Gis and Cartography
Gis and Cartography
Gis and Cartography
Introduction
Map-making is an ancient field dating from the first efforts by humans to create pictorial
representations of the world around them, by drawing on the walls or mud floors of
caves. Cartography, the modern discipline of map design, compilation, and publication, is
most often associated with the paper map, a flattened representation of the curved surface
of the Earth, but it may also refer to the creation of globes, and increasingly it refers to
the use of digital computers to manage the acquisition, manipulation, and eventual
display of geographic information on the screens of computers. The advent of digital
technology in the 1960s began a transition in the field of cartography that continues
today.
Geographic information systems (GIS) are computer applications concerned with the
manipulation of geographic information, and today these software packages are capable
of the representation, analysis, and visualization of virtually any form of information
about the distribution of features and phenomena on the surface of the Earth. The initial
developments of GIS occurred in the 1950s and 1960s, as computers became powerful
enough to manage the large volumes of data being collected from satellites or digitized
from paper maps, and today virtually all geographic information is in digital form, stored
in a computer, at some point in its life.
These definitions clearly overlap substantially, and while there might have been a period
in the 1960s and 1970s when it was possible to distinguish between a cartography
focused on the paper medium and a GIS focused on digital systems, today the distinction
is hopelessly blurred. Many cartographers now prefer the term geovisualization, and
many GIS professionals consider themselves expert in the design of maps. The following
sections explore the many dimensions of the relationship between cartography and GIS,
and end with a brief speculation on the relationships future.
allowed users to interact with maps in both visual and auditory senses. Thus one could
explore a digital representation of a wilderness area, zooming to greater detail or
accessing textual information associated with specific features, and even hearing the
sounds associated with places and their animal inhabitants. Many researchers found the
traditional term cartography too limiting as a description of this new, much richer world,
and began to describe their field as geovisualization.
Today the effects of this transition can perhaps be seen most clearly in the various virtual
globe services that are now available, typified by Google Earth. These services provide
access to vast stores of imagery and digital map data through a simple interface that a
child of ten can learn in a few minutes. The user is able to visualize geography as if
looking at and manipulating a globe, with none of the distortions that are introduced
when the Earth is flattened for a paper map. The virtual globe can be manipulated by
rotating and zooming, from resolutions of 10 km to less than a meter. Information can be
merged from a vast range of sources using the techniques now known popularly as
mashups, and these extend to three-dimensional representations of buildings and
structures. The users viewpoint can be shifted from vertically above to oblique, and it is
possible to simulate a flight over the surface.
Google Earth is in many ways more akin to the traditions of map-making than to GIS.
Little of the analytic power of the latter is present; instead the emphasis is on visual
representation of the Earths surface, and an important measure of the service is the
degree to which its results resemble the real thing. By emphasizing imagery, and
allowing the user to overlay other information on an imagery base, one ensures that the
result is familiar, recognizable, and readily understood. By contrast displays generated
using GIS tend to be highly abstracted, and while the information they present will likely
be precise and in most cases accurate, the GIS lens clearly provides a more analytic and
less intuitive view of the world. Indeed, it is common for researchers to combine the two
technologies, using GIS to analyze patterns, search for anomalies, test hypotheses, and
compute numerical results, and using Google Earth to provide additional contextual
information that may not be present in the highly abstracted GIS database.
GIS as threat
Google Earth abandons many of the cherished icons of cartography such as projections,
rendering them largely irrelevant, and in this sense can be seen as threatening to a
disciplines cherished expertise. Similarly GIS enables anyone equipped with data and a
few simple tools to produce a map that previously would have required the expertise of a
trained cartographer. While great efforts have been made by developers of the more
elaborate GIS packages to include support for sophisticated cartographic techniques, such
as methods of cartographic generalization and alternative techniques for assigning class
intervals to choropleth maps, nevertheless it is easy to find examples of the misuse of
GIS. For example, several recent news stories of the missile threat of North Korea have
included maps purporting to show the areas reachable by missiles of a given range
launched from Pyongyang by drawing concentric circles on a Mercator projection. Any
cartographer, and hopefully most trained GIS users, would know that the scale of the
Mercator projection changes rapidly at high latitudes, and that on this projection the locus
of equal distance from Pyongyang is only a circle when the distance is vanishingly small.
Unfortunately the result is a severe underestimate of the areas that can be reached (Figure
1).
[Figure 1 about here]
It is sometimes argued, therefore, that GIS, and the popularization of digital geographic
information technologies in general, represents a threat to the field of cartography that
GIS is killing cartography. Map-making, it is argued, is a sophisticated pursuit that is
best left in the hands of experts. Maps are often persuasive, capable of influencing
opinion and policy; and just as the playing of concert pianos is limited to a few experts,
so the creation of maps should be limited to trained professionals. Attempts have been
made in several jurisdictions to restrict mapping practice to accredited professionals, in
some cases with success.
