International Journal of Spatial Data Infrastructures Research, 2008, Vol. 3, 146-167.

Editorial

Next-Generation Digital Earth∗

A position paper from the Vespucci Initiative for the Advancement of
Geographic Information Science

Max Craglia1, Michael F. Goodchild2, Alessandro Annoni1, Gilberto Camara3,

Michael Gould4, Werner Kuhn5, David Mark6, Ian Masser7, David Maguire8, Steve
Liang9, Ed Parsons10
1
European Commission Joint Research Centre
[[email protected]; [email protected]]
2
University of California at Santa Barbara [[email protected]]
3
National Institute for Space Research, Brasil [[email protected]]
4
University Jaume I, Castellon [[email protected]]
5
University of Münster [[email protected]]
6
State University of New York at Buffalo [[email protected]]
7
University College London [[email protected]]
8
ESRI, Redlands [[email protected]]
9
University of Calgary [[email protected]]
10
Google [[email protected]]

Abstract
This position paper is the outcome of a joint reflection by a group of international
geographic and environmental scientists from government, industry, and
academia brought together by the Vespucci Initiative for the Advancement of
Geographic Information Science, and the Joint Research Centre of the European
Commission. It argues that the vision of Digital Earth put forward by Vice-
President Al Gore 10 years ago needs to be re-evaluated in the light of the many
developments in the fields of information technology, data infrastructures, and
earth observation that have taken place since. It focuses the vision on the next-
generation Digital Earth and identifies priority research areas to support this
vision. The paper is offered as input for discussion among different stakeholder
communities with the aim to shape research and policy over the next 5-10 years.

Keywords: Digital Earth, spatial data infrastructures, Earth Observation,

GEOSS, geobrowsers, voluntary geographic information.

∗
This work is licensed under the Creative Commons Attribution-Non commercial Works 3.0
License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc-nd/3.0/

DOI: 10.2902/1725-0463.2008.03.art9

146
 International Journal of Spatial Data Infrastructures Research, 2008, Vol. 3, 146-167.

1. Introduction

Ten years ago, U.S. Vice-President Al Gore articulated a vision of “Digital Earth”
as a multi-resolution, three-dimensional representation of the planet that would
make it possible to find, visualize, and make sense of vast amounts of geo-
referenced information on the physical and social environment. Such a system
would allow users to navigate through space and time, access to historical data
as well as future predictions based for example on environmental models, and
support access and use by scientists, policy-makers, and children alike (Gore
1998).

At the time, this vision of Digital Earth seemed almost impossible to achieve
given the requirements it implied about access to computer processing cycles,
broadband internet, interoperability of systems, and above all data organization,
storage, and retrieval. As an example, with the technology then available, it was
going to take more than 100 years to download the data needed to cover the
Earth’s surface at 1 meter resolution, and more than a human lifetime to view it.

Ten years later, many of the elements of Digital Earth are not only available but
also used daily by hundreds of millions of people worldwide thanks to innovative
ways to organize and present the data and rapid technological advancements1.
Geo-browsing (browsing digital geographic information over the web) has
become a major industry2 and introduced novel ways to explore data
geographically, and visualize overlaid information provided by both the public and
private sectors, as well as citizens who volunteer new data (Goodchild, 2007).
Several major initiatives, discussed later in this paper, have also been launched
at regional and global levels to increase our capacity to observe and understand
our planet, its environment, and the impacts on and by society.

It is now time to ask: has the vision of Digital Earth been achieved? Has the
Grand Challenge been met? This paper argues that it has not, because in
parallel to increased availability and access to information, our collective
awareness of the need to understand interdependencies of environmental and
social phenomena on a global scale has also increased. Major natural disasters
like the Indonesian Tsunami (2004), Hurricane Katrina (2005), and the Sichuan
earthquake (2008) to name but few have contributed to this raised awareness,
together with mounting evidence from the Intergovernmental Panel on Climate
Change (www.ipcc.ch) of the profound changes occurring in our ecosystem, and

1
These advancements include graphics co-processors with as much as 512 MB dedicated memory
(in 1998 64MB was predicted), clever server-side data caching, and multiresolution techniques
allowing for storage and progressive visualization of multiple Levels of Detail (LOD).
2
Although the market now seems quite large in terms of number of user, commercial geobrower
makers admit that these services have yet to be “monetised” via advertising or subscription fees.

147
 International Journal of Spatial Data Infrastructures Research, 2008, Vol. 3, 146-167.

the cost of non-action, estimated at a minimum of 5% of global GDP year on year

(Stern, 2006). Also technological expectations have risen over the past decade:
ordinary web users now expect information, from anywhere and about anything,
to be easily and quickly accessed, raising the bar for scientists and technologists
to properly organize, represent, and then serve, such data.

The more we understand the complexity of interactions and inter-dependencies

between environmental and social phenomena at different levels, local, regional,
global, the more we need dynamic information systems to provide reliable,
accurate, timely, and openly accessible information at the relevant geographic
and temporal scales. The more geographic information we have, the more we
see the need for sophisticated processing and analysis models that can turn
information into insight and intelligent action. It is now necessary therefore to
take stock of current developments and refocus the vision towards the next
generation Digital Earth.

2. Key Developments since Digital Earth

It may be useful to group key developments of relevance to this paper into four
related themes.

