Chapter 5. Realising digital opportunities in agriculture requires a data infrastructure

This chapter provides an overview of a range of factors conditioning the capacity of the agricultural sector overall (both public and private stakeholders) to embrace the digital transformation and a brief analysis of key policy issues for consideration by the governments. While it is acknowledged that a first constraint to the uptake of digital technologies is access to connectivity infrastructure, this chapter focuses on downstream issues and the use of available technologies. It is not intended to comprehensively deal with all of the relevant issues, but rather to provide an initial overview of some of the key issues of which policy makers need to be aware and to highlight areas where further work is needed.

    

This chapter briefly presents the “data infrastructure” which is at the core of the digital transformation of agriculture and which enables both the supply of new services in the agriculture sector and new forms of policy. The following sections then briefly consider several key issues related to digitalisation in the sector overall:

• access to farm-level agricultural data held by governments (section 5.1); and

• whether there are new roles for government in creating a data infrastructure for agriculture (section 5.2).

5.1. Realising digital opportunities in agriculture requires a data infrastructure

The capacity to create value in the food system or to create better policies using digital technologies depends not only on connectivity infrastructure (hard infrastructure), but also on the regulatory environment and institutional arrangements (soft infrastructure) which together govern access to and use of digital technologies and related data in the agriculture sector. These two elements together shape the creation of effective systems for digitalisation in agriculture, often called the “data infrastructure” or “data ecosystem” (OECD, 2015[1]). The data infrastructure is the system enabling and governing the collection, access and transfer of data (which together are referred to as data governance), as well as storage, and analysis of farm data to produce knowledge and advice (actionable insights) and feedback loops to stakeholders in the agriculture sector, including farmers as well as policy makers (Antle, Capalbo and Houston, 2014[2]). 1

Figure 5.1 sets out this data infrastructure, highlighting the flow of data at different stages, and outlining how data is collected, combined and analysed. In this figure, the data infrastructure is characterised as a chain or cycle of data and information flows. The figure shows key flows in relation to farm production systems; the flows of information for policy is depicted at the edge of the diagram as one of several different data feedback loops. One feature of the data infrastructure is the potential for feedback loops which operate in the complete absence of human intervention, via machine-to-machine flows and automation (referred to as “augmented behaviour” in the figure).

The policy and regulatory environments at each stage of the chain influence not only that stage, but also the ability to connect to the next stage. This influences the extent to which digital tools are available to farmers as well as to other actors in the system, such as governments, researchers and private sector service providers and hence the nature and use of digital infrastructure in the sector overall. For example:

• Digitalisation of farm or government activities is affected by regulations covering access to and use of remote and in-field sensors.

• The access to and transfer of farm data as well as the ability to link it with data from other sources is affected by regulations governing the flow of digital information and interoperability of systems between stakeholders, machines or individuals (data governance).

• Storage of data is affected by regulations influencing the location of data storage.

• Management and analysis of data (big data, models, algorithms, blockchain, etc.) is affected by regulations related to the use and agglomeration of data as well as measures regulating the provision of such services.

The following sections touch upon two elements of concern in relation to the data infrastructure, potentially constraining the uptake of digital technologies in agriculture. First, access to farm-level agricultural data held by governments is discussed. Second, the discussion puts forward potential new roles for the government in the data infrastructure.

Figure 5.1. The data infrastructure for agriculture

5.2. Access to farm-level agricultural data held by governments

Section 4.4.2 discussed options for improving access to agricultural data specifically for policy purposes, including policy-related research. Beyond that, there is rationale for improving access to government-held agricultural data more generally:

• For farmers, so they can better understand the environmental impacts of their decisions and how policies work, as well as learn from government-held information about the agriculture sector more generally;

• For the private sector and researchers, so they can develop and deliver better services for agriculture.

Data from the OECD questionnaire shows how access to agricultural data held by responding public organisations is currently differentiated based on the identity of the person seeking access (Figure 5.2).