Another dimension to this argument surfaced in the early 1990s as part of a rapidly
emerging social critique of GIS. In the late 1980s Brian Harley had introduced the
concepts of deconstruction to cartography, arguing that all maps were social
constructions that could be read as texts for evidence about the agendas of their makers.
These arguments found ready acceptance, since maps have always been important as
tools of power, and map-makers have often followed quickly on the heels of conquerers.
The selection of features for maps, and the styles in which selected features are rendered,
are part of the process of cartographic design, and clearly open to conscious or
subconscious manipulation for purposes that range from visual clarity to more sinister
coercion.
At the time, cartographers saw GIS as a new and wildly popular tool that appeared
insensitive to these ethical arguments. GIS seemed to be grounded in the nave
assumption that one could achieve a scientifically rigorous description of the world, and
store it in precise form in a digital computer that the contents of a GIS database
represented the results of objective, replicable scientific measurement. The ability to
compute measures such as area to large numbers of decimal places served to reinforce
this view, and it was clear that GIS was being marketed by commercial software
developers as a scientifically rigorous approach to geographic problems. To
cartographers influenced by Harley, GIS users were the barbarians, sensitized to none of
the nuances of mapping practice or to the tension between cartography as science and
cartography as art.
These critiques came to a head in the early 1990s, and led to a series of meetings in
which each side slowly achieved an understanding of the others position. Today, the
social context of GIS is one of the major themes of GIS research, and many
developments have addressed the issues raised by Harley and others. The notion of
objective truth, reflected in the use of such terms as accuracy and error, has been replaced
by concepts of uncertainty and of the relationship between truth and power, in the
realization that many aspects of GIS practice are not replicable, that many of the key
definitions are inherently vague, and that seemingly objective technologies can be
molded to the agendas of their owners and sponsors.
GIS as science
GIS is an attractive computer application, combining the visual and the numeric, and it is
not surprising therefore that it has captured the attention of many professionals with
backgrounds in science and engineering. Cartography may indeed be part art and part
science, but to these scientifically minded GIS users the new world of GIS offers the
opportunity to escape the constraints and vagueness of the past, by developing new ways
of describing the world that are indeed replicable and objective. Science has always
attempted to sweep away the vague and subjective, replacing terms such as cold, warm,
and hot with replicable measurements on standard scales why not, therefore, replace the
vague and partially subjective classifications of a soil map with measures of soil
properties that are both replicable and useful to farmers? From this perspective the power
of GIS lies in its ability to improve on past practice, rather than to replicate it in a digital
environment.
The transition to digital technology had both technical and economic impacts on mapmaking. On the technical side, it allowed entirely new approaches to map design that
escaped the constraints of manual cartography. For example, line widths could be varied
continuously, rather than being constrained by the width of a nib or scribe tool. Shading
and color could also be varied continuously, leading to vastly more possibilities in the
portrayal of spatially continuous phenomena such as topography or temperature. In short,
the new technology helped to underscore for the first time the degree to which
cartographic practice had been constrained by the technology of manual map-making
by the marks a human could make with a pen and ink.
Nowhere is this more evident than in the debates over interval-free choropleth maps.
These maps are used to portray statistical information, such as population density or
average income, for irregularly shaped reporting zones such as countries or postal-code
areas. The attributes of each area are shown using color or shading, each color or shade
corresponding to a defined part of the range of observed values. The relationship between
value and color or shade is usually shown in a legend on the map, and the set of possible
shades or colors is usually limited to less than ten, reflecting the difficulty of generating
many different shades or colors with traditional tools. There is a vast literature in
cartography on alternative methods of partitioning the range of values of defining what
are often known as class intervals.
In the late 1970s Waldo Tobler had suggested that in the digital world this use of a finite
set was no longer necessary, and that one could instead establish a continuous
relationship between observed value and color or shading density in what he termed
interval-free choropleth mapping. This would have the advantage that the information
lost during classification would be preserved, since it would no longer be necessary to
lump a range of values into each class, and as a result the map would show a more
accurate picture of the data (note, however, that choropleth maps also distort the spatial
aspects of data by averaging or lumping spatially continuous phenomena within discrete,
irregularly shaped areas). Accuracy and the avoidance of information loss are clearly
close to the heart of scientifically trained professionals.
The response from the cartographic community was overwhelmingly negative, and today
it is difficult to find mention of this technique in textbooks, or implementations in
software (ESRIs ArcGIS, the leading GIS software, supports many strategies for
defining class intervals but does not support interval-free choropleth mapping, or allow
the number of classes to exceed 32). Many cartographers argued that the assignment of
class intervals was a vital part of the cartographers art, and an essential tool in using the
map to tell his or her chosen story about the data (Figure 2).