2.1: Organizing Geographic Information

This theme includes the many initiatives since the early 1990s aimed at
increasing the availability and accessibility of geographic information through the
development of spatial data infrastructures (SDIs). By the mid 1990s, Masser
(1999) identified at least 11 SDIs at varying stages of development spanning
large countries like the USA, Canada, and Australia, small ones like the
Netherlands and Portugal, and developing nations like Malaysia, Indonesia, and
Qatar. That first generation of SDIs was largely led by national mapping agencies
and oriented towards the completion of national spatial databases addressing
topography and other key layers of general use. The documentation of existing
resources via metadata, and access mechanisms via catalogues and
clearinghouses were other key features of these early developments.

Since then we have seen a rapid diffusion of SDIs world wide facilitated by the
establishment in 1996 of the Global Spatial Data Infrastructures Association3 that
has helped the promotion of best practice and sharing of experiences, and
capacity building in the Americas, Africa, Asia and the Pacific and Europe. In
Europe a major recent development has been the adoption of a legal framework
in 2007 to establish a distributed Infrastructure for Spatial Information in Europe

3
www.gsdi.org

148
 International Journal of Spatial Data Infrastructures Research, 2008, Vol. 3, 146-167.

(INSPIRE)4 built on the SDIs of the 27 Member States of the European Union. At
the international level, it is also worth noting that in 2006 the UN Geographic
Information Working Group (UNIGWG)5 developed the vision, strategy and the
institutional governance framework for a United Nations SDI initiative. In its initial
phase (2008-2010) the UNSDI initiative is project-based, defined around projects
outputs that can involve non-UN partners but in the medium-term the UNIGWG
membership recognizes that the UNSDI will require legislative legitimacy.

The nature of these more recent SDIs has also shifted with an increased number
of stakeholder organizations engaged in the process. There is also a stronger
emphasis on distributed data and processes, and the interoperability of services
to discover, view, access, and integrate spatial information. The interoperability of
systems through services has been the major focus of the Open Geospatial
Consortium6 established in 1994 as an international partnership between
government agencies, industry, and academia. The OGC client-server interface
specifications and the standards adopted by the International Standards
Organization (ISO) have become the cornerstone of most SDIs in their current
form. In spite of this greater emphasis on interoperability through services, the
underlying basic approach to a SDI architecture has not evolved much during the
last 10 years.

2.2: Geography as a way to organize information

This theme includes the development of geo-browsers (e.g. Google Earth,

Microsoft Virtual Earth, NASA Worldwind, ESRI ArcGIS Explorer) which use the
globe as mechanism to pan, zoom, and fly over the Earth’s surface to areas of
interest much as in the original vision of Digital Earth. Associated to these 3D
representations of the Earth are also 2D applications (Google Maps, Microsoft
Live Search Maps) that also allow users to add and share information via simple
Application Programming Interfaces (APIs). Beyond visualization however,
geobrowsers have become client applications for accessing a more complex
infrastructure behind the scenes and for fulfilling a wider goal: organizing the
world’s information, in part spatially (Jones 2007).

Grossner et al. (2008) identify the “Digital Earth Initiative” (DEI) chaired by NASA
as a key milestone in trying to put the Al Gore vision into practice. The initiative
involved a number of US federal agencies and ran between 1998 and 2001
focusing on interoperability, infrastructure and organizational issues. A major step
forward came from the commercial sector in 2001 with the launch of Keyhole’s
Earth Viewer that demonstrated the technical feasibility of providing a global view

4
www.ec-gis.org/inspire
5
http://www.ungiwg.org/
6
www.opengeospatial.org

149
 International Journal of Spatial Data Infrastructures Research, 2008, Vol. 3, 146-167.

of the world using imagery on desktop computers, further enhanced by the

increased availability of powerful graphics cards in standard PCs. Google
acquired Keyhole in 2004 and launched its Google Earth in 2005, the same year
also making available its API for Google Maps thus making it possible for
anybody to add information to the Google platform. The Keyhole Markup
Language (KML) has also become a powerful way to document and index
information with a geographic reference, and display it on maps or globe
interfaces. The declared aim of Google is therefore not to organize geographic
information (as in SDIs) but to use geography as a way to search and view
information with a geographic footprint.

The success of Google Earth and Google Maps with hundreds of millions of
users, spurred Microsoft to accelerate its own developments and release at the
end of 2006 its alternative products, with a stronger emphasis on 3D
visualizations, particularly in urban areas. In parallel ESRI released a viewer
(ArcGIS Explorer) for its main GIS software platform also based on the globe
metaphor.

It is not to be unexpected that users focus their attention on the geobrowser

client, for after all it is this which must be installed on their desktop machine and
allows them to interact with geographic information. However, a strong case can
be made that the loosely connected information model, based on the standard
web architecture which allows geodata to be both found and published and the
servers that host data and process client requests are actually more important.
We could refer to this as the GeoWeb infrastructure to provide contrast to the
traditional SDI approach. The technical breakthroughs in horizontal (multiple
machines) and vertical (multiple hardware/software tiers, together with data
caching mechanisms have perhaps been the major contribution to the success of
the technology platforms and the biggest revolution in information organization
and access in many years.

Although these developments have been led by the private sector on a

commercial basis, their widespread success has also led an increasing number
of public-sector agencies to use these platforms to visualize their data. Examples
in this respect are the Cadastre of Spain7 and the European Environment
Agency8. Also worth mentioning is the leadership taken by China in promoting a
bi-annual International Symposium on Digital Earth starting in 1999. Since 2006
China acts as Secretariat to the International Society for Digital Earth9 promoting

7
http://www.catastro.meh.es/servicios/wms/wms.htm#_Buscar_parcelas_en_Google_Earth
8
http://www.eea.europa.eu/highlights/eea-and-microsoft-will-bring-environmental-information-to-
your-fingertips
9
http://www.digitalearth-isde.org/

150
 International Journal of Spatial Data Infrastructures Research, 2008, Vol. 3, 146-167.

international collaboration and scientific exchange to achieve the vision of Digital

Earth.

