Bryan Nanson, Director of Information Management, Neil Smith, Chief Science Advisor, Alison Davey, NGD Programme, Ordnance Survey.
Abstract
Recently there has been a growing appreciation world-wide of the importance of bringing together many existing geospatial databases. The prize from such data linkage is better quality data through identification of data inconsistencies and the generation of added value. Examples of this trend can be seen in many countries though the most well-known example is the Clinton Executive Order mobilising the National Spatial Data Infrastructure (NSDI) in the USA.
In Britain, the indications are that there is a wide range of geospatial data in existence and that much of this is of a higher quality than in certain other countries. Actions to facilitate data sharing in this country (such as the launch of the SINES metadatabase) have already taken place and even pre-date the NSDI. But there has been no articulation of what a National Geospatial Database (NGD) would consist of in Britain. The successful prototyping of the National Land Information Service and the Domesday Project are small, though significant, steps on the way to a NGD. Moreover, there seems to be a multiplicity of terms in common use which are used inconsistently.
This paper sets out a view of the nature of the British National Geospatial Database and how existing databases relate to it now and might do so in the future. It considers options for bringing the NGD into a working state, including use of data standards and the commercial and social aspects. The objective of the paper is to stimulate and advance debate on this important topic.
In order to overcome potential confusion over terminology with alternate use of terms we propose that the term 'geospatial' is any data that has associated with it some geographical reference. Examples include any data labelled with a postal address, a National Grid reference, a postcode, a standard area name (such as a county or parliamentary constituency), latitude and longitude or Global Positioning System co-ordinates. In mainland Europe, the term 'geographic information' (GI) is commonly used to mean the same thing. In North America the term 'spatial data' is commonly used. But in Europe, particularly France, the term 'spatial' has some connotations of extra-terrestrial activities. Consequently the term 'geospatial' has been coined in Britain to describe the same data to retain a link with location on or close to the earth, yet stressing the plurality of disciplines (not just geography) that are users and providers of such data.
Throughout the world there is an ever-increasing awareness of the potential from utilising advances in technology to bring together the many existing sets of data which contain a geospatial reference. It has been estimated that in the UK between 60% and 80% of all data held by government departments can be classified as geospatial. Over 500 major government datasets of this type (including, for example, population census data, school exam results data, much environmental data) are already known and many others are likely to exist. At present the great bulk of this rich resource is held in separate files in many different locations on different systems, often poorly documented and unknown to most potential users of it. Typically the only common property between different datasets is that they vary geographically - that is, over space. However the geospatial references or descriptions that are used at present may or may not be mutually compatible: this limits integration and hence the creation of added value. They may differ, for example in the referencing system (e.g. EUREF and Health Authority areas), the resolution of the reference (e.g. co-ordinate resolution of metres or kilometres) or the identifiers (the same thing referred to in different ways). The cost of collecting these data is not readily available but for the UK is clearly substantial. As an illustration, the annual investment in the USA to obtain the geospatial data required by Government has been estimated by the Federal GeoData Committee to be over $2 billion.
Examples of the use by the private sector and government of integrated geospatial data are set out below. New possibilities are being discovered continually - for example the UK Government's Technology Foresight exercise[1] has recently identified that to improve the UK retailing and distribution business as a whole there is a need to study the relationships between land use, demography and transport information.
Application Data
Setting insurance Geology, crime patterns, climate, topography,
premiums on a small area hydrology
basis
Planning applications Geology, topography, hydrology, land
registration
Location of shops, Land use, transport networks and
offices, distribution characteristics, demography, perhaps geology,
plant pedology
Social Security bad Weather data, climatic data, demographic
weather payments profiles
The most dramatic example of the effective use of such integration was perhaps the database built by the task force led by the United States Geological Survey following the 1993 flood disaster in the Upper Mississippi river basin. This brought together data from several government agencies and played a leading part in the management of the situation. Indeed it lead directly to the Clinton Executive Order mobilising the National Spatial Data Infrastructure. This infrastructure is based on:
However, the totality of this initiative is not yet universally recognised at the State or local levels. Major efforts are being made to build partnerships between Federal, State and local governments to expedite the NSDI through the USA. In some cases the lower tiers of government have different agendas in regard to their geospatial data. In some 19 States, for example, the information access laws have been modified to permit charging for geospatial data with the express aim that the States can recover from sales or leasing of data some of the costs of collection and maintenance.
