultation came at the Sept 2016 Orlando TC when the charter for the new Citizen Science Domain Working Group (DWG) was voted on and approved, as was the release of the OGC Discussion Paper ‘Standardized Information Models to Optimize Exchange, Reusability and Comparability of Citizen Science Data’ (published March 2017). COBWEB led on all the above and is explicitly acknowledged as doing so, project representatives in Orlando concluded by giving a live proof-of-concept demonstration of the documented approach in action.
At the time of writing, the work described above has already had impact and is a milestone in the standardisation of citizen sourced data. Several recently funded Horizon 2020 projects referenced the work in their proposals and are exploiting COBWEB results; this includes NextGEOSS and all four projects funded under H2020 SC5-17-2015: Demonstrating the concept of 'Citizen Observatories' - the latter was evident at the Brussels cluster meeting attended by the author in Nov 2016. Indications are that the OGC Citizen Science DWG will continue to be an important global forum for discussing common approaches to interoperability challenges in the citizen sourced data domain.
Longer term, the work may lead to the formation of a Standards Working Group and eventually a full international standard with broad uptake. As noted above, the potential impact of this is great; vast amounts of potentially valuable citizen sourced data are currently being locked in silos and opportunities are being lost now and will continue to be in the future because adherence to a well thought out data model is not happening. Whether an accepted standard emerges via the OGC route or not is dependent upon multiple different factors now outside the influence of COBWEB, although the COBWEB coordinator hopes to positively influence this as the current European co-chair of the new Citizen Science DWG.
Various other aspects of COBWEB have also filtered through into new projects. One of the H2020 SC5-17-2015 projects – LandSense, is reusing elements of the COBWEB approach to Quality Assurance (QA) and security.
As noted above, a major criticism of citizen sourced data is that it is often of insufficient quality to be of use for the kinds of scenarios typical of policy formation and delivery. The situation is compounded by the current lack of automated QA methods and bottlenecks caused by reliance upon experts visually checking the quality of submitted observations. Raising the TRL of the highly flexible standards based web services chaining approach, developed under COBWEB, to the point where it could be used in production to automate QA and add appropriate metrics to observation metadata would be valuable indeed.
The situation with the access management federation approach used for security and privacy in COBWEB is different. This is already proven to be high TRL, key components are, in some cases, eg, SAML based access management federations in the academic sector, already in use by literally millions round the globe. The barriers to widespread uptake are not technical; they are political, organisational and financial. Long term impact will depend upon whether other solutions to sharing data across administrative domains using interoperability standards emerge, or whether the lack of a widely accepted approach increases uptake. Over the lifetime of the project, COBWEB used the regular GEOSS Architecture Implementation Pilots to disseminate our knowledge of this area - current indications are that the lack of a widely adopted interoperable security solution continues to be an issue.
Another key dissemination channel used for summarising and communicating project results was the scientific paper ‘Citizen OBservatory WEB (COBWEB): A Generic Infrastructure Platform to Facilitate the Collection of Citizen Science data for Environmental Monitoring’ published in a special Citizen Science edition of the Joint Research Centres (JRC) International Journal of Spatial Data Infrastructures Research. Writing this paper in the final year of the project coincided with the special edition and JRC hosting a two day workshop (data and service infrastructures for Citizen Science) during Jan 2016 of a global grouping centred around the European Citizen Science Association focussed on interoperability of Citizen Science data. JRC subsequently agreed to be charter members of the new OGC Citizen Science DWG.
Other core dissemination activities communicating results towards the end of the project were a COBWEB article intended for a more general audience in the widely distributed Impact magazine and, with greater novelty value, the comic book: Crowd Power: the COBWEB Guide to Citizen Science, available off the COBWEB website . Here you can also find a short documentary video comprised of interviews with key project personnel recorded during the final consortium meeting in Sept 2016 .
From the outset, COBWEB committed to open standards, open data and open source software. In the last quarter of the project, effort was expended in open sourcing those high TRL outputs that the consortium considered to have the greatest exploitation potential. This resulted in the creation of Fieldtrip Open as a separate entity with an existence beyond the end of COBWEB.
Fieldtrip Open is comprised mainly of the generic COBWEB App, the survey designer and the middleware. It is fully documented and extensible allowing the incorporation of quality control processes and additional controls as required. Observations can be published in a variety of different formats/schemas and made available via OGC web services. The intention is that the software can be used with other well known open source geospatial projects, eg, GeoNetwork, that it forms the basis for multiple initiatives and that those benefiting contribute back to the codebase to create a viable sustained open source community.
In March 2017, the prospect of the latter happening received a boost through the Welsh Government’s awarding the year long Welsh Citizen Observatory Pilot (WCOP) contract to a subset of the COBWEB consortium. WCOP was created to sustain the existence of the COBWEB infrastructure as deployed in the Dyfi Biosphere Reserve to allow further opportunities to influence citizen science projects to adopt more mature data collection and management strategies and to demonstrate the benefits of standards based technology and open data models for government. The focus for WCOP will be on projects identified by Welsh Government as having the greatest impact in terms of environmental decision making and policy. Because the software is generic, flexible and does not require software engineering expertise to create surveys, the expectation is that it will also continue to be used by a variety of community groups, including those involved in co-design during the execution of COBWEB.
The Commission’s investment in COBWEB has clearly had an impact, this is particularly evident in how citizen involvement using new technology in environmental governance is viewed in Wales. This may be taken as additional evidence of the benefit of having direct involvement of government bodies in large collaborative research projects that aim to have policy impact. As Welsh Government is a devolved administration, exploitation of COBWEB results by the public sector to realise policy objectives is having impact, and will continue to have impact, not only in Wales, but across the UK through national fora such as the UK Environmental Observation Framework.
List of Websites:
https://cobwebproject.eu/