2.3:Geosensing the World

Geosensors can be defined as any device receiving and measuring

environmental stimuli that can be geographically referenced. As such they
include satellite-based sensors providing multi-spectral information about the
Earth’s surface (imagery, land cover, vegetation indices and so on), air-borne
sensors for detailed imagery but also for laser scans of physical or man-made
structures (LiDAR), and sensors near, on, or under the Earth’s surface measuring
anything from physical characteristics (pressure, temperature, humidity) and
phenomena (wind, rain, earthquakes), to the tracking of animals, vehicles, and
people. Large-scale networks of sensors have been in existence for several
decades. Examples include the network of the World Meteorological Organization
(Figure 1) and the Argos network of buoys measuring temperature and salinity of
the world’s oceans (Figure 2). What is novel is the web-enablement of these
sensors and their networks so that individual sensors can be discovered, tasked,
and accessed through web standards (sensor web), and that the networks can
exchange information through interoperability arrangements.

Figure 1: Weather and Climate measuring network

(Source: GEO, 2007a)

151
 International Journal of Spatial Data Infrastructures Research, 2008, Vol. 3, 146-167.

Figure 2: the ARGO Buoy Network

(Source: GEO, 2007a)

Linking existing systems and networks to achieve comprehensive, coordinated

and sustained observations of the Earth system is the objective of the Global
Earth Observation System of Systems (GEOSS). GEOSS is overseen by the
Group on Earth Observations (GEO)10, an intergovernmental organization at the
ministerial level comprising of 73 nations, the European Commission, and 52
international organizations. A major role of GEOSS is to promote scientific
connections between the observation systems that constitute the system of
systems, and promote applications across nine societal benefit areas (Figure 3).

Figure 3: GEOSS nine Societal Benefit Areas

Source: GEO, 2007b

10
http://earthobservations.org/

152
 International Journal of Spatial Data Infrastructures Research, 2008, Vol. 3, 146-167.

The GEOSS 10-Year Implementation Plan11 (2005-15) underscores the point that
the success of GEOSS will depend on data and information providers accepting
and implementing a set of interoperability arrangements, including technical
specifications for collecting, processing, storing, and disseminating shared data,
metadata, and products. GEOSS interoperability will be based on non-proprietary
standards, with preference to formal international standards. The architectural
design of GEOSS mirrors to a large extent that of the SDIs discussed in Section
2.1, with emphasis on metadata, catalogues, clearinghouses and portals, and
registries of existing standards across different communities.

It is worth noting that a new breed of sensor networks, called wireless sensor
networks (WSN), have demonstrated the potential to revolutionize the way we
acquire geospatial data. Different from the traditional, large-size, complex and
costly sensor stations, a WSN typically consists of miniature, battery-powered
nodes with power-efficient CPUs, short-range radios and low-cost sensors12. The
software that runs on the WSN nodes allows them to self-assemble into ad-hoc
networks, such that the network can be easily deployed (e.g., sensors can be
seeded from a low-flying airplane throughout hazardous areas) and data can be
relayed across multiple hops and from long distances.

As discussed previously, networks of sensors are not new. What is new is that
WSN changes sensor deployment strategies. WSN allows:

(1) High resolution spatial and temporal sensing: WSN’s self-assembled ad-hoc
network and the WSN nodes’ low unit-price makes very large scale deployments
economically feasible13, and also enables long-term data collection at scales and
resolutions that are difficult, if not impossible, to obtain otherwise;

(2) Pervasive and non-intrusive sensing: WSN nodes’ miniature size allows them
to be embedded in the physical world without disturbing the environment;

(3) Proactive sensing: WSN nodes are proactive and intelligent sensors rather
than passive data collectors. With embedded processors and radios, WSN nodes
can be programmed to share information with each other and modify their
behavior based on collected data. WSN nodes are, in fact, miniature computers
and have been projected to be the next computing class in 2010 (Bell, 2008).

11
http://earthobservations.org/documents/10-Year%20Implementation%20Plan.pdf
12
For example, a typical WSN node can be Crossbow Inc.’s MICA2 motes, which incorporates an
8-bit, 7 Mhz processor, a radio with a range of about 500 feet, 4KB or RAM, and 128KB of program
memory. (www.xbow.com)
13
Prices of a WSN node have been dropping steadily since these devices were introduced.
Projection from one vendor (www.dust-inc.com) suggests that the cost will drop to about $10 at
mass production stage.

153
 International Journal of Spatial Data Infrastructures Research, 2008, Vol. 3, 146-167.

This new class of sensing platform will provide SDIs and GEOSS with an
unprecedented volume of real-time geosensor data, along with high spatial and
temporal resolution.

