11 Distributed GIS

Until recently, the only practical way to apply GIS to a problem was to
assemble all of the necessary parts in one place, on the user’s desktop. But
recent advances now allow all of the parts – the data and the software – to
be accessed remotely, and moreover they allow the user to move away from
the desktop and hence to apply GIS anywhere. Limited GI services are already
available in common mobile devices such as cellphones, and are increasingly
being installed in vehicles. This chapter describes current capabilities in
distributed GIS, and looks to a future in which GIS is increasingly mobile and
available everywhere. It is organized into three major sections, dealing with
distributed data, distributed users, and distributed software.

Geographic Information Systems and Science, 2nd edition Paul Longley, Michael Goodchild, David Maguire, and David Rhind.
 2005 John Wiley & Sons, Ltd. ISBNs: 0-470-87000-1 (HB); 0-470-87001-X (PB)
242 PART III TECHNIQUES

Learning Objectives and tablet computers, which we have termed Hand-Held

GIS (Section 7.6.4). Such devices are now widely used
for field data collection, by statistical agencies such as
After reading this chapter you will understand: the US Bureau of the Census or the US Department of
Agriculture and by utilities and environmental companies.
Such organizations now routinely equip their field crews
■ How the parts of GIS can be distributed with PDAs, allowing them to record data in the field and
instead of centralized; to upload the results back in the office (Figure 11.1);
and similar systems are used by the field crews of
■ The concept of a geolibrary, and the utility companies to record the locations and status of
transformers, poles, and switches (Figure 11.2). Vendors
standards and protocols that allow remotely are also offering various forms of what we have termed
stored data to be discovered and accessed; Server or Internet GIS (Section 7.6.2). These allow GIS
users to access processing services at remote sites, using
■ The capabilities of mobile devices, including little more than a standard Web browser, thus avoiding
the cost of installing a GIS locally.
cellphones and wearable computers;
Certain concepts need to be clarified at the outset.
First, there are four distinct locations of significance to
■ The concept of augmented reality, and its distributed GIS:
relationship to location-based services;
■ The location of the user and the interface from which
the user obtains GIS-created information, denoted by
■ The concept of remotely invoked services,
U (e.g., the doorway in Figure 11.2).
and their applications in GIS.
■ The location of the data being accessed by the user,
denoted by D. Traditionally, data had to be moved to
the user’s computer before being used, but new
technology is allowing data to be accessed directly
from data warehouses and archives.
11.1 Introduction

Early computers were extremely expensive, forcing orga-

nizations to provide computing services centrally, from a
single site, and to require users to come to the comput-
ing center to access its services. As the cost of computers
fell, from millions of dollars in the 1960s, to hundreds
of thousands in the late 1970s, and now to a thousand or
so, it became possible for departments, small groups, and
finally individuals to own computers, and to install them
on their desktops (see Chapter 7). Today, of course, the
average professional worker expects to have a computer
in the office, and a substantial proportion of households
in industrialized countries have computers in the home.
In Chapter 1 we identified the six component parts of
a GIS as its hardware, software, data, users, procedures,
and network. This chapter describes how the network,
to which almost all computers are now connected, has
enabled a new vision of distributed GIS, in which the
component parts no longer need to be co-located. New
technologies are moving us rapidly to the point where it
will be possible for a GIS project to be conducted not only
at the desktop but anywhere the user chooses to be, using
data located anywhere on the network, and using software
services provided by remote sites on the network.
In distributed GIS, the six component parts may be
at different locations.
In Chapter 7 we already discussed some aspects of
this new concept of distributed GIS. Many GIS vendors Figure 11.1 Using a simple GIS in the field to collect data.
are now offering limited-functionality software that is The device on the pole is a GPS antenna, used to georeference
capable of running in mobile devices, such as PDAs data as they are collected
 CHAPTER 11 DISTRIBUTED GIS 243
In traditional GIS, three of these locations – U , D, and
P – are the same, because the data and processing both
occur at the user’s desktop. The subject location could be
anywhere in the world, depending on the project. But in
distributed GIS there is no longer any need for D and
P to be the same as U , and moreover it is possible for
the user to be located in the subject area S, and able to
see, touch, feel, and even smell it, rather than being in a
distant office. The GIS might be held in the user’s hand,
or stuffed in a backpack, or mounted in a vehicle.

In distributed GIS the user location and the subject

location can be the same.

Distributed GIS has many potential benefits, and

these are discussed in detail in the following sections.
Figure 11.2 Collecting survey data in the field, on behalf of a
Section 11.2 discusses the feasibility of distributing
government statistical agency (Courtesy: Julie Dillemuth, João data, and the benefits of doing so, and introduces
Hespanha, Stacy Rebich) the technologies of distributed data access. Section 11.3
discusses the technologies of mobile computing, and
the benefits of allowing the user to do GIS anywhere,
and particularly within the subject area S. Section 11.4
■ The location where the data are processed, denoted by
reviews the status of distributed GIServices, and the
P . In Section 1.5.3 we introduced the concept of a
benefits of obtaining processing from remote locations.
GIService, a processing capability accessed at a
It is important to distinguish at the outset between the
remote site rather than provided locally by the user’s vision of distributed GIS, and what is practical at this
desktop GIS. time, and so both perspectives are provided throughout
■ The area that is the focus of the GIS project, or the the chapter.
subject location, denoted by S. All GIS projects Critical to distributed GIS are the standards and
necessarily study some area, obtain data as a specifications that make it possible for devices, data,
representation of the area, and apply GIS processes to and processes to operate together, or interoperate. Some
those data. of these are universal, such as ASCII, the standard for

