Conference paper Open Access
Cameron, Kath; Janusz, Stefan; Pilley, Vanessa; Amin, Farhana; Wilkinson, Steve; Hicks, Jonathan; Gauntlett, Francesca; Kass, Gary; Rose, Mike; Oliver, Sophia; Miles, Emily; Boyd, Ian
Defra (the Department for the Environment, Food and Rural Affairs) and its network of arm’s length bodies – the Environment Agency, Natural England, the Marine Management Organisation, Kew Gardens, the Centre for Environment, Fisheries and Aquatic Science and the Animal and Plant Health Agency – produce large amounts of data. This has been instrumental in providing evidence to support the development and delivery of key policies.
Historically, expertise has been embedded in subjectspecific areas, resulting in data existing in silos with varying degrees of accessibility for those within the department, wider government and wider society. In 2015, Defra’s Secretary of State declared that more than 8000 of Defra’s datasets would be made freely available for anyone to access, use and share. Doing so creates opportunities for everyone – not just those making their living in food, farming and the environment. Opening access to Defra’s data is intended to both provide opportunities to those who wish to use it to exploit its business potential, and to improve policy delivery by engaging a wider community in solving problems.
The Department has been focused on making datasets available as open data; ensuring future data collection and publication approaches are open by default, taking a transparent approach from the outset. This will make it easier to work more collaboratively with other government bodies, external partners and the public.
Alongside growing support for citizen science activities, new technologies are changing the way in which data flows. This presents opportunities for government as much as for others. Today’s smartphones, for example, have the computing power of the supercomputers of 30 years ago. When coupled with a growing range of accurate sensors,
smartphones potentially allow the move away from environmental monitoring depending on sparse, fixed, high-accuracy, expensive monitoring stations, replacing them with mobile, cheap, lower accuracy but more densely populated networks, which could be carried on vehicles such as buses or taxis in addition to being supported by the capability of smartphones.
There is already a move towards crowdsourced data for biodiversity measurement (supported by the National Biodiversity Network). Many people, connected together and using mobile phone technology, could revolutionise country-wide data collection that is sufficient to address policy needs. In future, machine learning may allow species recognition by the smartphone in real time when connected to cloud-based software. This kind of completely integrated software and hardware capability is broadly described by the potential of internet-of things-devices. These have the potential to revolutionise the farm: from making comparisons on productivity, to tracing the journey of food from field to fork.
Defra is also explore the use of real-time data from earth observation and remote sensing to respond to incidents such as floods, invasive species, pollution and epidemics affecting livestock. Opportunities form these types of data sources are often balanced by challenges associated with transforming systems to be more data-driven and to ensure that, in areas that are business-critical, the data sources are assured.
Complex decision-making is already supported by modelling that often requires data flows from multiple sources. Machine learning has the potential to increase the automation and speed of data assimilation in to models For example, this could mean using these technologies to inform policies that reward farmers for acting in ways that maximise positive national outcomes. In this particular instance and more widely, however, this comes at the risk of taking human subjectivity and its political and moral weight out of processes.
In future data volumes will be a major challenge. While compression algorithms may improve it is not going to be possible to store all raw data. This is going to require real-time processing of data to provide intermediate outputs. Judgements will need to be made about what data needs to be kept and how to implement intelligent compression focused on the ultimate use.
The main ethical challenges around data concern personal information, including observance of legislation on data protection and more recent legislation on the right to be forgotten. Defra is working both internally, among its network of agencies, and with its data users to ensure these rights are fully respected. However, data ethics extends beyond just consideration of personal data:
there are sensitivities around some datasets for which Defra shares custody, and others where there are implications for civic, national or cyber security. Some data are also commercially sensitive. There may be additional sensitivities where data concerns reporting of environmental monitoring and enforcing regulation.
Funding reforms have also been a major driver for change. Finding innovative ways to save money, but deliver a high quality service, remains a departmental mission – new data policies can help to transform some of the department’s ways of working. Adopting an ‘open by default’ approach has impacts beyond data, as working openly allows more efficient collaborative working that avoids duplication, saves money, and allows crosspollination of ideas across the Defra group.
Defra is looking more broadly at how to obtain, analyse and manage data (tools, techniques and storage) to further drive innovation. This has been demonstrated by the recently established Earth Observation Centre of Excellence. This centre aims to ensure satellite data is used to its full potential in policy development and operations across Defra by 2020, via a collaborative group of Defra organisations and external partners. A recent success was during the severe weather in December 2015: the centre facilitated better involvement and communication across departments on flood estimates of non-urban areas using radar satellite systems