A set of developments within the category of geosensors is that of citizens as

sensors, volunteering geographic information. Goodchild (2007) argues that there
is a long tradition of non specialists contributing to the collection of scientific
information such as the case of the Christmas Bird Count
(http://www.audubon.org/bird/cbc) or the collection of weather information in the
GLOBE programme (http://www.globe.gov) but that only recently the
convergence of greater access to broadband connections, the availability of
Global Positioning Systems at affordable prices, and more participative forms of
interaction on the Web (Web 2.0) have enabled vast numbers of individuals to
create and share geographic information. Platforms such as Google Maps and
Microsoft Live Search Maps have made it possible to publish and make
geographically searchable user-generated content to an unprecedented rate.
Initiatives such as Wikimapia14 and OpenStreetMap15 show how organised
volunteered information can challenge traditional data suppliers with good-quality
products that are openly accessible to all. As observed by Goodchild, the
potential of up to 6 billion human sensors to monitor the state of the environment,
validate global models with local knowledge, and provide information that only
humans can capture (e.g. emotions, and perceptions like fear of crime) is vast
and has yet to be fully exploited. In general SDIs as currently designed and
implemented do not handle well geosensor-based data (beyond remotely sensed
imagery snapshots), which tend to arrive in real-time, are more continuous in the
sense that changes over time can be considered at much higher frequency, and
are available in much smaller packages than traditional georesources: entire
images or maps.

2.4 Innovation in supporting technology

Digital Earths built today will be able to benefit from some very significant
technology breakthroughs. Any Digital Earth will most likely be built using a
technology stack comprising a lightweight client, a high speed computer network
and a sophisticated server infrastructure. This type of computer system
architecture in which data and processing resources are managed on remote

14
(http://www.wikimapia.org)
15
(www.openstreetmap.org)

154
 International Journal of Spatial Data Infrastructures Research, 2008, Vol. 3, 146-167.

servers accessed over the Internet is now being referred to as ‘cloud

computing’16 This is a type of ‘utility’ or service-based computing model.

The fact that today geobrowsers function at very high speeds while accessing,
on-line, vast qualities of global-scale geodata is due in large part to
advancements in the underlying technology. On the client side (local processing),
the geobrowser today has access to many times more basic system memory
than was available in desktop computers a decade ago. In its October 27, 1997
issue, Electronic Times told of home PCs being ready to make the jump from 32
MB to 64 MB of main memory, and for increases to be on the order of 20 MB
annually through 2001. These estimates have been greatly surpassed, such that
domestic laptop computers now boast up to 2-3 GB of main memory. More
importantly the same has been true for graphics memory, key to making
geobrowsers “fly” in real-time rather than as pre-recorded flights. While Keyhole
co-founder Avi Bar-Zeev tells us (Crampton 2008) that at the time of Gore’s
speech PCs were capable of “fluidly drawing a 3D earth”, we must remember that
such capability was not in the hands of ordinary programmers. Today, however,
all graphics programmers benefit from the evolution over the past decade of
graphics hardware, from rendering thousands to several millions of polygons per
second. Smooth graphic animation is also achieved by client (and server)
caching of information that is progressively streamed over a network connect
from a system of servers.

Given the vast quantities of information that are needed to represent Earth
digitally (not to mention information licensing requirements) it is impractical for
each user to maintain their own local copy of a Digital Earth. A more practical
solution is to store the information of a collection of shared servers that can be
accessed over a high-speed network connection, Although multiple protocols are
available in theory, the major technology providers are rapidly converging on
using Web protocols (especially http and XML).

On the server side, Google (and later Microsoft) boldly illustrated the speed
accruing from distributing its index and its stored data across thousands of
servers around the world. These servers cache copies of the most frequently
used or viewed data, including map tiles. They are also capable of processing
vast quantities of information in order to respond to client queries. These
developments now make available to ordinary PC users what in Al Gore’s speech
was envisaged as requiring expensive and specialized head-mounted displays.

16
The term ‘cloud’ originates from the symbol used to represent the Internet on early conceptual
diagrams of how distributed computer systems worked.

155
 International Journal of Spatial Data Infrastructures Research, 2008, Vol. 3, 146-167.

3. Strengths and weaknesses of current developments

The preceding section has provided a broad-brush overview of developments

providing context for our vision of a Digital Earth as a framework to harmonise
our efforts in the geospatial information realm, and to increase our collective
understanding of the state of our planet and the interactions between physical
and societal environments on it.

It is important to acknowledge that each set of initiatives introduced in Section 2

has its own characteristics in terms of leading actors, drivers, target audience,
and implementation mechanism as summarized in Table 1. For the sake of
clarity, GEOSS and volunteered geographic information (VGI) are treated
separately in the table.

Table 1: Key features of recent developments

Lead Actors Key Main Implementation

Drivers Target mechanism
Audience
SDI Public sector Inform Analysts in Legal
policy public framework (not
sector always)
Geobrowsers Private sector Market Mass Market +
share + market voluntary
Advertising
Revenue
GEOSS Public sector Inform Scientists Voluntary +
policy political
VGI Individuals/groups Social Society Voluntary
networking

As indicated in the Table, SDIs and GEOSS have many similarities in respect to
leading actors, drivers, and primary target audiences. Both have a focus on data
search, retrieval, and access, which then needs processing by experts in the
different domains. Their appeal to non-expert users is therefore limited, until such
time when many services are developed on top of these infrastructures to
process the data and return information understandable and relevant to non-
experts. Their basic architecture is also similar, and based on distributed nodes
and systems that rely on metadata, catalogues, and a set of ancillary services to
search, and retrieve the data. GEOSS emphasizes also dissemination services
able to cater for different communities with varying level of access (e.g.
GEONetcast services for data dissemination via satellite communication in
developing countries).

156
 International Journal of Spatial Data Infrastructures Research, 2008, Vol. 3, 146-167.

There are some differences between SDIs and GEOSS, particularly in relation to
implementation mechanism: whilst SDIs are by and large being developed within
the bounds of legal and administrative jurisdictions which may avail themselves
of legal tools to impose specifications and data sharing policies, GEOSS is
entirely voluntary (although with strong political backing), and needs to balance
the respect for existing systems, with their technical and data policy
characteristics, with the need to develop agreed interoperability arrangements
and data sharing principles to support the long term sustainability and use of
earth observations. As progress is being made on interoperability at the technical
level, greater attention is now been given to the data policy dimension, which is
particularly challenging for the implications it may have on organizational
arrangements and dependencies.

