Abstract
There are claimed to be National Spatial Data Infrastructures under development in about 40 countries. In addition, many lower level governments are creating them and attempts are being made to create a Global Spatial Data Infrastructure. This paper summarises the overall concepts and progress, using the USA and the UK as particular (and disparate) examples. It highlights the problems that have been faced and the lessons learned.

Introduction
National Spatial Data Infrastructures are being created in many different countries (see table 1). The ‘drivers’ for this include the development of Geographical Information Systems (GIS); without routinely useable tools to collate, refine, analyse and distribute information, many decision-making processes are hampered. Other relevant factors include the advent and widespread use of the Internet, globalisation of many businesses and the export of certain types of management approaches to the public sector in many countries (see Longley et al 2001)



Table 1 Countries which have assembled or are assembling National Spatial Data Infrastructures as of mid-2000. Source H Tom, Oracle Corporation.

The history of GIS has been summarized by Coppock and Rhind (1991) and Foresman (1997). What began as a hugely expensive - and thus rare - mainframe computer tool for inventorying and mapping purposes in the mid-1960s blossomed in the 1990s. The statistics suggest that there are now about two million users of GIS world-wide and this figure is growing at about 20% per annum. Many of the active users of GIS are carrying out tasks on behalf of citizens, notably those carried out in governments. In addition, there are many everyday users of services which include GIS elements, sometimes without this being obvious (e.g. in many Internet searches). The annual commercial revenues generated from sales of GIS software are now over $1 billion world-wide. Typically this leverages a ratio of about 15 in terms of expenditures on hardware, staff training, data collection and operations. Thus the total annual GIS expenditure is now likely – in so far as it can be estimated – to be of the order of $15 to 20 billion. This figure is supported by estimates of expenditure by known users e.g. cadastral and mapping agencies. One surrogate for the growth of GIS is the number of people attending conferences: the numbers attending the user conference of the most popular GIS software have grown from 23 in 1981 to over 10000 in 2001. All this is not surprising for the range of applications of GIS is now truly extraordinary (Longley et al 2001). GIS has thus moved in 30 years from an esoteric side issue, based on the creation of software by academics and government, to a fully fledged global business. The great bulk of the software revenues are earned by a small number of US-based firms but other GIS expenditures are typically more local.

A key requirement is for appropriate ‘fuel’ for GIS. Mapping is important but the range of geographical information required is typically far beyond that held in map form (e.g. demographic, health, environmental, sales and market opportunities information). Most countries have governmental bodies charged with providing key data sets, especially ‘framework mapping’. But the ideal of having these information in machine form, manufactured to common and publicly defined standards, to be ‘interoperable’ and with ready accessibility is not met in any country. It might be thought that the situation is best in the most economically advanced countries. That is not necessarily true: the problems with the existing ad hoc situation in the USA have been summarised by the US Federal Geographic Data Committee (1997) as follows

‘In the United States, geographic data collection is a multibillion-dollar business. In many cases, however, data are duplicated. For a given piece of geography, such as a state or a watershed, there may be many organizations and individuals collecting the same data. Networked telecommunications technologies, in theory, permit data to be shared, but sharing data is difficult. Data created for one application may not be easily translated into another application. The problems are not just technical - institutions are not accustomed to working together. The best data may be collected on the local level, but they are unavailable to state and federal government planners. State governments and federal agencies may not be willing to share data with one another or with local governments. If sharing data among organisations were easier, millions could be saved annually, and governments and businesses could become more efficient and effective.

Public access to data is also a concern. Many government agencies have public access mandates. Private companies and some state and local governments see public access as a way to generate a revenue stream or to recover the costs of data collection. While geographic data have been successfully provided to the public through the Internet, current approaches suffer from invisibility. In an ocean of unrelated and poorly organized digital flotsam, the occasional site offering valuable geographic data to the public cannot easily be found.