[Figure 2 about here]
Today there continues to be a strong association between GIS and science. The term
geographic information science was coined in 1992, and serves two somewhat distinct
purposes: first, to describe the set of fundamental scientific questions and principles that
underlie GIS, and second, to describe the use of GIS in a scientifically rigorous context.
Cartography as a discipline clearly overlaps strongly with both interpretations of
GIScience, but while there have been efforts to emphasize the scientific aspects of
cartography, most notably in the analytic cartography pioneered by Waldo Tobler, the
artistic underpinnings of the field remain as important as ever.
Current research
Both cartography and GIScience have developed and published strongly overlapping
research agendas in recent years, and a comparison makes interesting reading, for both
the similarities and the differences. Both emphasize representation, and the need for
continuing research on how to extend existing methods to capture phenomena that are
complex, dynamic, and three-dimensional. Automation of the map-making process, and
particularly the thorny problem of automatic generalization, remains a concern of
cartographers, as does the design of user interfaces, particularly in devices such as mobile
phones that offer only limited area for visual display, and the role that cognitive science
can play in helping to address such issues. Both fields are concerned about uncertainty,
and how to convey what is known about it to the user; and both fields are concerned
about the rapidity of technological change, and the abilities of researchers and developers
to exploit innovation.
software engineering) tools, have made it possible to represent phenomena in GIS that
were never suitable topics for map-making, including events and transactions, flows, and
fine-resolution time series.
[Figure 3 about here]
At the same time maps and visual displays remain a central part of all geographic
information technologies, and provide much of the attraction that draws new recruits to
the field of GIS. Maps are in many ways visual trophies that can be incorporated to great
effect in presentations and posters, and hung on walls for purposes that go far beyond the
objective representation of information. GIS is still explained to the uninitiated using the
metaphor of maps, and students are still judged in GIS classes in part on the visual
quality of the maps they produce. Students are trained to be sensitive to the degree to
which maps can mislead and even lie, and to the importance of careful visual design in
telling a desired story.
The legacy of Brian Harley lives on in the general perception that cartography is the more
artistic, subjective side of geographic information technology, and GIS the more
scientific, objective side. Scientists using computers to make maps are more likely to call
what they do GIS than cartography, and more likely to assume that the information they
work with is the result of scientific observation and measurement.
To the outside world, however, these nuances are largely invisible. The general public is
by now well aware that important things are happening in geography, whether it be as a
result of a GPS-enabled navigation system in a car, or exploring the world through
Google Earth, or geotagging a photograph. In this world the concepts and principles in
which both cartographers and GIS professionals are trained are largely invisible, and it is
possible for entire books to be written on the subject of map-making with Google Maps
that make not a single reference to the literature of either field. Formal training in what it
means to think and express oneself spatially remains the preserve of a few, in contrast to
formal training in thinking numerically or in the use of language.
Further reading
Clarke, K. C. (1990). Analytical and computer cartography. Englewood Cliffs, NJ:
Prentice Hall.
Clarke, K. C. (2001). Getting started with geographic information systems. Upper Saddle
River, NJ: Prentice Hall.
Dykes, J., MacEachren, A. M., and Kraak, M.-J. (2005) Exploring geovisualization.
Amsterdam: Elsevier.
Erle, S., Gibson, R., and Walsh, J. (2005). Mapping hacks: tips and tools for electronic
cartography. Sebastopol, CA: OReilly Media.
Goodchild, M. F. (1988). Stepping over the line: technological constraints and the new
cartography. American Cartographer 15: 311-319.
Jones, C. B. (1997). Geographical information systems and computer cartography.
Harlow, UK: Longman.
MacEachren, A. M. and Kraak, M. -J. (2001). Research challenges in geovisualization.
Cartography and Geographic Information Science 28(1): 3-12.
McMaster, R. B. and Usery, E. L. (eds.) (2005). A research agenda for geographic
information science. Boca Raton, FL: CRC Press.
Slocum, T. (2003). Thematic cartography and geographic visualization. Upper Saddle
River, New Jersey: Prentice Hall.
Figure captions
1. A circle centered on Pyongyang and superimposed on a Mercator projection. Similar
maps appeared in several newspapers and magazines to illustrate the threat of North
Korean missiles with 10,000 km range. However the locus of points 10,000 km from
Pyongyang is not a circle on the Mercator projection, and includes the North Pole and
much of the continental US.
2. Two choropleth maps of Milwaukee census tracts, showing percent black. (A) uses
five classes following the Jenks natural breaks method of class interval selection. (B) uses
32 classes, and approaches the Tobler interval-free technique.
3. An analysis of the relationship between vegetation cover (each green area is a different
vegetation cover type) and elevation (the black and white image). The results of such a
GIS analysis are more likely to be presented in numeric form in a table (the column
headed MEAN shows the average elevation for each SPECIES).