Biographical Box 11.1

David Schell, proponent of open GIS

David Schell earned degrees in mathematics and literature from Brown

University and the University of North Carolina. Initially, he worked in
system development at IBM. Later, during the 1980s, David managed
marketing and business development programs for various Massachusetts
workstation startups. In these positions, he was actively involved in the
development of Unix software libraries and the integration of productivity
tools for office automation ‘measurement and control’ and scientific
applications. In the late 1980s, while working at real-time computing
pioneer Masscomp (later Concurrent Computer), David had the opportunity
to work closely with the Corps of Engineers Geographic Resources Analysis
Support System (GRASS) development team on technology transfer and
the integration of GRASS open source with commercial GIS and imaging
systems. This integration work resulted in an early prototype of the
principles of ‘Web mapping’. Carl Reed and John Davidson of Genasys
developed an operational system in which a user could access both
GenaMap and GRASS seamlessly from the same command line and where Figure 11.3 David Schell,
map graphics from both systems displayed into the same X-Window. President, Open Geospatial
Consortium
This resulted in the development of some of the earliest interface-based
interoperability models for geospatial processing.
244 PART III TECHNIQUES

In 1992, with the aid of a Cooperative Research and Development Agreement with the US Army Corps of
Engineers USACERL, David founded the Open GIS Foundation (OGF) to facilitate technology transfer of COE
technology to the private sector as well as continuing the commercial interoperability initiatives begun with
GenaMap. In 1994, responding to the need to engage geospatial technology users and providers in a formal
consensus standards process, he reorganized OGF to become the Open GIS Consortium (now the Open
Geospatial Consortium). With the support of both public and private sector sponsors, the OpenGIS Project
began an intense and highly focused industry-wide consensus process to create an architecture to support
interoperable geoprocessing. David was cited in 2003 as one of the 20 most influential innovators in the IT
community by the editors of CIO Magazine and received the CIO 20/20 Award for visionary achievement.
Speaking about the industry’s evolution, David Schell said: ‘It is most important for us to realize the far-
reaching cultural impact of geospatial interoperability, in particular its influence on the way we conceptualize
the challenges of policy and management in the modern industrialized world. Without such a capability,
we would still be wasting most of our creative energies laboring in the dark ages of time-consuming
and laborious data conversion and hand-made application stove-pipes. Now it is possible for scientists and
thinkers in every field to focus their energies on problems of real intellectual merit, instead of having to
wrestle exhaustingly to reconcile the peculiarities of data imposed by diverse spatial constructs and vendor-
limited software architectures. As our open interfaces make geospatial data and services discoverable and
accessible in the context of the World Wide Web, geospatial information becomes truly more useful to more
people and its potential for enabling progress and enlightenment in many domains of human activity can
be more fully realized.’

coding of characters (Box 3.1), and XML, the extensible

markup language that is used by many services on the
Web. Others are specific to GIS, and many of these have
been developed through the Open Geospatial Consortium
(OGC, www.opengeospatial.org), an organization set
up to promote openness and interoperability in GIS (see
Box 11.1). Among the many successes of OGC over the
past decade are the simple feature specification, which
standardizes many of the terms associated with primitive
elements of GIS databases (polygons, polylines, points,
etc.; see also Chapter 9); Geography Markup Language
(GML), a version of XML that handles geographic
features and enables open-format communication of
geographic data; and specifications for Web services
(Web Map Service, Web Feature Service, Web Coverage
Services) that allow users to request data automatically Figure 11.4 The power of this home computer would be
from remote servers. wasted at night when its owner is sleeping, so instead its power
Distributed GIS reinforces the notion that today’s has been ‘harvested’ by a remote server and used to process
computer is not simply the device on the desk, with signals from radio telescopes as part of a search for
its hard drive, processor, and peripherals, but some- extra-terrestrial intelligence
thing more extended. The slogan ‘The Network is the
Computer’ has provided a vision for at least one com- use of computer power wherever it can find it on the Inter-
pany – Sun Microsystems (www.sun.com) – and pro- net (Figure 11.4). Grid computing is a generic term for
pels many major developments in computing. The term such a fully integrated worldwide network of computers
cyberinfrastructure describes a new approach to the con- and data.
duct of science, relying on high-speed networks, mas-
sive processors, and distributed networks of sensors and
data archives (www.cise.nsf.gov/sci/reports/toc.cfm).
Efforts are being made to integrate the world’s computers
to provide the kinds of massive computing power that are 11.2 Distributing the data
needed by such projects as SETI (the Search for Extra-
Terrestrial Intelligence, www.seti.org), which processes
terabytes of data per day in the search for anomalies that Since its popularization in the early 1990s the Inter-
might indicate life elsewhere in the universe, and makes net has had a tremendous and far-reaching impact on
 CHAPTER 11 DISTRIBUTED GIS 245
the accessibility of GIS data, and on the ability of GIS provide useful information on the dataset’s contents. In
users to share datasets. As we saw in Chapter 9, a large the case of remotely sensed images, this may include