In addition to these similarities and differences, SDIs and GEOSS share a

number of limitations in the components of their infrastructures and approach:

Metadata: useful to make visible the wealth of information resources already

existing and to allow the opportunity for re-use. However, current standards-
based approaches to metadata require considerable human input, and are
difficult to maintain up-to-date. Moreover, they primarily represent the perspective
of the data producer on the quality and utility of the data, and do not allow for
users to express their measures of fitness-for-purpose. Also they are focused on
datasets, not features themselves, and they do not adopt recent innovations in
related fields of multimedia data exploitation, which embed a significant amount
of metadata with the data themselves.

Catalogues: stores of searchable metadata reflecting a library metaphor.

Separating the metadata from the data they refer to poses challenges of
synchronization in the event of change. Searching across distributed catalogues
is time consuming, and current OGC specifications are not tight enough to allow
for unequivocal implementation, so that individual adapters may be required for
each catalogue, minimising scaleability17. Multilingual search and retrieval are
also very challenging and in need of further work.

Geoprocessing services: very few are available to process data into information
relevant to different categories of users. The description of services in respect to
processes, relevance, limitations, expected outcome, reliability, and
trustworthiness needs significant additional work, as well as the chaining of
services which is still at an early stage and requiring considerable expert input
(Crosier et al., 2003). A promising development is the emergent interest in
GEOSS for grid processing to support complex model processing, and countries

17
See http://inspire.jrc.it/reports/DistributedCatalogueServices_Report.pdf

157
 International Journal of Spatial Data Infrastructures Research, 2008, Vol. 3, 146-167.

that don’t have adequate computer facilities. Geoprocessing services are offered
(shared) by GEO members (same approach for large data repositories). This
provides an interesting potential link with the activities taking place in e-science18
and cyber-infrastructures19, although the synergy between these activities,
GEOSS, and SDI developments needs to be developed further.

Portals and view services: many geo-portals exist worldwide, providing entry
points to SDIs (Maguire and Longley, 2005). Useful as they are, they often still
display a 2-D cartographic background, and a GIS layered view of the world
which makes them unattractive to non-experts and distant from the Digital Earth
vision and the user experience of geobrowsers.

Data: even for expert users, the ability to integrate data from distributed sources
(and disciplines) is hampered by the lack of explicit documentation of the
semantic properties of the data and mechanisms to refer such properties to
agreed semantic reference systems in a similar way as they would do when
transforming data to a spatial or temporal reference system (Kuhn, 2003).

Time: current SDIs address relatively static data in time slices, similar to
traditional GIS. As more and more dynamic data becomes available, through
geosensors for example, better ways to integrate spatial and temporal data,
analysis, and modeling are needed.

Model interoperability: GEOSS is aiming to link systems that often deal with n-
dimensional data which are needed to gain a better understanding of the
complex processes taking place in the Earth system, and to model future
scenarios. Model and process interoperability across multiple disciplines takes
the semantic reference system challenge one step further. Some initial research
has been undertaken in the context of the GEOSS Interoperability Process Pilot
Project (IP3) (Khalsa et al. 2008, 2007), and further work is expected in the
coming years but there is little doubt that this is one area of major scientific
challenge for the foreseeable future.

Policies: searching, finding, and accessing data and services require not only a
technical infrastructure but also a policy one. Some international agreements and
treaties exists in respect to Intellectual Property rights (WIPO Berne Copyright
Convention 1976, and WIPO Copyright Treaty 1996), the environment and
environmental information (e.g., Aarhus Convention, and Convention on
Biodiversity), together with UN principles relating to Remote Sensing of Earth
from Space (UNGA 1986). The extent to which further agreements are needed to
address access and use of heterogeneous and distributed data sources can be

18
www.rcuk.ac.uk/escience/
19
http://www.nsf.gov/dir/index.jsp?org=OCI

158
 International Journal of Spatial Data Infrastructures Research, 2008, Vol. 3, 146-167.

assessed by the fact that at the present time the OECD, GEOSS, and the
European Commission are all engaged in developing principles and guidelines
for data sharing, access and re-use (see for example OECD, 2008). Although
addressing slightly different audiences, these efforts all need to address the
balance between economic interests of data providers and member states with
considerations of openness, equity, and access particularly in matters relating to
environmental and human well being.

Equity: appropriate technology and policies to enable access to data and

information are crucial but cannot go without the infrastructure to develop
capacity, skills, and knowledge at the level of the individual, organization, and
institution so that information can be turned in actionable empowerment. All too
often the development of SDIs and related information infrastructure seems to
underestimate the importance of investing also in the education and capacity-
building programmes necessary for these infrastructures to support development
and not widen the digital divide.

4. Overcoming the limitations

What can the current generation of SDIs and GEOSS learn from geobrowsers,
social networking, and volunteered geographic information (VGI) to help
overcome some of the limitations highlighted above? Clearly the success of
geobrowsers with hundreds of millions of users demonstrates the power of the
Digital Earth vision, and of simple and free applications (Crampton, 2008).
Intuitive interfaces, speed in search and retrieval, and the use of imagery which is
often more understandable than maps to non-expert users have contributed to
the widespread use of these tools, and the popularization of geography as a way
to organise information. They have created the informational and technical
infrastructure, with hundreds of thousands of computers organised in server
farms (the so-called “cloud computing” architectural style), upon which many
other applications can be built, including VGI. Harvesting data and metadata and
building centralised indexes overcomes existing limitations of distributed
searches via catalogues. At the same time, what is technically achievable by
single industries may not be politically acceptable if attempted by government
organizations at national or international levels for perceived loss of sovereignty,
ownership, and privacy. It is noticeable how individuals may be less concerned
about giving away personal information to a private company than to a
government organization (Economist, 2008a). Moreover, the development of
large server farms begins to raise issues about their environmental footprint i.e.
energy consumption (Economist, 2008b).