The US vision of the NSDI could probably not be made to work in Britain because:
But:
It is therefore obvious that though simple cloning of developments elsewhere are unlikely to work in the United Kingdom there is a good national basis on which to proceed. This requires us to explore what a vision of the National Geospatial Database could - and should - be.
The National Geospatial Database is not seen as a single, all-embracing collection of data sets held in one massive computer system. Neither is it the property of or under the control of any one organisation. Attempts to obtain control of it will inevitably fail given the nature and structure of British government bodies at all levels - the primary source of geospatial data in Britain.
The National Geospatial Database is seen as being a 'virtual database'. It will comprise the totality of many individual geographical data sets collected and held separately by many different organisations. Any one data set will qualify to be part of the NGD provided that the following conditions exist:
Figure 1 shows our concept of the NGD in diagrammatic form, with standards forming the deep foundations and applications sitting on top of all other data items.
Concepts and initiatives such as we propose are not new. The National Land Information Service pilot scheme is one example of how organisations with disparate databases can link their data together to add value and provide information for a specific market - those involved in conveyancing and land and property management. Investment in the development of various standards such as BS 7567 (National Transfer Format) and BS 7666 Parts 1 to 3 (with 4 in work) provide definition of national systems for gazetteers, contributing significantly to the process of achieving common agreement about important issues in the area of geospatial information handling. There are many players - Ordnance Survey, HM Land Registry (HMLR), the Local Government Management Board (LGMB), the Inter-departmental Group on Geographical Information (IGGI) among others.
This paper looks beyond the present situation. It defines the concept of a National Geospatial Database, the existence of which would facilitate wider use of many existing datasets and the gradual harmonisation of their characteristics. The existing initiatives in standardisation and harmonisation can for the most part be readily fitted into the concept. If the National Geospatial Database is achieved, then maximum value can be obtained from the substantial investment already made by the taxpayer and other parties.
The National Geospatial Database will
and
The concept of the NGD as a virtual database is only sustainable if certain standards exist, are accepted by participants and followed in all significant respects. It is the only way in which such a loose structure can be made to work.
The relevant standards are those that govern processes and procedures and also standard datasets that provide a definitive description of particular phenomena.
Process and procedure standards will include:
Many of the above are already in place or are being devised at the British, European or international levels. Work will be needed to implement them and ensure that they are appropriate. For this reason OS, among other British government organisations, is active in participating in the development of various standards through the British Standards Institute, Comité Européen de Normalisation (CEN) and the International Standards Organisation (ISO). These initiatives are facilitated by the Association for Geographic Information (AGI). An important but purely British initiative exists in the concept of the Standard Geographic Base of UK proposed by IGGI which concentrates in particular on the definition of hierarchies between geospatial objects at different levels. The results of all these exercises will need to be subsumed in the operating standards of the National Geospatial Database.
Clearly a fundamental requirement is for knowledge of what data sets exist, who they are held by and whether and how they available. SINES - the service run on behalf of IGGI by OS - provides a first step in this. It describes some 500 government held geospatial datasets and allows them to be selected on the basis of a keyword, location and organisation. SINES at present can simply give information that provides a high level overview of a dataset with a contact address for further information. It is not yet a 'one stop shop' for data that would allow both selection and downloading from a single transaction. Nonetheless, SINES is progressing. When introduced in 1994 it was simply a 'phone-in' service. Since then it has allowed email contact and, from October 1995, a service is available over the World Wide Web.
In the future, more detailed information will be needed within SINES and its successors to support the applications that the NGD will allow. Quality information, increasingly quantitative, will allow rational selection of datasets with links to transfer - and invoicing - systems. The contents of the SINES metadatabase will need to be expanded and updated more frequently. It is essential that the 'one stop shop' happens if our vision of the National Geospatial Database is to succeed.
No attempt is made at this stage to define in detail the services that will be supported from within the National Geospatial Database. These are likely to be varied, ranging from direct on-line access by an end user to mediated services through brokers and expert interpreters. Some of the latter may also be data providers, for example for geological data.