and increasing number of websites offer GIS data, for the percentage of cloud obscuring the scene, or whether
free, for sale, or for temporary use, and also provide the scene contains particularly useful instances of specific
services that allow users to search for datasets satis- phenomena, such as hurricanes.
fying certain requirements. In effect, we have gone in
a period of little more than ten years from a situa- Object-level metadata are formal descriptions of
tion in which geographic data were available only in datasets that satisfy many different requirements.
the form of printed maps from map libraries and retail-
OLM generalize and abstract the contents of datasets,
ers, to one in which petabytes (Table 1.1) of information
and therefore we would expect that they would be smaller
are available for download and use at electronic speed
in volume than the data they describe. In reality, however,
(about 1.5 million CDs would be required to store 1
petabyte). For example, the NASA-sponsored EOSDIS it is easy for the complete description of a dataset
(Earth Observing System Data and Information System; to generate a greater volume of information than the
spsosun.gsfc.nasa.gov/New EOSDIS.html) archives and
actual contents. OLM are also expensive to generate,
distributes the geographic data from the EOS series of because they represent a level of understanding of the
satellites, acquiring new data at over a terabyte per day, data that is difficult to assemble, and require a high
with an accumulated total of more than 1 petabyte at this level of professional expertise. Generation of OLM for
site alone. a geographic dataset can easily take much longer than it
takes to catalog a book, particularly if it has to deal with
Some GIS archives contain petabytes of data. technical issues such as the precise geographic coverage
of the dataset, its projection and datum details (Chapter 5),
The vision of distributed GIS goes well beyond the and other properties that may not be easily accessible.
ability to access and retrieve remotely located data, Thus the cost of OLM generation, and the incentives that
however, because it includes the concepts of search, motivate people to provide OLM, are important issues.
discovery, and assessment: in the world of distributed For metadata to be useful, it is essential that they
GIS, how do users search for data, discover their follow widely accepted standards. If two users are to be
existence at remote sites, and assess their fitness for use? able to share a dataset, they must both understand the
Three concepts are important in this respect: object-level rules used to create its OLM, so that the custodian of
metadata, geolibraries, and collection-level metadata. the dataset can first create the description, and so that the
potential user can understand it. The most widely used
standard for OLM is the US Federal Geographic Data
11.2.1 Object-level metadata Committee’s Content Standards for Digital Geospatial
Metadata, or CSDGM, first published in 1993 and now the
Strictly defined, metadata are data about data, and object- basis for many other standards worldwide. Box 11.2 lists
level metadata (OLM) describe the contents of a single some of its major features. As a content standard CSDGM
dataset by providing essential documentation. We need describes the items that should be in an OLM archive, but
information about data for many purposes, and OLM try does not prescribe exactly how they should be formatted
to satisfy them all. First, we need OLM to automate or structured. This allows developers to implement
the process of search and discovery over archives. In the standard in ways that suit their own software
that sense OLM are similar to a library’s catalog, which environments, but guarantees that one implementation
organizes the library’s contents by author, title, and will be understandable to another – in other words, that
subject, and makes it easy for a user to find a book. the implementations will be interoperable. For example,
But OLM are potentially much more powerful, because ESRI’s ArcGIS provides two formats for OLM, one using
a computer is more versatile than the traditional catalog the widely recognized XML standard and the other using
in its potential for re-sorting items by a large number of ESRI’s own format.
properties, going well beyond author, title, and subject, CSGDM was devised as a system for describing
and including geographic location. Second, we need geographic datasets, and most of its elements make
OLM to determine whether a dataset, once discovered, sense only for data that are accurately georeferenced and
will satisfy the user’s requirements – in other words, to represent the spatial variation of phenomena over the
assess the fitness of a dataset for a given use. Does it Earth’s surface. As such, its designers did not attempt
have sufficient spatial resolution, and acceptable quality? to place CSGDM within any wider framework. But in
Such metadata may include comments provided by others the past decade a number of more broadly based efforts
who tried to use the data, or contact information for have also been directed at the metadata problem, and at
such previous users (users often comment that the most the extension of traditional library cataloging in ways that
useful item of metadata is the phone number of the make sense in the evolving world of digital technology.
person who last tried to use the data). Third, OLM must One of the best known of these is the Dublin Core (see
provide the information needed to handle the dataset Box 11.3), the outcome of an effort to find the minimum
effectively. This may include technical specifications of set of properties needed to support search and discovery
format, or the names of software packages that are for datasets in general, not only geographic datasets.
compatible with the data, along with information about Dublin Core treats both space and time as instances
the dataset’s location, and its volume. Finally, OLM may of a single property, coverage, and unlike CSGDM
246 PART III TECHNIQUES