Whilst geobrowsers have democratized access to geographically organised

information, their stated mass-market orientation means they are less concerned
with scientific and policy domains. They therefore do not support an

159
 International Journal of Spatial Data Infrastructures Research, 2008, Vol. 3, 146-167.

understanding of the current state of the Earth, changes over time, causes of
change and relationships between physical phenomena and human activities.
Data is generally not accessible, and the notions of a laboratory for multi-
disciplinary science and a bridge between science and society are not amongst
their objectives. This argues for the development of not one Digital Earth, but
multiple interoperable ones, addressing different audiences but with the ability
nevertheless to communicate, share, and learn from each other.

Social networking, Web 2.0, and VGI offer also enormous opportunities to
develop SDIs for scientific and policy-support purposes which are yet to be
exploited. As an example, some of the current limitations of metadata could be
overcome with more participative methods of user classification and feedback,
similarly to what is already common practice in commercial services such as
Amazon or eBay. Equally, offering users information about what previous users
have found interesting and useful may facilitate searching and retrieval beyond
the current flat database structures of catalogues. The opportunity for GEOSS to
include information provided by local communities about their environment should
also be grasped in order to achieve its stated vision of realizing “a future wherein
decisions and actions for the benefit of humankind are informed via coordinated,
comprehensive and sustained Earth observations and information” (GEO, 2005).
Not only are local communities the best source of knowledge about local
conditions and changes, but their engagement in a shared framework would also
address the equity, access, and empowerment issues highlighted above as one
of the important limitations of current approaches. Clearly, there are also issues
to be still fully addressed like the motivation people have to volunteer information,
the process needed to assure the quality of the information provided, and
appropriate technologies to enable people in every part of the world to participate
regardless of their social, economic, and cultural contexts.

5. Towards the next-generation Digital Earth

How the Earth’s environment is changing, and what are the consequences for
human civilization, are among the fundamental questions of our time. Sound
scientific knowledge, and reliable and up-to-date environmental data and
information, are necessary to address these questions and underpin the policies
necessary to affect change. As indicated earlier significant progress is being
made in the availability and quality of environmental and geographic information
at our disposal, and in connecting information systems and new sources of data.
The diffusion of SDIs, the efforts of GEOSS, and developments in industry and
civil society have major contributions to make in answering these critical
questions. Taken individually however, none of these developments can achieve
the objective or get us to the vision of Digital Earth put forward more than 10
years ago. Now is the time to set a new vision of what can be achieved within the

160
 International Journal of Spatial Data Infrastructures Research, 2008, Vol. 3, 146-167.

next 5-10 years, building on what is existing, bridging the gaps, and overcoming
the limitations identified. Below are the initial elements of such a vision.

1. Not one Digital Earth, but multiple connected globes/infrastructures

addressing the needs of different audiences: citizens, communities, policy-
makers, scientists, educationalists.
Each audience has a distinct set of needs for information about Earth and
its future, so we anticipate that each would be accommodated by a
specially designed Digital Earth. One might encourage members of the
general public to contribute their own observations, while another would
present only the most rigorously obtained scientific results. Each would
be a view, however, of a single coordinated, distributed data resource.
Different views would require different levels of detail, of geometry,
imagery, attributes (semantics), etc.

2. Problem oriented: e.g. environment, health, societal benefit areas, and

transparent on the impacts of technologies on the environment
While it is important that specific problems be addressed in focused
ways, Digital Earth should still clarify the interactions between problems
and objectives – the difficulties of achieving one objective, such as
reducing the costs of energy, with others such as impacts on the
environment and food production.

3. Allowing search through time and space to find similar/analogue situations

with real time data from both sensors and humans (different from what
existing GIS can do, and different from adding analytical functions to a virtual
globe)
While there are strong affinities between the current generation of virtual
globes and earlier GIS technology, it is clear that users expect virtual
globes to answer a different kind of query, one that is less precise and
quantitative, and more attuned to exploration and browsing. For example,
one popular use of Google Earth is to search the globe for similar
conditions – for example, to find areas that are as vulnerable to tsunamis
as the coasts of the Indian Ocean.

4. Asking questions about change, identification of anomalies in space in both

human and environmental domains (flag things that are not consistent with
their surroundings in real time)
One of the most compelling benefits of satellite images, maps, and virtual
globes lies in their ability to provide context, by displaying information in
its correct geographic position. The next generation of Digital Earth
should allow for rapid search for geographic anomalies, that is, situations
that are inconsistent with their geographic context, such as outbreaks of
disease, biodiversity hotspots, or anomalous levels of air pollution.

161
 International Journal of Spatial Data Infrastructures Research, 2008, Vol. 3, 146-167.

5. Enabling access to data, information, services, and models as well as

scenarios and forecasts: from simple queries to complex analyses across the
environmental and social domains.
One of the challenging issues today is to combine environmental
modeling and forecasting with its socio-economic impacts. Traditional
flood forecasting or mapping of natural hazards risk zones loose their
value if their social impact is not assessed. Meanwhile such models have
immediate economic consequences (e.g. property value reduction in the
case of identified risk) and for this reasons both model reliability and
appropriate communication tools (including visualization) are strongly
required.