Figure 5.2. Accessibility of farm data held by government organisations

Note: N = 47, except for LPIS question, where N=14. LPIS = Land Parcel Identification System.

Source: OECD Questionnaire.

As this figure shows, around 70% of respondents indicated that farmers are able to access their own data held by the organisation relatively freely. Ease of access to farm level data decreases markedly when actors other than farmers are considered. Access for other government organisations most commonly is provided on a restricted basis (55% of respondents); however 24% of respondents reported they do not allow any access to farm level data by other government organisations. Italy (Ministry of agriculture, food, and forestry / Ministero delle politiche agricole, alimentari e forestali) was the only respondent who indicated that other government organisations could freely access farm level data. Access for non-government third parties is even lower, with 45% of respondents not allowing access to such actors, and a further 41% allowing access only on a restricted basis.2

As discussed in section 2.1.3, recent advances in technologies to improve access to and sharing of agricultural data, as well as advances in institutions for data sharing, can help improve access to agriculture data for farmers and for the private sector, while maintaining confidentiality and privacy where needed. Questionnaire respondents were also asked whether their organisation had adopted any innovations to make agri-environmental data more publicly accessible; 34 of 48 organisations have adopted such initiatives. The majority of these initiatives were technical solutions, involving some combination of:

• Increasing the amount of open data available.

• Developing new web applications or portals for viewing or interacting with agri-environmental data.

• Investing in infrastructure which automatically generates agri-environmental data (e.g. new connected weather stations).

• Developing application program interfaces (APIs) to allow for increased interoperability and new ways to use agri-environmental data.

• Publishing data using cloud-based documents (e.g. Google spreadsheets).

5.2.1. Concluding recommendations about agricultural data held by governments

There appear to be opportunities for governments to improve access to agricultural data they hold. As shown in Chapters 2 and 3, there are a variety of solutions which can help improve access to agricultural data, while maintaining appropriate protections (e.g. maintaining data security, protecting privacy, confidentiality, intellectual property, etc.). It is not clear that one particular solution is superior; rather, governments could take a tiered approach, as follows:

• Invest in data services such as providing linked datasets to increase the usefulness of government data collections for policy-making and related research. One important aspect of this to consider is how, and when, to link farm financial datasets with physical data such as soils, precipitation, and other climate variables.

• Increase use of secure remote access mechanisms to reduce transaction costs of allowing trusted actors (e.g. policy researchers) to access agricultural micro data held by governments.

• Explore how new data sharing technologies such as confidential computing could avoid the traditional confidentiality-accessibility dilemma.

• Take a risk-based approach towards access to agricultural data held by governments: consider and clearly articulate reasons why specific data or classes of data cannot be openly provided. This could be accompanied with commitments to periodically review pre-existing legislative requirements to protect confidentiality of agricultural data.3

Government organisations which collect or store agricultural data could work together with data providers and data users to establish clear frameworks governing data access and use. It is important to emphasise that such frameworks should be coherent with broader policies governing such issues, as well as with underlying legislation authorising government agencies to collect agricultural data.

In seeking to improve publicly-held agricultural datasets, data-collection agencies can explore how the burden of existing data collection by government organisations can be lessened while maintaining or strengthening data collection through the use of digital technologies, including considering how digital tools could be used to gather data via alternative pathways. Data management frameworks could also support the evaluation of data quality for data from alternative sources and planning. Finally, government organisations have a role in ensuring the longevity and robustness of these data sources.

Governments should also explore ways to incentivise provision of private sector data for public use and for agricultural research. This should include consideration of providing incentives for farmers to allow their data to be shared for policy purposes; options include monetary incentives (i.e. payments for data provision) and non-monetary incentives, such as provision of regulatory safe-harbours for data providers or provision of services which use data that has been provided (e.g. benchmarking services).

More broadly, while further work is needed to evaluate existing regulatory and governance frameworks, there seems to be a role for governments to help stakeholders clearly understand different available governance arrangements and to provide clearly articulated underpinning regulatory frameworks that other users can build on.