Technical Box 11.2

Major features of the US Federal Geographic Data Committee’s Content Standards

for Digital Geospatial Metadata

1. Identification Information – basic attributes, and the domains from which

information about the dataset. attribute values may be assigned.
2. Data Quality Information – a general 6. Distribution Information – information
assessment of the quality of the dataset. about the distributor of and options for
obtaining the dataset.
3. Spatial Data Organization
Information – the mechanism used to 7. Metadata Reference
represent spatial information in the Information – information on the
dataset. currentness of the metadata information,
and the responsible party.
4. Spatial Reference Information – the
8. Citation Information – the recommended
description of the reference frame for, and
reference to be used for the dataset.
the means to encode, coordinates in the
dataset. 9. Time Period Information – information
about the date and time of an event.
5. Entity and Attribute Information – details
about the information content of the 10. Contact Information – identity of, and
dataset, including the entity types, their means to communicate with, person(s) and
organization(s) associated with the dataset.

Technical Box 11.3

The fifteen basic elements of the Dublin Core metadata standard

1. TITLE. The name given to the resource by 7. DATE. The date the resource was made
the CREATOR or PUBLISHER. available in its present form.
2. AUTHOR or CREATOR. The person(s) or 8. RESOURCE TYPE. The category of the
organization(s) primarily responsible resource, such as home page, novel, poem,
for the intellectual content of the working paper, technical report, essay,
resource. dictionary.
3. SUBJECT or KEYWORDS. The topic of the 9. FORMAT. The data representation of the
resource, or keywords, phrases, or resource, such as text/html, ASCII, Postscript
classification descriptors that describe the file, executable application, or JPEG image.
subject or content of the resource. 10. RESOURCE IDENTIFIER. String or number
4. DESCRIPTION. A textual description of the used to uniquely identify the resource.
content of the resource, including abstracts 11. SOURCE. The work, either print or
in the case of document-like objects or electronic, from which this resource is
content description in the case of delivered, if applicable.
visual resources.
12. LANGUAGE. Language(s) of the intellectual
5. PUBLISHER. The entity responsible for content of the resource.
making the resource available in its present
13. RELATION. Relationship to other resources.
form, such as a publisher, a university
department, or a corporate entity. 14. COVERAGE. The spatial locations and
temporal duration characteristic of
6. OTHER CONTRIBUTORS. Person(s) or
the resource.
organization(s) in addition to those
specified in the CREATOR element who 15. RIGHTS MANAGEMENT. The content of this
have made significant intellectual element is intended to be a link (a URL or
contributions to the resource, but whose other suitable URI as appropriate) to a
contribution is secondary to the individuals copyright notice, a rights-management
or entities specified in the CREATOR statement, or perhaps a server that would
element. provide such information in a dynamic way.
 CHAPTER 11 DISTRIBUTED GIS 247
does not lay down how such specific properties as The term geolibrary has been coined to describe
spatial resolution, accuracy, projection, or datum should digital libraries that can be searched for information about
be described. any user-defined geographic location. A US National
The principle of establishing a minimum set of proper- Research Council report (see Further Reading) describes
ties is sharply distinct from the design of CSGDM, which the concept and its implementation, and many instances
was oriented more toward the capture of all knowable and of geolibraries can be found on the Web (see Box 11.4).
potentially important properties of geographic datasets. Of Digital Earth is a vision of a geolibrary in which the user
direct relevance here is the problem of cost, and specifi- interacts with a visual representation of the surface of
cally the cost of capturing a full CSGDM metadata record. the Earth, rather than with a digital representation of the
While many organizations have wanted to make their data traditional card catalog.
more widely available, and have been driven to create
A geolibrary can be searched for information
OLM for their datasets, the cost of determining the full set
of CSDGM elements is often highly discouraging. There about a specific geographic location.
is interest therefore in a concept of light metadata, a lim-
ited set of properties that is both comparatively cheap to 11.2.3 Collection-level metadata
capture, and still useful to support search and discovery.
Dublin Core represents this approach, and thus sits at the Thus far we have seen how metadata can be useful in
opposite end of a spectrum from CSGDM. Every orga- the description of individual datasets. There are several
nization must somehow determine where its needs lie on other forms of metadata of potential interest to GIS users,
this spectrum, which ranges from light and cheap to heavy and this section describes the issues associated with one
and expensive. of them – metadata that describe entire collections of
datasets, rather than individual datasets.
Light, or stripped-down, OLM provide a short but Many different collections or archives of geographic
useful description of a dataset that is cheaper datasets exist on the Internet, and Box 11.4 and Table 9.4
to create. identify several examples. The Geospatial One-Stop is
structured as a mechanism that allows custodians to make
their data visible to users who access the portal, and that
11.2.2 Geolibraries allows users to search across multiple servers using a sim-
ple protocol, assess the appropriateness of datasets, and
The use of digital technology to support search and access them through their own GIS. The US Geological
discovery opens up many options that were not avail- Survey supports several major collections of geographic
able in the earlier world of library catalogs and book- data, notably at the EROS Data Center in Sioux Falls,
shelves. Books must be placed in a library on a per- South Dakota (edcwww.cr.usgs.gov), and major com-
manent basis, and there is no possibility of reordering mercial collections of data can be found at the sites of
their sequence – but in a digital catalog it is possible to Space Imaging (www.spaceimaging.com), Terraserver
reorder the sequence of holdings in a collection almost (www.terraserver.com), GlobeXplorer (www.globex-
instantaneously. So while a library’s shelves are tradi- plorer.com), and at many other locations. Clearinghouses
tionally sorted by subject, it would be possible to re-sort are also sponsored by many other countries and by several
them digitally by author name or title, or by any property US states.
in the OLM catalog. Similarly the traditional card cata- But this abundance of readily accessible data presents
log allowed only three properties to be sorted – author, a problem – how does the user know which collection to
title, and subject – and discouraged sorting by multiple search for a given dataset? Although the Geospatial One-
subjects. But the digital catalog can support any number Stop hopes to be a single portal for a large proportion
of subjects. of geographic data, its emphasis is on the US, and no
Of particular relevance to GIS users is the possibility site, however successful, will ever have a monopoly of
of sorting a collection by the coverage properties: location all data for all parts of the world. The term collection-
and time. Both the spatial and temporal dimensions are level metadata (CLM) defines the information needed to
continuous, so it is impossible to capture them in a single make an intelligent choice, based on general knowledge of
property analogous to author that can then be sorted each collection’s contents. For example, the EROS Data
numerically or alphabetically. But in a digital system Center is a good site to search for images and maps of the
this is not a serious problem, and it is straightforward US. Microsoft’s Terraserver (terraserver.microsoft.com)
to capture the coverage of a dataset, and to allow the user is a good source for imagery of any part of the world. In
to search for datasets that cover an area or time of interest looking for maps of a specific country, it makes sense
defined by the user. Moreover, the properties of location to try to find a site sponsored by that country’s national
and time are not limited to geographic datasets, since mapping agency, and similarly, information about a state
there are many types of information that are associated may exist in a clearinghouse sponsored by that state.
with specific areas on the Earth’s surface, or with specific
Collection-level metadata describe the entire
time periods. Searching based on location or time would
enable users to find information about any place on the collection available through a single website.
Earth’s surface, or any time period – and to find reports, Unfortunately there has been little consistent effort
photographs, or even pieces of music, as long as they to develop standards for CLM, unlike its individual
possessed geographical and temporal footprints. dataset counterpart (OLM). Knowledge of CLM is still
248 PART III TECHNIQUES

Technical Box 11.4

Exemplar geolibraries
The Alexandria Digital Library (www.alexandria. distributed resources of geographic data.
ucsb.edu) provides access to over 2 million maps Managed by the Department of the Interior,
and images using a simple method of search it presents a catalog of datasets residing in
based on user interaction with a world map, federal, state, local, tribal, and private archives,
placenames, or coordinates. Many of the maps and allows users to search across these resources
and images are stored in the Map and Imagery independently of their storage locations.
Laboratory at the University of California, Its benefits include reduced duplication, by
Santa Barbara, but the Alexandria catalog encouraging users to share the data of others
also identifies materials in other collections. rather than creating them themselves, and
Figure 11.5 shows a screen shot of the library’s increased collaboration between agencies. Users
Web interface. can volunteer their own datasets by registering
The US Geospatial One-Stop (www.geo-one- (‘publishing’) them to the portal. Geoportals
stop.gov; Figure 11.6) is intended to provide can also act as directories to remotely accessible
a single portal (a geoportal) to vast and GIServices (Section 11.4).