6. Supporting the visualization of abstract concepts and data types (e.g. low
income, poor health, and semantics)
Advances in dynamic visualization environments (see, for example,
www.gapminder.org) show strong potential for decision support and
increased understanding of global, complex, and abstract phenomena.
Bringing these capabilities to the next generation Digital Earth will turn
these into important tools for education, awareness-rainsing, and
informed decision making. Different perspectives on phenomena like
poverty or health and their indicators can now be made explicit through
ontologies, and mappings between them have become possible.

7. Based on open access, and participation across multiple technological

platforms, and media (e.g. text, voice and multi-media)
The geoinfo community may have a great deal to learn from the wider
multimedia community. For this to happen the emphasis on maps (fixed
geometry plus labels) as central object of inquiry, study and navigation,
must shift and dynamic elements must be incorporated. There is
promising work already underway that integrates georeferenced moving
video with other static geographic data sets.

8. Engaging, interactive, exploratory, and a laboratory for learning and for multi-
disciplinary education and science.
The notion of virtual collaboratories is a key feature of e-Science (Access
grid20) to support the multidisciplinary exchange of knowledge across
scientific teams dispersed at multiple locations. In other fields interactive
learning tools, distance learning but also location-based games offer
platforms and lessons that can be built upon to develop teaching,
learning, and sharing environments for multiple audiences.

20
www.accessgrid.org/

162
 International Journal of Spatial Data Infrastructures Research, 2008, Vol. 3, 146-167.

6. Research needed to achieve the vision

1. Information integration (multi source and heterogeneous, multi-disciplinary,

multi-temporal, multi-resolution, and multi-media, multi-lingual)
Despite substantial progress, our ability to integrate geographic
information from multiple sources is still quite limited. We need a better
understanding of the statistical problems of integration across scales, the
linguistic problems of integrating across languages, and the semantic
problems of integrating across disciplines. This will require a substantial
effort by a number of collaborating disciplines: computer science,
information science, and the domain disciplines.

2. Space-time analysis and modelling (i.e. universal elements and language

for dynamic modelling, algebra of space-time change)
The next generation of Digital Earth should provide a powerful platform
for simulating the human and physical processes that operate on the
Earth’s surface. While such models have been developed in many
domains, they use a myriad of approaches that are impossible to couple
or integrate. Fundamental research is needed to develop a
comprehensive language for simulation, and the software components
needed to make simulations easily interoperable across disciplinary
boundaries.

3. Schemes for tiling the curved surface of the Earth and for use in data
management, analysis, simulation, visualization
Each of the current generation of virtual globes uses its own approach to
structuring data. These systems are optimized for storage and display
purposes, but have limitations for the analysis of global scale data and
processes. Research on optimal structuring and indexing schemes has
been under way for the past two decades, but we do not yet know how to
design an optimal scheme that can support massive simulation of Earth-
surface processes.

4. Intelligent descriptions (automatic, user driven) of data, services,

processes, models, searching and filtering
Good progress has been made on standardizing the description of data,
through metadata and syntax standards. Still missing, however, are
adequate standards for the description of models, processes, and
services that can support search and the assessment of fitness for use.

5. Visualization of abstract concepts in space

Transformations from lower level, physical observations through
indicators defined on them, to abstract concepts like quality of life or
vulnerability need to be modelled and implemented, to support

163
 International Journal of Spatial Data Infrastructures Research, 2008, Vol. 3, 146-167.

visualization and reasoning. The wealth of data sources that can now be
tied into Virtual Globes, together with advances in complex system
modelling and semantic mappings, provide a richly equipped laboratory
for domain and information scientists to enable new uses of Virtual
Globes. Progressing from specific, well-defined case studies on well-
defined scientific phenomena to more complex cases of socially defined
and negotiated notions promises gradual, but significant progress.

6. Computational infrastructures to implement vision (architecture, data

structures, indexing, interfaces)
Advances in technology, information structuring, and organization of the
IT infrastructure have already made it possible to come close to the vision
of Digital Earth. The interoperability of multiple systems delivering data,
information, and models in real-time from multiple sources, including
passive sensors and humans, can increase by several orders of
magnitude the quantity of information posing new challenges to deliver
reliable, and timely quality information to multiple and diverse audiences
through multiple access media including voice.

7. Trust, reputation and quality models for contributed information and

services
Progress from traditional provider perspectives on data quality to broader
notions of fitness for use, trust, and reputation is already happening in the
context of Volunteered Geographic Information. Models of spatio-
temporal expertise can be developed and used in reasoning about the
suitability of data and services for specific applications.

8. Governance models and collaborative frameworks (business, institutional,

voluntary, communities of practice)
The emergence of hybrid infrastructures can already be observed
combining both voluntary and institutional data (e.g. using the Google
platform). Without a mechanism to clearly distinguish the different nature
of the data, users will be reluctant to take any formal decision (still
recognising the value of non-institutional data). An important challenge is
to build new models for reciprocal validation of data made available
through collaborative frameworks (e.g. validation of precise quantitative
comprehensive information about air quality collected through few
professional measuring stations to be combined with air quality data
collected through mobiles equipped with appropriate nanotechnologies
measuring only a limited number of parameters). Communities of practice
should be better engaged in Governmental decisions sharing their data
and knowledge in an easier and effective manner

164
 International Journal of Spatial Data Infrastructures Research, 2008, Vol. 3, 146-167.

9. Data sharing and open access policies

More systematic and comparable evidence is needed on the impact of
different access policies on organizations and society in the face of
multiple pressures on government budgets and increasing inter-
organizational competition for funding. If such evidence was already in
place, we would not still be facing tensions surrounding data access and
sharing policies. More work is also needed on incentives and barriers to
data sharing at individual, inter, and intra organizational levels.