5.3. Are there new roles for the government in the data infrastructure?

According to the type of public services required, and the institutional environment and initial conditions, enabling development of a data infrastructure might require different types of actions and roles for the government, whether as co-ordinator, as a regulator setting interoperability standards or to directly develop the data infrastructure and create markets for usage rights. The role of the government is likely to change according to how advanced those networks are, and whether the service provided can be marketable.

Provision of physical infrastructure (e.g. connectivity infrastructure, sensor networks, physical elements of tracking and traceability systems, etc.) faces traditional issues for infrastructure in network industries, particularly the question of where the role of government stops and that of the private sector starts. There might be cases requiring broader government support for the financing of network infrastructure, including in less economically important areas (areas not cultivated, of low productivity, but nevertheless important from an equity perspective or to be able to have a holistic approach to data acquisition). In particular, the creation of a network of sensors and of information needed to monitor the environment in ways that allow the provision of public services such as drought early warning systems, and to inform water policy and management, requires coverage of all geographic areas, whether cultivated or not (see example of soil moisture in Case Study 9). This could suggest a role for the government as it might not be economically viable for the private sector to develop infrastructure in some areas, which are nevertheless important for the understanding of ecosystems dynamics and forecasting.

In addition, questions about the sharing of data according to the definition and value (economic and social) provided to the different use of data produced by private systems remains an issue. Discussions at the OECD Global Forum for Agriculture in May 2018 highlighted a range of views in relation to data ownership, privacy and the types of information and derived conclusions that can be left with the private sector and those which need to be managed (governed) by public authorities. For instance, consider wireless sensors networks (WSN), which can provide data of public interest, but which could also underpin development of decision support systems which could be sold to farmers. Such WSN could produce a lot of information, especially in high density farming areas, to which services could be added to make investment profitable. However, there might still be constraints to the sharing of the data. Therefore, there may be a role for governments to develop at least the basic WSN and allow for the private sector to build on this and develop marketable services. In addition, the quality and veracity of data obtained via private application of new digital technologies to support policy-making would need to be ensured.

Three case studies presenting different elements of data infrastructures currently being developed for agriculture were explored in order to further identify some potential roles of governments, the constraints they faced and how they dealt with them. The case studies are:

• Case Study 8: Estonia e-government and the creation of a comprehensive data infrastructure for public services and agriculture policy implementation.

• Case Study 9: Connecting the dots to create a data infrastructure: the US National Soil Moisture Network.

• Case Study 10: Data infrastructure and the potential role of the government supporting the data infrastructure: the example of the Akkerweb in the Netherlands.

Some common lessons drawn from these case studies are presented in this section.

5.3.1. Data quality and trusted algorithms

One first element identified in which governments have a role is that the performance of the data infrastructure to support decision-making depends on the quality of data and the trust in algorithms.

Without good quality data, even the most refined algorithm will not be able to provide good information. For example, big data is the capacity to aggregate a large amount of data, but big data only makes sense if it can be used to produce quality analysis. Other new digital technologies, such as blockchain or artificial intelligence (algorithms), are sophisticated programmes, the value of which also depends on the quality of the data they use. Moreover, if bad quality data is used in automation, it can potentially have important negative consequences. However, quality data and “fit for use” data can be expensive to produce.

Governments can play a role in ensuring that good quality data is used in algorithms and artificial intelligence:

• Governments can use a range of measures to improve access to farm-level data held by government agencies, particularly in relation to access for policy and research purposes (Section 5.1).

• Governments can encourage good data management practices by participating in or leading development of high quality metadata standards.

• Governments can consider the merits of shift towards an “open first” approach to allowing access to data held by government, in which data is encouraged to be open or re-useable as a default, rather than inaccessible by default. This openness can enable users to identify and notify problems with data and serve a quality control function, as well as to help ensure the best available data is used in algorithms and artificial intelligence.