(A)

Figure 11.5 Example of a geolibrary: the Alexandria Digital Library (alexandria.ucsb.edu). (A) shows the interface to the
library in a standard Netscape browser. The user has selected an area of California to search, and the system has returned
the first 50 items in the library whose metadata match the search area footprint and the user’s request for scanned images of
1:24 000 topographic maps. (B) shows one of the ‘hits’, part of the San Rafael quadrangle centered on Muir Woods
National Monument
 CHAPTER 11 DISTRIBUTED GIS 249

(B)

Figure 11.5 (continued)

Figure 11.6 The US Department of the Interior’s Geospatial One-Stop, another example of a geolibrary. In this case the
user has zoomed in to the area of Goleta, California, and requested data on transportation networks. A total of 66 ‘hits’
were identified in this search, indicating 66 possible sources of suitable information
250 PART III TECHNIQUES
largely unorganized, and tends to accumulate slowly in the
minds of geographic information specialists (or SAPs, see 11.3 The mobile user
Section 1.4.3.2). Knowing where to look is still largely a
matter of personal knowledge and luck.
Various possible solutions to this problem can be Computing has become so much a part of our lives that
identified. In the future, we may find the means to for many people it is difficult to imagine life without it.
develop a new generation of Web search engines that Increasingly, we need computers to shop, to communicate
are able to discover geographic datasets automatically, with friends, to obtain the latest news, and to entertain
and catalog their OLM. Then a user needing information ourselves. In the early days, the only place one could
on a particular area could simply access the search compute was in a computing center, within a few meters
of the central processor. Computing had extended to the
engine’s results, and be provided with a list of appropriate
office by the 1970s, and to the home by the 1990s. The
collections. That capability is well beyond the reach of
portable computers of the 1980s opened the possibility of
today’s generation of search engines, although several
computing in the garden, at the beach, or ‘on the road’ in
format standards, including GeoTIFF, already add the
airports and airplanes. Wireless communication services
tags to geographic datasets that clever search engines such as WiFi (the wireless access technology based on
could recognize. Alternatively, some scheme might be the 802.11 family of standards) now allow broadband
developed that authorizes a uniform system of servers, communication to the Internet from ‘hot-spots’ in many
such that each server’s contents are defined precisely, hotels, restaurants, airports, office buildings, and private
using geographic, thematic, or other criteria. In such homes (Figure 11.7). The range of mobile computing
a system all data about locations within the server’s devices is also multiplying rapidly, from the relatively
nominal coverage area, such as a state or a range cumbersome but powerful laptop weighing several kg to
of latitude and longitude, would be found on that the PDA, tablet computer, and the cellphone weighing a
server and no other. Such a system would need to be hundred or so grams. Within a few years we will likely
hierarchical, with very detailed data in collections with see the convergence of these devices into a single, low-
small coverage areas, and coarser data in collections weight, and powerful mobile personal device that acts as
with larger coverage areas. But at this time the CLM computer, storage device, and cellphone.
problem remains a major impediment to effective sharing In many ways the ultimate end point of this progression
of geographic data. is the wearable computer, a device that is fully embedded

Figure 11.7 Map of WiFi (802.11) wireless broadband ‘hotspots’ within 1 mile of the White House (1600 Pennsylvania Ave,
Washington, DC) and using the T-Mobile Internet provider
 CHAPTER 11 DISTRIBUTED GIS 251
as from actually being there. The expense and time of
traveling to Australia would be avoided, and the analysis
could proceed almost instantaneously. Some aspects of
the study area would be missing, of course – aspects of
culture, for example, that can best be experienced by
meeting with the local people.
Research environments such as this are termed virtual
realities, because they replace what humans normally
gather through their senses – sight, sound, touch, smell,
and taste – by presenting information from a database.
In most GIS applications only one of the senses, sight,
is used to create this virtual reality, or VR. In principle,
it is possible to record sounds and store them in GIS
as attributes of features, but in practice very little use is
made of any sensory channel other than vision. Moreover,
in most GIS applications the view presented to the user
is the view from above, even though our experience at
looking at the world from this perspective is limited (for
most of us, to times when we requested a window seat in
Figure 11.8 A wearable computer in use. The outfit consists an airplane). GIS has been criticized for what has been
of a processor and storage unit hung on the user’s waist belt; termed the God’s eye view by some writers, on the basis
an output unit clipped to the eyeglasses with a screen that it distances the researcher from the real conditions
approximately 1 cm across and VGA resolution; an input experienced by people on the ground.
device in the hand; and a GPS antenna on the shoulder. The
batteries are in a jacket pocket (Courtesy: Keith Clarke) Virtual environments attempt to place the user in
distant locations.
in the user’s clothing, goes everywhere, and provides More elaborate VR systems are capable of immersing
ubiquitous computing service. Such devices are already the user, by presenting the contents of a database in a
obtainable, though in fairly cumbersome and expensive three-dimensional environment, using special eyeglasses
form. They include a small box worn on the belt and or by projecting information onto walls surrounding the
containing the processor and storage, and an output user, and effectively transporting the user into the envi-
display clipped to the user’s eyeglasses. Figure 11.8 ronment represented in the database. Virtual London (Box
shows such a device in use. 13.7 and see www.casa.ucl.ac.uk/research/virtuallon-
Computing has moved from the computer center don.htm) and the Virtual Field Course (www.geog.le.
ac.uk/vfc/) are examples of projects to build databases
and is now close to, even inside, the human body.
to support near-immersion in particular environments.
The ultimate form of mobile computing is the Some of the most interesting of these are projects that
wearable computer. recreate historic environments, such as those of the Cul-
What possibilities do such systems create for GIS? Of tural VR Lab at the University of California, Los Ange-
greatest interest is the case where the user is situated in les (www.cvrlab.org), which supports roaming through
the subject location, that is, U is contained within S, and three-dimensional visualizations of the Roman church of
the GIS is used to analyze the immediate surroundings. Santa Maria Maggiore (Figure 11.9) and other classi-
A convenient way to think about the possibilities is by cal structures.
comparing virtual reality with augmented reality. Roger Downs, Professor of Geography at Pennsylvania
State University, uses Johannes Vermeer’s famous paint-
ing The Geographer (Figure 11.10) to make an important
point about virtual realities. On the table in front of the
11.3.1 Virtual reality and figure is a map, representing the geographer’s window on
augmented reality a part of the world that happens to be of interest. But
the subject figure is shown looking out of the window,
One of the great strengths of GIS is the window it provides at the real world, perhaps because he needs the informa-
on the world. A researcher in an office in Nairobi, Kenya tion he derives from his senses to understand the world
might use GIS to obtain and analyze data on the effects of as shown on the map. The idea of combining informa-
salinization in the Murray Basin of Australia, combining tion from a database with information derived directly
images from satellites with base topographic data, data through the senses is termed augmented reality, or AR.
on roads, soils, and population distributions, all obtained In terms of the locations of computing discussed earlier,
from different sources via the Internet. In doing so the AR is clearly of most value when the location of the user
researcher would build up a comprehensive picture of a U is contained within the subject area S, allowing the
part of the world he or she might never have visited, all user to augment what can be seen directly with informa-
through the medium of digital geographic data and GIS, tion retrieved about the same area from a database. This
and might learn almost as much through this GIS window might include historic information, or predictions about
252 PART III TECHNIQUES
(Figure 11.11), includes a differential GPS for accurate
positioning, a GIS database that includes very detailed
information on the immediate environment, a compass
to determine the position of the user’s head, and a pair
of earphones. The user gives the system information
to identify the desired destination, and the system then
generates sufficient information to replace the normal role
of sight. This might be through verbal instructions, or by
generating stereo sounds that appear to come from the
appropriate direction.