Once found, digital data may be incomplete or incompatible, but the user may not know this because many data sets are poorly documented. The lack of metadata or information on the “who, what, when, where, why, and how” of databases inhibits one’s ability to find and use data, and consequently, makes data sharing among organizations harder…If finding and sharing geographic data were easier and more widespread, the economic benefits to the nation could be enormous’.

This statement actually understates the scale of the problems encountered in typical GIS projects or programmes. These normally involve dealing with the assembly of data from multiple sources and coping with scarce staff skills, ‘state of the art’ technology and frequently contested outcomes. Making effective and safe use of data encoded by different people to different levels of resolution and accuracy, collected at different times and without certification of quality, is often non-trivial. These are some of the problems that Spatial Data Infrastructures are designed to resolve.

What do we mean by a Spatial Data Infrastructure?
Formal definitions of SDIs vary significantly around the world. A definition of the US National Spatial Data Infrastructure is shown below, together with that of the putative Global Spatial Data Infrastructure (GSDI):

[NSDI is] ‘the technology, policies, standards, and human resources necessary to acquire, process, store, distribute, and improve utilization of geospatial data’. Source: Presidential Executive Order #12906: ‘Co-ordinating Geographic Data Acquisition and Access: The National Spatial Data Infrastructure’, 1994

‘The GSDI is the broad policy, organizational, technical and financial arrangements necessary to support global access to geographic information’. Source: Steering Committee of the Global Spatial Data Infrastructure (see http://www.gsdi.org)

The differences in these are subtle but important. For instance, the financial underpinnings of NSDIs were largely ignored in the early years, as were the commercial and private sector contributions. Many originally saw the entire NSDI enterprise being driven by new technologies, especially Geographical Information Systems. The progressive acceptance of Geographical Information Science has broadened the concept of NSDIs to include a stronger research and education agenda. In the USA in particular, the NSDI was originally driven by federal government agencies and the Federal Geographic Data Committee in particular; this has led to much debate on how to create leadership in an enterprise that necessarily spans many organisations of different sizes, value systems and other characteristics in the government, private, not-for-profit and academic sectors. No universal answer has yet emerged – or is likely to do so.

In the UK, a fundamentally different – and somewhat smaller scale and pragmatic – approach was followed. Beginning formally in 1995, the National Geospatial Data Framework (NGDF) was established with the primary objective of ‘unlocking geographic information’. Relatively little attempt was made to describe what is the NGDF; rather a strategy was defined and a series of tasks identified whose successful completion would achieve that objective. Details of the NGDF approach are given by Hadley’s paper to this conference.

Brief summaries and evaluations of the US NSDI and the UK NGDF
The US scene – an outsider’s view
The Presidential Executive Order (see above) identified three primary areas to promote development of the NSDI. The first was the development of standards, the second improvement of access to and sharing of data by developing a National Geospatial Data Clearinghouse, and the third was the development of the National Digital Geospatial Data Framework. All of these efforts were to be carried out through partnerships among federal, state and local agencies, the private and academic sectors, and non-profit organizations. Presidential Executive Orders in the USA are only enforceable in regard to federal agencies: such cross-sectoral adherence could only be achieved by influence and leverage obtained through federal funding.

The Federal Geographic Data Committee (FGDC) operates as a bureaucracy through a series of subcommittees based on different themes of geospatial data (e.g. soils, transportation, cadastral), each chaired by a different federal agency. Many of these groups have developed standards for data collection and content, classifications, data presentation and data management to facilitate data sharing. For example, the Standards Working Group developed the metadata standard, which was formally adopted by the FGDC in mid-1994 and has since been adapted in the light of international and other national developments. All of the FGDC-developed standards undergo an extensive public review process. This includes nationally advertised comment and testing phases plus solicitation of comments from state and local government agencies, private sector firms, and professional societies. The NSDI Executive Order mandated that all federal agencies to use all FGDC-adopted standards.