Box 5.1. Case Study 8: Estonia e-government and the creation of a comprehensive data infrastructure for public services and agriculture policy implementation

This case study illustrates how digital technologies can be used to improve the administration of government systems and the provision of public services, including in relation to agriculture, using the example of e-Estonia, an initiative by the Estonian government to facilitate citizen interactions with the state through the use of electronic solutions.

The development of the Estonian e-Government is based on the Principles of the Estonian Information Policy, adopted by the Estonian Parliament in 1998. Through this, the government initiated a digital transformation to increase efficiency of its processes as well as how efficiently it delivers public services. The Estonian government made two critical technology choices: a compulsory digital identity (ID-card, proof of concept and ecosystem built in 2004) allowing the real world to match the digital. The second choice was to develop the X-Road, the data management infrastructure based on an innovative decentralised linked government data infrastructure preventing data redundancy and using Blockchain technology to create transparency about data access.

Among a range of applications, this infrastructure is used for agriculture policy and regulations. The Estonian national paying agency has been using satellite imaging and remote sensing since 2005 and controls of mowing requirements under the EU CAP, attached to financial support from the European Commission, have been increasingly automatised from 2011. With remote sensing and automation of processes, the percentage of checks has gone from 5% on site to almost 100% performed remotely.

A range of digital services is also now available to farmers, including digital registers. For instance, whereas information was previously recoded using a paper-based system, farmers are now able to provide information via an e-register. As of August 2018, 64% of documents and 89% of notifiable animal events were submitted using the e-services register.

Finally, the Ministry of Rural Affairs has initiated a feasibility study for development of an agricultural big data system. The aim is to create a central electronic system to link and integrate existing data with analytical models and practical applications. Data linked in this system must be harmonised, compatible, updated, linked to spatial data, and transferable from the producer to the system and from the system to the producer, enabling access to potential models and applications. The system will provide useful practical information flow for the farm management decisions (e.g. machine-readable data for the precision farming machinery). The system will also enable to collect more precise farm data with less effort. This improves the quality of statistical data and enables more comprehensive analyses. This one-year duration project started in September 2018.

Source: Case Study 8, Part IV.

But beyond the data itself, it can also be important to ensure the quality of algorithms used to process it (which will also affect the quality of inputs downstream). Governments can also play a role in ensuring that algorithms are able to be appropriately scrutinised, while also recognising intellectual property or commercially sensitive elements relating to algorithm design:

• Governments can provide a model of good practice for responsible and transparent use of algorithms as a tool for public analysis and decision-making (section 4.3.3).

• Governments can build farmers’ confidence in using algorithms as aids for decision-making by ensuring that algorithm designers and providers of algorithm-based AI services are subject to standard conflict of interest regulation (including declaration requirements) and that farmers are aware of these obligations. For example, if a seed or fertiliser company designs algorithms to provide planting or fertiliser application maps based on precision agriculture machinery, there is a risk that the algorithm could be designed to maximise profits to the company rather than benefits to the farmer. This underscores both the need for regulation to prevent such practices, as well as for educating the agriculture sector about relevant regulatory frameworks, and available recourse.

It is worth noting that the issue of whether farmers have confidence in use of algorithms to support policy (discussed in section 4.4.1) is similar to the issue of whether farmers have confidence in use of algorithms by service providers (e.g. farm advisory services provided by the private sector). There may be a need to invest in developing farmers’ understanding of how algorithms are used (in a general sense): otherwise technologies may appear to be a “black box” and farmers may oppose policy recommendations or may not act on recommendations due to a lack of confidence or trust (for example, in a context where an agri-environmental programme uses an algorithm to develop on-farm conservation recommendations; or where a service provider’s recommendations are based on an algorithm).

A second element to take into account when identifying the role of the government is that relevant regulations affecting the quality of the data infrastructure may not concern the sector itself but may relate to other sectors that produce intermediate goods and services for the agriculture sector or which buy from the agriculture sector. That is, effective policy-making for digitalisation in agriculture may require going beyond the agriculture sector. In addition to core connectivity infrastructure, the functioning of the data infrastructure requires access to goods (sensors) and services (connectivity providers, as well as business services producing actionable insights sold back to the farmer). A combination of policies and regulations beyond the agriculture sector (e.g. goods and services trade policies), as well as innovation policy more generally, can therefore influence the business strategy of actors in the data infrastructure.