Issues that need to be addressed to ensure maximum accessibility of the data within the National Geospatial Database include:
Different data sets become more valuable when they can be used in combination: with 2 data sets for the same area, one combination exists but with 20 data sets, over 1 million combinations exist and many more applications are possible. Since 'the geography' of each data set may be defined in different ways (e.g. postcodes or co-ordinates or references to other objects), tools to link the data sets must exist. Ordnance Survey provides the data tools for some linkages through its ADDRESS-POINTTM, OSCAR® and Land-Line® products which are all derived largely from the same and hence consistent ultimate source. Moreover many key datasets (such as property boundaries, geological and statistical areas) have been compiled on the same basis and are therefore compatible. Service vendors provide the software tools to use these data tools to link other data sets together. Clearly all these tools will need to be brought together by service providers to ensure that the underlying technology is transparent to the (non-expert) user.
Because data sets can be linked together and used in combination does not mean that the results will be always be meaningful for every task. Often some considerable understanding of the specialist characteristics of the data is required to use it safely. Two examples are given below:
For this reason, the NGD will also identify areas where specialist consultancy might be required in order to guide the unwary. It will be necessary or desirable to permit 'advertising' of these services for professionally accredited service providers.
It is the Ordnance Survey view that the National Geospatial Database will be created gradually by the process of data providers entering into partnerships - joining the club - when the time is appropriate for them to do so. This will be facilitated by the creation of the overriding data architecture described above and a technical delivery architecture to integrate datasets and to deliver the applications. It can be aided by Ministerial encouragement and any move towards a more corporate policy for British government information.
We are not advocating a 'big bang' approach with the consequential requirement for massive funding. Instead - and inevitably in present circumstances - we envisage a modest development building upon existing or already planned initiatives. Current initiatives such as the National Land Information Service (NLIS) and its equivalent in Scotland are entirely complementary to the concept of the National Geospatial Database espoused here, not least because they build upon the same Ordnance Survey-supplied topographical framework. Figure 2 shows how the NLIS concept fits neatly into the concept of the NGD. It is but one of the many such systems or services (see figure 3) that might be created, all of which will interact because they all employ the same set of standards and framework.
All of the OS and other bodies experience to date indicates that, though the technical problems in bringing about the National Geospatial Database are not trivial, they pall into insignificance compared to those arising from use of different standards and conventions, from different institutional and financial incentives on the part of different data custodians and the inevitable suspicions of other bodies endemic in a changing world. It seems likely that the private sector can help create individual applications on the National Geospatial Database and perhaps even some of the technical infrastructure. But the diverse sourcing of the data and the reality of the government machinery ensures that only government itself - pressed by potential users of the National Geospatial Database - can bring this vision to fruition. Moreover, it is essential to ensure that the National Geospatial Database is organic and evolves in terms of its quality, extent, consistency and use. Only user pressure will lead to changes in the data collected, albeit over a long period. Finally, to achieve the whole vision of the National Geospatial Database without some form of central control and ownership will be extremely difficult if not impossible. At the minimum, some organisation to monitor and encourage adherence and to publicise unrectified shortcomings will be necessary if the National Geospatial Database is not to be compromised by inconsistency and non-conformance to the rules.
Clearly initiatives such as the AGI Round Tables on Copyright and Government Information play a part in the development of a National Geospatial Database. Ordnance Survey believe that there is now a need to bring all the initiatives into a combined programme of research and development with the aim of progressively refining the concept and practice of a National Geospatial Database. To facilitate this, OS will be setting up a small group - entirely separate from its commercial operations and acting in the public interest - to help bring the National Geospatial Database concept into reality. Discussions have already been held with various bodies which will inevitably be key players in an National Geospatial Database and Ordnance Survey will seek to extend the range of those involved. Organisations that feel that they wish to be involved or even to be kept in touch with developments should contact the Director General of Ordnance Survey.
There is a long-standing principle in field survey that 'the whole is greater than the sum of the parts'. The same principle applies to the National Geospatial Database. Collaboration as outlined in this paper can multiply benefits to all stakeholders - the taxpayer, organisations themselves and for individual staff - all of whom should be able to will benefit from ever-increasing amounts of ever more complex and inter-related geospatial data.
[1] Office of Science and Technology, Progress through Partnership 15 - Report of the Technology Foresight Panel on Retail and Distribution, HMSO, 1995
[2] Ordnance Survey, Executive Agency Framework Document, Ordnance Survey 1995
Figure 1 NGD - A Conceptual View
Figure 2 NLIS - NGD in microcosm
Figure 3 National Geospatial Database -
different applications using related data on a common framework