The second activity area is intended to facilitate access to data, minimize duplication and assist partnerships for data production where common needs exist. This has been done by helping to advertise the availability of Geographical Information through development of a National Geospatial Data Clearinghouse. Agencies producing data describe its existence with metadata and serve those metadata on the Internet in such a way that they can be accessed by commonly used Internet search and query tools. This has been rendered practicable by the development of metadata-creating software within GIS by the larger vendors. As a result of all this, nearly all federal agencies, as well as most States and numerous local jurisdictions, have become active users of the Internet for disseminating geospatial data. This model does not necessarily assume that GI will be distributed for free. Obtaining some of the data sets requires the payment of a fee, others are free. The Clearinghouse can also be used to help find partners for database development by advertising interest in or needs for GI.

The third NSDI activity area is the conceptualisation and development of a US geospatial framework. Framework data are those which underpin all GIS operations; they typically include geodetic control, topography, hydrography, transportation, the geography of governmental units, place names and (sometimes) cadastral information. The framework forms the foundation for the collection of other data, minimises data redundancy and facilitates the integration and use of geospatial data in combination. Such GI is typically collected by multiple agencies, usually within the boundaries of national or administrative territories; and some of these agencies have historically collected data primarily to suit their own local purposes. It should be pointed out that the USA is generally less well endowed in relation to a consistent geographical framework than many other nations. The federal nature of the country has ensured a multi-level and inconsistent hierarchy of geographical information to be produced by the federal, state and local governments. This contrasts with India, Britain and many other countries where many elements of a truly national framework have long existed.

The Executive Order directed the FGDC to develop a plan for completing initial implementation of the framework by the year 2000 but this has proved somewhat more difficult than the other two activities. A popular solution has been to create one metre resolution geometrically correct orthophotomaps in place of a conventional digital map framework, not least because this could be financed jointly by states and the federal government and implemented via the private sector.

The collaboration to create or refine the national NSDI has led to various tensions. These have occurred between federal agencies, between them and the states and local governments and – to a more limited extent because of their more modest involvement until the late 1990s – the private sector. These tensions arise partly because the NSDI permits the questioning of traditional, legacy roles for government bodies: the concept of assembling national databases by sewing together detailed ones held by states directly undermines the traditional role of some federal agencies. In addition, the NSDI has made public latent demands which have ramifications for budget holders: in many cases the beneficiaries of spending are not those called on to divert resources from other activities. Most recently, a US national report on NSDI pointed out the consequences of seeking a private/public sector partnership in developing the National Spatial Data Infrastructure. These consequences include an obvious requirement for the private sector to generate profits from trading in information and in financing infrastructure. These would necessitate, it was argued, a more cohesive and industry-friendly policy on information ownership and massive investment by government to convert its data into more coherent, consistent and object-oriented form. Partnerships of this kind – as opposed to normal contractual arrangements with the private sector working for the government - are difficult to make work. ‘Where next?’ for the US NSDI is not presently clear.

The UK scene
In the UK, the problem was not one of inconsistent or unavailable framework data – all these have existed in digital map form (derived from national coverage of up-to-date OS 1:1250 and 1:2500 scale mapping) since 1995 and are continuously up-dated. The key problems were enhancing knowledge of what could be done beyond a cadre of technical specialists, making known what GI was held by government departments and other players and securing greater transparency – and some greater consistency – of the rules under which GI was made available. The last point can only be understood in the light of policies to enhance the performance of the UK public service in operation throughout the 1990s: this was achieved by breaking the Civil Service into policy and delivery sectors, the latter provided by Executive Agencies. These Agencies have considerable discretion on how to achieve the key public service targets set by Ministers. In some cases (e.g. Ordnance Survey, the Meteorological Office, HM Land Registry) these were required to cover all or large fractions of their costs from user fees.

From the outset, NGDF embraced the private sector – two GI (not software) industry leaders were founder members of the Board and contributed very different perspectives to the discussions. The major constraints in the early days were

• in gaining legitimacy for the Board of NGDF. It was attacked by some users as a club of (largely) data providers and friends (even though the Association of Geographic Information, Britain’s ‘user body’ was formally represented on the Board).
• in forming enough of a consensus on what should be tackled – indeed what was properly in the remit of NGDF Board.
• how decisions could be made by a Board which then had to be funded by some of the organisations represented on it, and
• the low level of ‘spare change’ funding available.