5.3.2. From regulatory oversight to acting as an investor and co-ordinator

More broadly, there could be a role for the government to support the development of infrastructure for the datafication of agriculture, from regulatory oversight to acting as an investor and co-ordinator when there is a collective gain but few private incentives.

This can be particularly important in case of infrastructure in network industries, as illustrated by the Case Study 9.The opportunity costs for policy management from the lack of coordination of soil moisture data across the United States triggered an effort to promote their better integration, under the National Soil Moisture Network (NSMN) initiative. But in addition to coordination (see McNutt Verdin and Darby, (2013[3])), the initiative recommended early on that the increase in the number of monitoring sites would be the most important improvement in the overall depiction of soil moisture. Drought risk and water flows do not finish at regional borders, nor are they only an issue at the level of agriculture lands, nor only for highly productive areas: water management, policies and drought risk require a comprehensive understanding of soil water dynamics at a high resolutions across potentially large and highly varied landscapes. This suggests that there may be a role for the government as it might not be economically viable for the private sector to develop infrastructure in some areas, which are nevertheless important for the understanding of ecosystems dynamics and forecasting.

Box 5.2. Case Study 9: Connecting the dots to create a data infrastructure: The US National Soil Moisture Network (NSMN)

Two types of technologies are used for the monitoring of soil water content in the United States: direct in situ instruments and remote sensing. Each approach has strengths and weaknesses. Remote sensing has the advantage of allowing contiguous data coverage across the United States and progress in its precision has resulted in increasing use for agriculture services and policy implementation. However, data provided is still at a relatively coarse level of resolution. In-situ measurements group diverse types of networks. Some, such as wireless sensors networks (WSN), provide data at the farm level and can be integrated into decision systems for precision agriculture or water management. However, these are often private and systems are proprietary and focus on the farm level. In addition, the data belongs to either the farmers or the company providing the service and is therefore not easily accessible by other stakeholders, including researchers and the government.

Most data used by researchers is still mostly at the 30 km scale. These mesoscale networks, also called mesonet, have principally resulted from initiatives at the State level. As a consequence, they are distributed unevenly across the United States, with some geographic areas more densely covered than others. In addition, they are not always publicly accessible and some are protected by paywalls. While the mesonet is very useful for some applications, understanding a range of natural phenomenon requires broader coverage. In addition, understanding the dynamics of soil moisture in ways that can be useful for policy management and decision making requires more information than soil moisture data point estimates. Needed information—such as soil characteristics, composition across multiple soil depths, weather patterns, and land use information—is available but in disparate data networks and from different sources.

While a large amount of data exists and could support researchers and policy makers, it is not used to its full potential. This is due to a lack of technical capacity (data processing and management) but also to the independent and non-coordinated development of networks across the United States. The production of an accurate representation of soil moisture at an informative scale has therefore remained a challenge, and soil moisture observations have been poorly integrated into assessments of vulnerability, such as early warning systems for droughts and floods.

In 2013, the realisation by the policy and research community of the need to improve metadata and calibration and validation of soil moisture data as well as data integration resulted in the development of a Coordinated National Soil Moisture Network (NSMN) The objective is to develop a high-resolution gridded soil moisture resource, accessible to the public through a web portal. The project brought together in situ measurements of soil moisture from the federal networks, in combination with a range of other databases, including the NRCS SSURGO, which provide a unique gridded database of soil properties and satellite (PRISM) data. Challenges highlighted in the feasibility study included data transfer protocols, storage, and data gaps from intermittent connectivity to stations.

Source: Case Study 9, Part IV.

5.3.3. Governments might need to rethink the way they are operating, as well as their role as a provider of public services.

Implementing a new data infrastructure policy requires awareness of a range of issues, from how to create interoperability between agencies and between database and stakeholders, to how to ensure the protection of government data and who has a right to access it. Estonia dealt with such problems using a decentralised system and cryptography.