Augmented reality combines information from the

database with information from the senses.

Steven Feiner, Professor of Computer Science at

Columbia University, has demonstrated another form
Figure 11.9 Image generated from a three-dimensional digital of AR that superimposes historic images and other
reconstruction of the Basilica of Santa Maria Maggiore by the information directly on the user’s field of view. For
Virtual Reality Laboratory of the University of California, Los example, in Figures 11.12 and 11.13, a user wearing a
Angeles (www.cvrlab.org). The basilica was built in Rome head-mounted device coupled to a wearable computer
around 440 AD and has long since been destroyed (Reproduced is seeing both the Columbia University main library
by permission of Regents of the University of California building, and also an image generated from a database
showing the building that occupied the library’s position
prior to the University’s move to this site in 1896: the
Bloomingdale Insane Asylum.

Figure 11.10 Johannes Vermeer’s painting The Geographer

from 1669

the future, or information that is for some other reason

invisible to the user.
AR can be used to augment or replace the work
of senses that are impaired or absent. A team led
by Reginald Golledge, Professor of Geography at the Figure 11.11 The system worn here by Reg Golledge (a
University of California, Santa Barbara (see Box 1.8), leader of the development team) uses GIS and GPS to augment
has been experimenting with AR as a means of helping the senses of a visually impaired person navigating through a
visually impaired people to perform the simple task complex space such as a university campus (Reproduced by
of navigation. The system, which is worn by the user permission of Reg Golledge)
 CHAPTER 11 DISTRIBUTED GIS 253

11.3.2 Location-based services

One of the four big trends in software is

location-based applications.
Bill Gates, Wireless 2000 Conference, March 2000

A location-based service (LBS) is defined as an

information service provided by a device that knows
where it is and is capable of modifying the information it
provides based on that knowledge. Traditional computing
devices, such as desktops or laptops, have no way of
knowing where they are, and their functions are in no
way changed when they are moved. But increasingly
the essential information about a device’s location is
available, and is being used for a wide range of purposes,
some beneficial and some distinctly questionable (see
Box 5.2).

A location-based service is provided by a

computing device that knows where it is.

The simplest and most obvious form of locationally

enabled device is the GPS receiver, and any computer
that includes a GPS capability, such as a laptop or PDA
with an added PCMCIA card or chip, is capable of
providing LBS. But the most ubiquitous LBS-capable
Figure 11.12 An augmented reality system developed at device is the modern cellphone. A variety of methods
Columbia University by Prof. Steve Feiner and his group  exist for determining cellphone locations, including GPS
1999, T. Höllerer, S. Feiner & J. Pavlik, Computer Graphics & embedded in the cellphone itself, measurements made by
User Interfaces Lab., Columbia University. Reproduced with the cellphone of signals received from cellphone towers,
permission and measurements made by the towers of signals received

Figure 11.13 The field of view of the user of Feiner’s AR system, showing the Columbia University main library and the insane
asylum that occupied the site in the 1800s  1999, T. Höllerer, S. Feiner & J. Pavlik, Computer Graphics & User Interfaces Lab.,
Columbia University. Reproduced with permission
254 PART III TECHNIQUES
from the phone. As long as the phone is on, the operator Emergency services provide one of the strongest
of the cellphone network is able to pinpoint its location at motivations for LBS.
least to the accuracy represented by the size of the cell, of
There are many other examples of LBS that take
the order of 10 km, and frequently much more accurately.
advantage of locationally enabled cellphones. A yellow-
Within a few years it is likely that the locations of the vast page service responds to a user who requests information
majority of active phones will be known to an accuracy on businesses that are close to his or her current loca-
of 10 m. tion (Where is the nearest pizza restaurant? Where is the
One of the strongest motives driving this process is nearest hospital? Where is the nearest WiFi hotspot?) by
emergency response. A large and growing proportion of sending a request that includes the location to a suitable
emergency calls come from cellphones, and while the Web server. The response might consist of an ordered list
location of each land-line phone is likely to be recorded presented on the cellphone screen, or a simple map cen-
in a database available to the emergency responder, in a tered on the user’s current location (Figure 11.7). A trip
significant proportion of cases the user of a cellphone is planner gives the user the ability to find an optimum driv-
unable to report his or her current location to sufficient ing route from the current location to some defined desti-
accuracy to enable effective response. Several well- nation. Similar services are now being provided by public
publicized cases have drawn attention to the problem. transport operators, and in some cases these services make
The magazine Popular Science, for example, reported a use of GPS transponders on buses and trains to provide
case of a woman who lost control of her car in Florida information on actual, as distinct from scheduled, arrival
in February 2001, skidding into a canal. Although she and departure times. AT&T’s Find Friends service allows
called 911 (the emergency number standard in the US and a cellphone user to display a map showing the current
Canada), she was unable to report her location accurately locations of nearby friends (provided their cellphones are
and died before the car was found. active and they are also registered for this service). Under-
cover (www.playundercover.com) is an example of a
One solution to the 911 problem is to install GPS in
location-based game that involves the actual locations of
the vehicle, communicating location directly to the dis-
players (see Box 11.5) moving around a real environment.
patcher. The Onstar system (www.onstar.com) is one Direct determination of location, using GPS or mea-
such system, combining a GPS device and wireless com- surement to or from towers, is only one basis on which
munication. The system has been offered in the US to a computing device might know its location, however.
Cadillac purchasers for several years, and provides vari- Other forms of LBS are provided by fixed devices, and
ous advisory services in addition to emergency response, rely on the determination of location when the device
such as advice on local attractions and directions to local was installed. For example, many point-of-sale systems
services. When the vehicle’s airbags inflate, indicating a that are used by retailers record the location of the sale,
likely accident, the system automatically measures and combining it with other information about the buyer
radios its location to the dispatcher, who relays the nec- obtained by accessing credit-card or store-affinity-card
essary information to the emergency services. records (Figure 11.18). In exchange for the convenience