Successes have been both measurable and less tangible. The measurable ones include the creation and operation of a Data Locator (see http://www.ngdf.org.uk) and the progress in creating a UK Standard Geographic Base. Recent changes in policy on Geographic Information – GI from many government departments is now available at very low cost, except for those organisations on Trading Funds (where they have to operate like a private company and generate returns covering all costs, including interest on capital) - may have been influenced by lobbying from NGDF members and the private sector. Funding was generated through success in government investment competitions, obtained by arguments for the long term benefits which would flow from widespread use in government of better GI. Intangible benefits certainly include a significantly greater level of awareness of GIS/GI and their potential benefits. This awareness extends to Ministers, government officials, the press, private sector management consultants and other ‘players’. In part, all of this was good fortune for the new government of 1997 came into office convinced of the need for more ‘joined up government’. This necessitated departments behaving less like information and policy ‘silos’ and working together to minimise unintended consequences of policy interactions. A common geographic framework, well documented metadata, easily available GI in digital form, understanding of GIS and appropriate staff skills are all required to make this a reality.

The global dimension: a Global Spatial Data Infrastructure (GSDI)
GSDI is the result of a voluntary coming together of national mapping agencies and a variety of other individuals and bodies (see Holland paper to this conference). The organization has sought to define and facilitate creation of a GSDI by learning from national experiences. As part of this process, it funded a scoping study carried out by Australian consultants. They outlined in early 2000 their conceptual view of how the inputs to decision-making were linked to a SDI and the potential benefits that would arise from having a global one. The consultants pointed out however that proof of the reality of the predicted benefits was difficult to find. The fundamental question for any GSDI business case is: what exactly is the product that governments and other agencies will be asked to fund? The consultants said bluntly that ‘The broad definitions of GSDI are not sufficiently operational (or practical) to build a clear business case around’. They went on to raise some key questions - which they recognized as being difficult to answer - but had to be tackled to give credibility to a formal business case for GSDI. These included

• Is GSDI a real Global SDI or a federation of national or regional SDIs?
• Is GSDI really about encouraging developing countries (or at least those that do not currently place emphasis on geospatial data to do so)? In this case, the product may be a suite of policy recommendations and technical assistance approaches that development agencies can apply in their client countries.
• Is GSDI a process, a general framework, or is it a particular product such as a world map or a comprehensive database?
• Who are the stakeholders and how do we establish the demand side of GSDI?

Given all this – and not surprisingly in view of the lack of any global government and the complications in building a national, let alone pan-national NSDI – progress on establishing a visible GSDI has been slow to date. The real reason is not about technology but about the need to clarify real need, the sources of finance and politics; the similar reasons for demise of the International Map of the World project (Longley et al 2001) indicates the fundamental barriers to global consistency have not changed. Ventures such as the Global Map, to which 77 countries have made some form of commitment, may be a pragmatic way of bringing about some of the GSDI’s ambitions.