In addition to these technical questions, governments should also think about the use of the envisaged tool. In particular, the data infrastructure can potentially smooth communication between all stakeholders and can be thought of as serving not only policy makers and administrators, but also farmers. Moreover, while one of the roles of the government may be to gather relevant information for policy implementation, governments should take a multi-functional approach to its data collection and management, considering the merits of also including in their databases information not directly of use for policy-making, but which could be useful for farmers when combined with government data. Both the Estonia and the Akkerweb case studies involve systems that allow private sector access to government. It is envisaged that in Estonia the system could be based on an agreement with farmers by data type. The data infrastructure created by Estonia, clearly identifies who the data has been registered by or referred to through the eID-card.

The second issue highlighted by the NSMN case study is the need for interoperability standards. Any network or platform, whether publicly or privately administered, is developed to answer specific questions, or achieve certain purposes. Therefore, they often adopt different approaches to data creation, management and codification. As a consequence, the data produced might not be “fit for purpose” and could create biases if used in modelling and analytics that depart from the initial goals. While there can be collective gains to coordination, there might not necessarily be private incentives. In such cases, networks lend themselves to some form of regulatory oversight or a central planner.4

With the creation of the NSMN, an important need was for the data produced to be usable for a diversity of objectives and by a diversity of end users. The first step of the NSMN was the co-ordination of existing networks, bringing together current entities in a common format. As such, the NSMN also acted as a standard setter; effectively leveraging the full variety of existing networks and modelling efforts relied on consistent calibration and validation practices and metadata characterisation.

Box 5.3. Case Study 10: Data infrastructure and the potential role of the government supporting the data infrastructure—Example of the Akkerweb in the Netherlands

This case study provides a practical example of how an open data infrastructure can facilitate the creation and uptake of value adding services by the private sector, supporting productivity and sustainability in agriculture, using the example of the Akkerweb digital platform and data repository. Akkerweb is a foundation, founded by both Wageningen University and Research (WUR) and a farmer association, Agrifirm. Scientific knowledge and a practical approach to farmers’ problems are combined to develop successful applications. Some data and applications are made available by the WUR research team, others are added by the private sector.

In the Netherlands, a plethora of unrelated systems have been accumulating data about on-farm activities, farm performance (e.g. yield variation) and the characteristics of production assets, resulting in a fragmentation of data. In addition, while a large amount of data is being used and acquired, most is not actionable, meaning that it cannot be directly used (or re-used) for further production of information feeding into decision processes (analytics). Akkerweb is a digital repository and work bench upon which applications, ranging from data visualisation to analytics and decision support, can be built by both the public and the private sector.

Farmers can access a free account and add information that is securely managed on the platform. The platform provides a variety of agriculture related applications readily usable by farmers, using their data, and providing support to decision making to optimise production objectives. In Akkerweb, the farmer can combine his or her farm specific data with data from public sources (satellites, soil maps, weather data, parcel maps from the Netherlands Enterprise Agency (RVO) etc.) with proprietary data sources such as sampling bodies, parties in the chain, farm management systems, own sensors etc. In particular, WUR currently provides free satellite data already translated, using complex computation, into in vegetation indices (indication of the amount of vegetation, distinguishing between soil and vegetation etc.). This data is then combined with other commercial data (for example drone data) for a range of advisory services.

Farmers can also access government data. For instance, active links are available with the data store of the national Paying Agency (RVO) and with other farm management systems, to prevent double entry of data. Only the farmer has access to their own data but they can grant access to others at their discretion, making it a type of “controlled access” data governance. In this way, they can give access to their advisors to help them monitor the crops or interpret a soil analysis. Farmers are therefore free to share enriched data with advisers and other users on the platform, to obtain practical recommendations to optimise crop production. The system itself provides interoperability of data. Any data provider can link their data (e.g. soil laboratories) and make them available to farmers.

Source: Case Study 10, Part IV.