10. Social and economic impacts of Digital Earth

Appropriate theoretical and methodological frameworks to assess the
social and economic impacts of geo-spatial information, and related
infrastructures are still poorly developed but are urgently needed to justify
the initial investment and the long-term sustainability of the infrastructure.
The costs of not acting also need attention (in a similar vein as the Stern
(2006) Review on Climate Change) at different scales from global, to
regional, national, and local to inform appropriate funding and business
models.

7. Next steps

This position paper is the outcome of a specialist meeting on Virtual Globes

organised in June 2008 by the Vespucci Initiative for the Advancement of
Geographic Information Science and the Joint Research Centre of the European
Commission. It represents the views of the scientists coming from a range of
disciplinary and professional experiences in the geographic, earth observation,
and environmental fields. Over the coming eighteen months the paper will be
presented to a wider range of stakeholders in these and related fields with a view
to seek feedback as well as additional ideas both on the vision and the means to
achieve it. At the end of this period we will report back on the feedback received,
and monitor regularly the progress made on the Vespucci web site
(www.vespucci.org).

References

Bell, G., 2008. Bell’s Law for the Birth and Death of Computer Classes,
Communications of the ACM, January 2008, volume 51(1), pp. 86-94.
Crampton, J., 2008. Keyhole, Google Earth, and 3D Worlds: An Interview with Avi
Bar-Zeev, Cartographica, volume 43(2), pp. 85–93
Crosier S.J., Goodchild M.F., Hill L.L., and T.R. Smith (2003) Developing an
infrastructure for sharing environmental models. Environment and Planning
B: Planning and Design 30: 487–501.

165
 International Journal of Spatial Data Infrastructures Research, 2008, Vol. 3, 146-167.

Goodchild M.F. 2007. Citizens as voluntary sensors: spatial data infrastructure in

the world of Web 2.0. International Journal of Spatial Data Infrastructures
Research 2: 24–32.[http://ijsdir.jrc.it/editorials/goodchild.pdf]
Gore Albert, 1999. The Digital Earth: Understanding our planet in the 21st
Century. Photogrammetric Engineering and Remote Sensing vol 65 (5), p.
528.
Grossner K., Goodchild M.F., and K. Clarke. 2008. Defining a Digital Earth
System. Transactions in GIS, 12(1), 145-160.
Group on Earth Observaton (GEO) Secretariat (Ed.) 2005. GEOSS 10-Year
Implementation Plan: Geneva: GEO
[http://earthobservations.org/documents/10-
Year%20Implementation%20Plan.pdf]
Group on Earth Observaton (GEO) Secretariat (Ed.) 2007a. The Full Picture.
Geneva: GEO. [http://earthobservations.org/documents/the_full_picture.pdf]
Group on Earth Observaton (GEO) Secretariat (Ed.) 2007b. The First 100 Steps
to GEOSS: Annex of Early Achievements to the Report on Progress 2007
Cape Town Ministerial Summit. Geneva: GEO.
[http://earthobservations.org/documents/2007_%20Annex%20of%20Early%2
0Achievements%20to%20the%20Report%20on%20Progress.pdf]
Economist. 2008a. Identity Parade. Special Report Technology and Government.
14th February.
[http://www.economist.com/specialreports/displaystory.cfm?story_id=106381
96]
Economist. 2008b. Down on the Server Farm. 22nd May.
[http://www.economist.com/business/displayStory.cfm?story_id=11413148&fs
rc=nwlehfree]
Jones, M.T., 2007. Google’s geospatial organizing principle. IEEE Computer
Graphics and Applications, July/August 2007, 8-13.
Khalsa S.J.S., , Nativi S., Geller G. And R. Lumsden. 2008. How GEOSS IP3
explores and enables interdisciplinary science Geophysical Research
Abstracts, Vol. 10,
[http://www.cosis.net/abstracts/EGU2008/04949/EGU2008-A-04949-
1.pdf?PHPSESSID=]
Khalsa, S. S.J., Nativi, S. Ahern, T. Shibasaki, R. and D. Thomas. 2007. The
GEOSS interoperability process pilot project Geoscience and Remote
Sensing Symposium, 2007. IGARSS 2007. IEEE International, 293-296. DOI:
10.1109/IGARSS.2007.4422788

166
 International Journal of Spatial Data Infrastructures Research, 2008, Vol. 3, 146-167.

Kuhn W. 2003. Semantic Reference Systems. International Journal of

Geographic Information Science, 17(5), 405-409.
Maguire D.J., Longley P.A. 2005. The Emergence of Geoportals and Their Role
in Spatial Data Infrastructures. Computer, Environment and Urban Systems,
29: 13-14.
Organization for Economic Cooperation and Development (OECD). 2008. OECD
Recommendations of the Council for Enhanced Access and More Effective
Use of Public Sector Information [C(2008036]. Paris: OECD.
[http://www.oecd.org/dataoecd/0/27/40826024.pdf]
Stern N. 2006. Stern Review on the Economics of Climate Change. Cambridge
University Press. [http://www.hm-
treasury.gov.uk/independent_reviews/stern_review_economics_climate_chan
ge/sternreview_index.cfm]

167