Biographical Box 11.5

Antonio Câmara, designer of location-based services

Antonio Câmara (Figure 11.14) is the CEO and co-founder of YDreams

(www.ydreams.com), a company specializing in location-based games and
novel applications of computer graphics. A Professor at the New University
of Lisbon in Portugal, he has been involved in GIS since 1977. Antonio
Câmara was a member of the team that developed SNIG, the first spatial
data infrastructure on the Internet, in 1995.
The mobile-phone-based entertainment market is expected to reach
between 8 and 25 billion Euros a year in revenues by 2008. Games where the
player’s location is known are becoming a prominent component of that
market. Undercover (www.playundercover.com), developed by YDreams, Figure 11.14 Antonio Câmara,
was the first location-based game with a visual interface. Undercover is a CEO and co-founder of YDreams, a
multi-player game where players can act solely or in a group. The purpose company specializing in
location-based games (Reproduced
is to eliminate terrorists, which are either virtual or result from an infection
by permission of Antonio Câmara)
acquired by a player. The positioning of players and terrorists can be seen
on the mobile-phone screen (Figure 11.15).
 CHAPTER 11 DISTRIBUTED GIS 255

Figure 11.15 Cellphone display for a user playing Undercover: scanning for friends (blue) and foes (red) (Reproduced by
permission of Antonio Câmara)

Figure 11.16 The Lex Ferrum location-based game developed by YDreams relies on the display of three-dimensional characters on
the cellphone screen (Reproduced by permission of Antonio Câmara)

Figure 11.17 A Formula 1 race played on a large screen using mobile phones. The project was developed by IDEO and YDreams
for Vodafone (Reproduced by permission of Antonio Câmara)
256 PART III TECHNIQUES

In the first version, Undercover relied on simplified maps where only main points of interest were
considered. Positioning was based on Cell-ID techniques provided by the operator. Undercover’s second
version will use city maps in the background. Positioning may use a range of systems including GPS units
communicating via Bluetooth with the mobile phones. YDreams has also developed a proximity-based
location game called Lex Ferrum (Figure 11.16) for the NOKIA N-Gage. In this game, the player’s location is
scanned via Bluetooth or GPRS/UMTS. Local combat is done using Bluetooth. Remote combat may use the
cellular network.
Such games will also benefit from the widespread deployment of large screens in cities. Mobile phones can
then be used as controllers of such screens. Imagine a user visualizing a three-dimensional representation of
the city, querying the underlying GIS and playing games such as an imaginary Formula 1 race (Figure 11.17).

information that may appear on Web pages. Companies

such as InfoSplit (www.infosplit.com; Figure 11.19)
specialize in gathering and providing lists of such
locations to customers. A search-engine company such
as Google would be interested in knowing the locations
of computers that send it queries, because it would
then be possible to sort responses by distance from the
computer – ranking in this way would be of commercial
interest to retailers, for example, because it would make
their sites more conspicuous to local customers.

11.3.3 Issues in mobile GIS

GIS in the field or ‘on the road’ is very different from GIS
in the office. First, the location of the user is important,
and directly relevant to the application. It makes good
sense to center maps on the user’s location, to provide
the capability for maps that show the view from the
user’s location rather than from above, and to offer
maps that are oriented to the user’s direction of travel,
Figure 11.18 Credit card in use (Reproduced with permission rather than north. Second, the field environment may
of PhotoDisc, Inc.)
make certain kinds of interaction impractical, or less
desirable. In a moving vehicle, for example, it would
of a credit card, the user effectively surrenders some be dangerous to present the driver with visual displays,
degree of location privacy to the company whenever the unless perhaps these are directly superimposed on the field
card is used. One benefit is that it is possible for the of view (on the windshield) – instead, such systems often
company to analyze transactions, looking for patterns of provide instructions through computer-generated speech,
purchase that are outside the card user’s normal buy- and use speech recognition to receive instructions. With
ing habits, perhaps because they occur in locations that wearable devices that provide output on minute screens
the user does not normally frequent, or at anomalous attached to the user’s eyeglasses there is no prospect of
times. Many of us will have experienced the embar- conventional point-and-click with a mouse, so again voice
rassment of having a credit card transaction refused in communication may be more appropriate. On the other
an unfamiliar city, because the techniques used by the hand, many environments are noisy, creating problems
company have flagged the transaction as an indicator for voice recognition.
the card might have been stolen. In principle, a store- One of the most important limitations to mobility
affinity card gives the company access to information remains the battery, and although great strides have been
about buying habits and locations, in return for a modest made in recent years, battery technology has not advanced
discount. as rapidly as other components of mobile systems, such as
processors and storage devices. Batteries typically account
Location is revealed every time a credit, debit, or for the majority of a mobile system’s weight, and limit
store affinity card is used. its operating time.