5.3.4. Path dependency, infrastructure and regulatory environment: governments have to be aware of their starting point

Finally, policy makers have to consider a degree of path dependency in policy-making and infrastructure development. In the case of the NSMN, the devolution of investment decisions to sub-national scales led to a lack of coordination and alignment of objectives that created inefficiencies in terms of data creation and management. In this context, the NSMN initiative acted as a catalyst, bringing together institutions and creating awareness about the specificity of soil moisture monitoring. But instead of recreating a new infrastructure, they decided to reuse previous, still relevant ones, but created a push for further investments in maintaining and developing it.

This approach contrasts with the Estonian case, which is very different in its intent, timeframe and scale. The government data infrastructure in Estonia was a long-term plan to build a holistic government data infrastructure. Although implementation began in 1998, the system is flexible enough to incorporate new technologies (use of the blockchain) and to add new functionalities. Estonia has a relatively small population, and while it is true that this possibly made the transition and communication about the initiative easier, this should not understate the success of this government administration make-over. While not without problems, the change has proved reliable and flexible.

These case studies demonstrate that there is not a one-size-fits-all solution for the creation of a data infrastructure. Countries need to balance opportunities for coordination and reuse of existing infrastructure with a level of flexibility and potentially changes to the role of the government.

Effective communication and collaboration is an important part of the implementation and adoption of the data infrastructure. The data infrastructure should provide the right incentives with flexibility to implement change and avoid barriers to adoption. Collaboration and communication will be needed both within the government and between the government and citizens.

In Estonia, the regulatory environment has been used to set the incentives for the implementation and use of e-government by government agencies, by centralising policy development, and allowing the Ministry of Economic Affairs and Communication to develop the principles of information policies and supportive legislation, and take over responsibility for supervision of relevant state organisations. Subsequent implementation was decentralised, with e-Government developments done mainly by responsible ministries and state agencies. Accordingly, every government department, ministry or business, gets to choose its own technology, based on commonly agreed principles.

It appears that in the case of Estonia, there have been few barriers to adoption, whether from the institutional side or from that of the users. Various factors account for this, including the population size that helped make implementation more straightforward and communication about initiatives more efficient. On the agriculture side, the fact that data provided by farmers is used to provide support, and not (as in other countries) to verify that the farmers are complying with regulations, had an important role in the level of adoption, as were the services provided digitally by government bodies.

Collaboration requires trust, and the adoption of a data infrastructure requires creating a regulatory environment guaranteeing such trust in the new system based on transparency. A clear regulatory environment about the use and protection of data is reassuring for stakeholders. For example, data security is considered to be the most important feature allowing the Estonian digital society to function. Anyone with a social security code can look up their information online, and thanks to the blockchain technology, they can see who has accessed their data and when. It is also possible to ask about any single query, which allows for a higher transparency in the services. Some core principles, adopted by the Estonian parliament as early as in 1998 and reviewed and updated in 2006 in the course of preparing the Estonian Information Society Strategy 2013, have been driving the development of the Estonian e-government, some backed by legislation (see Part IV for details of relevant legislation).

One core principle is that while the public sector has a role in leading the way towards the development of what is more broadly referred to as the information society, developments require co-operation between the public and private sectors, and perhaps with the public more broadly. Therefore, and in order to reassure the Estonian society about the use of their data, a range of legislation has been passed to ensure the protection of fundamental freedoms and rights, personal data and identity. In particular, individuals are the controllers of their personal data and they have an opportunity to decide how their personal data are used.

5.3.5. Governments should ensure there is co-operation and communication between stakeholders

Co-operation and communication is also needed to ensure relevance, uptake as well as prevent unintended consequences on all stakeholders. To be successful, digital technologies have to be designed based on expressed user needs and create positive outcomes from use for all stakeholders.