The lessons learned
These can be summarised as set out below

• Typically, present arrangements for the creation, validation and exploitation of Geographic Information are based on legacy arrangements. Many different bodies collect information to suit their own or historical specifications. The existence, characteristics and availability of what is collected are not widely known. As a result, there is a considerable shortfall in the return on investment because we can not use the information in combination to get added value – despite the GIS technology.
• Whilst governments still create the bulk of the core Geographic Information, the private sector is playing an increasingly important role in data collection and exploitation of GI. It therefore must be factored into any NSDI as a central player from the outset.
• To make an NSDI a success requires a vision of what can be – but such visions can easily raise unrealisable expectations and subsequent disillusionment.
• Political leadership has proved essential to get government ‘buy in’ and significant funding in both the USA and UK
• That said, GI is not very exciting politically. It has been ‘sold’ to the most senior politicians in the USA as exciting new technology which will help US competitiveness and on its benefits in creating a better informed citizenry. In the UK, it has succeeded most commonly by portrayal as a necessary under-pinning of other government policies on transport, equal access to opportunities, environmental sustainability, etc i.e. GIS, GI and NGDF are not important in themselves but only for what they can help to achieve.
• The definition of what is needed for a successful NSDI is non-trivial and contentious. Academics have typically proposed grand schemes embracing the technology, information sources, institutional arrangements, legal frameworks and the ‘people skills’ and education. Most governments have been more pragmatic and tackled essential short term measures. The likely beneficiaries of NSDIs are often not the budget holders and those with budgets rarely have ‘spare funds’ or even mission approval for such activities.
• It is also not clear what constitutes success in creating an NSDI. Certainly measuring it in standard ways is not easy. Some of the gains are intangible ones. Set against this is the considerable expenditure of senior staff time – a precious commodity.
• The biggest problem is that a large number of players – including data collectors, software vendors, data brokers, citizens and end users – see themselves in having a central stake in NSDI. Even within government, multiple departments are inevitably involved, all with their own agendas set by different Ministers. Co-ordination is a major problem, not least because having anyone ‘in overall charge’ is difficult in any line management sense. Longley et al (2001) illustrate this by a simulation exercise.
• Whilst the general problems of disparate GI - specified for local tasks and controlled by individual players and whose existence or characteristics are little-known - is common in most countries, the most pressing issues are nationally (or even regionally)-specific. For instance, no known NSDI has yet had to consider the availability of GI as a security issue as may be the case in India.
• Finally, the GIS community – of which the author is a long-standing part – has only itself to blame for some of the problems it faces. One apparently trivial one is actually fundamental. Because we come largely from technical backgrounds, we have delighted in using terminology meaningless and alienating to decision-makers and citizens alike. NSDI, NGDF and GSDI are ‘insider’ terms. We have sometimes delighted in needless complexity, which also alienates.

The merits of establishing an NSDI are widely assumed by those involved to be self-evident, even if no compelling formal business case can be established for it. Why else would we have 40 countries be committed to establishing one? Many of these lessons however suggest that the case is not clear-cut. That is not this author’s considered conclusion. Geographic Information in one way or another underpins many of the activities of the state and of business. Given that, there are significant potential benefits to the taxpayer, to government and to business from review and reform of national arrangements for collecting, combining and exploiting Geographical Information. But to expect it to be easy is very unwise and to excite high expectations is to risk disillusion. Creating a successful NSDI is a essentially a highly challenging management task.

Acknowledgements
Though thanks are due to many past and current colleagues for their help and advice, the paper reflects the personal views of the author. Some of the text is derived from a new textbook on GIS, Geographic Information and GI Science (Longley et al 2001) with permission from James Wiley and Sons, the publishers.

Further information
Some web sites of selected NSDIs

• http://www.fgdc.gov/ 
• http://www.fgdc.gov/I-Team.html 
• http://www.ngdf.org.uk 
• http://www.cgdi.gc.ca 
• http://www.auslig.gov.au/index_sdi.htm 
• http://www.gsdi.org  
• http://www.gsdi.org/canberra/masser.html 
• http://www.anzlic.org.au/anzdiscu.htm 
• http://www.land.state.az.us/agic/agichome.html 

References

• Coppock J T and Rhind D W (1991) The history of GIS. In Maguire D J, Goodchild M F and Rhind D W (eds.) Geographical Information Systems, pp. 21-43, Longman Scientific and Technical, Harlow
  
  
• FGDC (1997) A strategy for the National Spatial Data Infrastructure. Federal Geographic Data Committee, Washington, D.C.
  
  
• Foresman T W (1997) (ed.) The history of GIS, Prentice-Hall, Upper Saddle River, NJ.
  
  
• Longley P, Goodchild M, Maguire D and Rhind D (2001) Geographical Information Systems and Science. Wiley and Sons, Chichester and New York (in association with ESRI Inc). Published April 2001.