Another technique relies on the locations given when The battery remains the major limitation to LBS
an Internet IP address is registered, augmented by location and to mobile computing in general.
 CHAPTER 11 DISTRIBUTED GIS 257

Figure 11.19 Companies such as InfoSplit (www.infosplit.com) specialize in determining the locations of computers from their IP
addresses. In this case the server has correctly determined that the ‘hit’ comes from Santa Barbara, California

Although broad band wireless communication is

possible using WiFi, connectivity remains a major issue. 11.4 Distributing the software:
Wireless communication techniques tend to be:
GIServices
■ Limited in spatial coverage. WiFi hotspots are limited
to a single building or perhaps a hundred meters from
the router; cellphone-based techniques have wider This final section addresses distributed processing, the
coverage but lower bandwidth (communication notion that the actual operations of GIS might be
speeds), and only the comparatively slow provided from remote sites, rather than by the user’s
satellite-based systems approach global coverage. own computer. Despite the move to component-based
software (Section 7.3.3), it is still true that almost all of
■ Noisy. Cellphone and WiFi communications tend to the operations performed on a user’s data are executed
‘break up’ at the edges of coverage areas, or as in the same computer, the one on the user’s desk. Each
devices move between cells, leading to errors in copy of a popular GIS includes the same functions, which
communication. are replicated in every installation of that GIS around
■ Limited in temporal coverage. A moving device is the world. When new versions are released, they must
likely to lose signal from time to time, while some be copied and installed at each site, and because not
devices, particularly recorders installed in commercial all copies are replaced there is no guarantee that the
vehicles and PDAs used for surveys, are unable to GIS used by one user is identical to the GIS used by
upload data until within range of a home system. another, even though the vendor and product name are
■ Insecure. Wireless communications often lack the same.
adequate security to prevent unwanted intrusion. It is A GIService is defined as a program executed at a
comparatively easy, for example, for someone to tap remote site that performs some specific GIS task (see also
into a wireless session and obtain sensitive Section 7.2). The execution of the program is initiated
information. remotely by the user, who may have supplied data, or may
rely on data provided by the service, or both. A simple
Progress is being made on all of these fronts, but it will example of a GIService is that provided by wayfinding
be some years before it becomes possible to ‘do GIS’ as sites like MapQuest (www.mapquest.com) or Yell.com
efficiently, effectively, and safely anywhere in the field as (see Section 1.4.3.1). The user’s current location and
it currently is in the office. desired destination are provided to the service (the
258 PART III TECHNIQUES
current location may be entered by the user, or provided and other sites. In general, however, the characteristics
automatically from GPS), but the data representing the that make a GIS function suitable for offering as a service
travel network are provided by the GIService. The results appear to be:
are obtained by solving for the shortest path between
the origin and the destination (often a compromise ■ Reliance on a database that must be updated
between minimizing distance and minimizing expected frequently, and is too expensive for the average user
travel time), a function that exists in many GIS but in to acquire. Both geocoding (Box 5.3) and wayfinding
this case is provided remotely by the GIService. Finally, services fall into this category, as do gazetteer
the results are returned to the user in the form of driving services (Section 5.9).
instructions, a map, or both.
■ Reliance on GIS operations that are complex and can
A GIService replaces a local GIS function with one be performed better by a specialized service than by a
provided remotely by a server.
generic GIS.

In principle, any GIS function could be provided in The number of available GIServices is growing
this way, based on GIS server software (Section 7.6.2). steadily, creating a need for directories, portals, and other
In practice, however, certain functions tend to have mechanisms to help users find and access them, and
attracted more attention than others. One obvious problem standards and protocols for interacting with them. The
is commercial: how would a GIService pay for itself, Geography Network (www.geographynetwork.com;
would it charge for each transaction, and how would Figure 11.20) provides such a directory, in addition to
this compare to the normal sources of income for GIS its role as a geolibrary, as does the Geospatial One-
vendors based on software sales? Some services are Stop (Box 11.4). Generic standards for remote services
offered free, and generate their revenue by sales of are emerging, such as WSDL (Web Services Definition
advertising space or by offering the service as an add- Language, www.w3.org/TR/wsdl) and UDDI (Universal
on to some other service – MapQuest and Yell are good Description, Discovery, and Integration, www.uddi.org),
examples, generating much of their revenue through direct though to date these have found little application
advertising and through embedding their service in hotel in GIS.

Figure 11.20 The Geography Network provides a directory of remote GIServices. In this case a search for services related to
transportation networks has identified four
 CHAPTER 11 DISTRIBUTED GIS 259
devices in field settings; limitations placed on communi-
11.5 Prospects cation bandwidth and reliability; and limitations inherent
in battery technology. Perhaps, at this time, more prob-
lematic than any of these is the difficulty of imagining the
Distributed GIS offers enormous advantages, in reducing full potential of distributed GIS. We are used to associ-
duplication of effort, allowing users to access remotely ating GIS with the desktop, and conscious that we have
located data and services through simple devices, and pro- not fully exploited its potential – so it is hard to imagine
viding ways of combining information gathered through what might be possible when the GIS can be carried any-
the senses with information provided from digital sources. where and its information combined with the window on
Many issues continue to impede progress, however: com- the world provided by our senses.
plications resulting from the difficulties of interacting with

Questions for further study Further reading

1. Design a location-based game based on Câmara A.S. 2002 Environmental Systems: A Multidimen-
GPS-enabled cellphones. sional Approach. New York: Oxford University Press.
National Research Council 1999 Distributed Geolibraries:
2. Find a selection of geolibraries on the Web, and Spatial Information Resources. Washington, DC:
identify their common characteristics. How does each National Academy Press.
of them: 1) allow the user to specify locations of Peng Z.-H. and Tsou M.-H. 2003 Internet GIS: Dis-
interest; 2) allow the user to browse through datasets
tributed Geographic Information Services for the Inter-
that have similar characteristics; and 3) allow the user
net and Wireless Networks. Hoboken, NJ: Wiley.
to examine the contents of datasets before
acquiring them?
3. To what extent do citizens have a right to locational
privacy? What laws and regulations control the use of
locational data on individuals?
4. Is an LBS a type of GIS, or is a GIS a type of LBS,
or are they quite different forms of computer
application? If they are different, are they likely to
merge in the future, and what will the result
look like?