In order to support the shift from paper to digital, the government of Estonia supported different advertising campaigns to communicate advantages to farmers, including a more rapid identification and treatment of errors. This provided a positive outcome to the digitalisation of farmers’ information. Advisory services are free for farmers, who benefitted from a smooth transition and lesser administrative burdens and a shorter time to rectify errors. In addition, as administrative processes are managed faster, payments are more rapidly transferred to farmers. The Estonian LPIS (land parcel identification system) and animal data are also used by statistical offices and for the cadastre system as well as by the environment agency, allowing for cross checks with different agencies. Benefitting directly from positive outcomes facilitates the transition to digital government services.

Another example is the NSMN which created awareness among stakeholders about first, the importance of soil moisture data not only for researchers but also for policy makers and second, about what would be possible with cooperation enabling data already produced to be better exploited simply by cutting across administrative borders. In particular, both the public and private sectors – farmers, policy makers and the community – would benefit from better preparation and resilience to drought.

Finally, the Akkerweb provides the example of functionality design that is based on expressed user needs. The platform also partnered with a private sector firm to develop an application for visualisation and analysis of satellite- and drone data. Farmers get access not only to vegetation indices, but to maps, for example scouting maps and task maps.

The success of a range of digital technologies relies on the integration of all stakeholders, and collaboration between the public and private sectors for the creation of information has to be fostered. For example, one success of Akkerweb is the strong connection with stakeholders from government, research, agriculture and the ICT sector, providing both scientific backstopping to models and algorithms, and a practical approach to functionality. Public bodies are participating in the data repository construction by linking their agriculture policy data to the platform and supporting the pre-processing of satellite data.

A final element of importance in which government has a role is to provide a regulatory environment enabling to create transparency and tackle the issue of data quality, which leads the way to making the best out of big data and ensuring trust in data-enabled decision making.

In Estonia, a new project is investigating the possibility of bringing big data to farmers. A feasibility study is currently assessing the needs and roles of stakeholders, data storage systems and evaluation of existing data quality. The concept for the big data system will include the technical, legal and economic analyses and the roadmap for implementation. The project includes training for farmers to explain the potential of big data for farm management decisions, to introduce practical applications and models and to demonstrate the technologies for precision farming. The next phase will be the implementation of the system based on the results of the feasibility study.

References

[2] Antle, J., S. Capalbo and L. Houston (2014), “Towards a Knowledge Infrastructure for Science-Based Policy and Sustainable Management of Agricultural Landscapes”, http://www.foodandagpolicy.com.

[3] McNutt, C., J. Verdin and L. Darby (2013), Developing a Coordinated National Soil Moisture Network: Findings from a meeting in Kansas City, Missouri, Nov.13-14, 2013, https://www.drought.gov/drought/documents/developing-coordinated-national-soil-moisture-network (accessed on 21 September 2018).

[1] OECD (2015), Data-Driven Innovation: Big Data for Growth and Well-Being, OECD Publishing, Paris, https://doi.org/10.1787/9789264229358-en.

[4] OECD (2008), Broadband Growth and Policies in OECD Countries, OECD Publishing, Paris, https://doi.org/10.1787/9789264046764-en.

Notes

← 1. The OECD’s Directorate for Science, Technology and Innovation has an ongoing work programme examining broadband developments and related policies. This work “highlights challenges such as connecting users to fibre-based networks or coverage of rural areas” (OECD, 2008[4]). Further information is available at http://www.oecd.org/sti/ieconomy/.

← 2. EU member respondents were additionally asked about access to farm level data held in EU Member country Land Parcel Identification Systems (LPIS). These respondents generally indicated that a farmer or landowner can freely access and download their own LPIS data at any time (71% of EU member respondents); however, a number of countries only allow access on a restricted basis.

← 3. Note that this recommendation does not presume that an open data approach will be appropriate in all cases. Rather, it is recommended that governments consider the possibility of opening datasets as a useful conceptual starting point so that the case for confidentiality requirements can be appropriately (re-)evaluated and transparently made.

← 4. World Meteorological Organization (WMO) is in the process of developing standards for soil moisture network development.

Chapter 5. Realising digital opportunities in agriculture requires a data infrastructure