An Overview of Geographic Information Systems and Science

An Overview of Geographic Information Systems and Science
A group of engineers look out across a large field, trying to imagine
the roads, infrastructure, and buildings that will be on the site in less
than a year. The locations of everything from powerlines to plants to
streetsigns need to be planned by a diverse group of organizations,
and then built in a timely fashion by an even more diverse group.
Not only must the design, work, and ultimately finished community
conform to land parcels, zoning, and aesthetics, but features such as
sewers have design restrictions regarding slope, placement relative
to water supply lines, and spacing of accessways.
The engineers turn to a laptop running Geographic Information
Systems software. On-screen they can see the existing topography
and neighboring infrastructure. They have access to high resolution
air photographs, survey information, and the existing parcel fabric
for the area. The analysis toolkit in the GIS supports network analysis including slope and routing restrictions, and furthermore supports distance constraints, overlap analysis, and support for links to
GANTT charts and other organizational aids.
The down-side of the GIS tools is that they are complex, and like
many specialist tools, jargon-laden.
This short note is about these tools, what they do, and what some
of the GIS jargon means!
Introduction
Although we don’t necessarily think about it all that often, humans live and breathe in space and time. It makes sense then that
many of the decisions we make in our daily lives are profoundly
influenced by where and when issues.
Many of the tools we surround ourselves with are fundamentally
about space and time. Telephones bridge communications across
space. Voice mail allows us to leave messages across time. The
Internet removes the idea of space from the idea of computing.
The Web turns a network of computers into a shared document
library. Despite this, many tools that support decision making do
a very poor job of including space and to some extent time in the
process of analysis: for example, spreadsheets such as Excel make
simplistic assumptions about the nature of earth locations that
invalidate some statistical approaches.
Rob Harrap
Queen’s University
[email protected]
ArcGIS, a modern Geographic Information System.
(Source: www.esri.com)
Geographic Information Systems are tools that support the collection, analysis, and representation of spatial information, and to a
lesser extent, temporal information. They extend the spreadsheet
and database idea from computer applications to the ‘real-world’
by including the shape of the Earth, spatial relationships, and
spatially accurate statistics in the analytical ‘toolbox’ available to
decision makers.
Geographic Information Science is the collection of principles
and techniques that support data collection, preparation, analysis and presentation. More than most other scientific disciplines,
GIScience as it is sometimes called is inclusive - we include whatever we need for our work inside our discipline, happily absorbing
approaches from across the scientific and artistic spectrum. This
reflects the profoundly artistic nature of early maps, which were
simultaneously works of great beauty and storehouses of hardearned geographic knowledge.
History
Since large parts of GIScience make use of maps to show results
- in fact, to many users, the map is the whole point - one could
argue that GIS originates in early attempts to map the Earth and
the Heavens.
Many of the techniques and ideas used in GIScience, such as the
principles of geometric surveying and the latitude-longitude coordinate systems, have their origin in ocean-voyage navigation, and
in early scientific attempts to determine the shape, size, and mass
of the Earth.
Blaeu’s map of Cyprus
For many years the ultimate realization of geographic knowledge was ornate maps such as the one shown at left. By the 20th
Century, however, simply managing spatial knowledge was a
major problem far exceeding simple cartography. Increasingly
commercial and military interests were concerned with what was
where and would be transported when. A number of fields sprang
up around the handling of such information, but until the 1960’s
these efforts tended to be non-geographic: space wasn’t treated as
an especially important variable.
The first true GIS tools were developed to manage forest information by Roger Tomlinson, a Canadian Geographer, in the early
1960’s. By the late 1970’s the idea of using computers to support
spatial decision making was no longer particularly outrageous,
and groups such as the Harvard Graphics Lab were developing
interactive mapping tools using state of the art Vector-Graphics
display tools. Research projects sprang up worldwide and the
first commercial software packages became available in the early
1980’s. At first, these tened to have simplistic functionality aimed
at simply drawing points and lines and attaching descriptive
information to these, but by the mid 1980’s GIS tools capable of
supporting surveying, spatial analysis, and sophisticated cartography were available.
Early GIS tools emphasized vector features - points, lines, and
polygons - and had only limited support for imagery such as
photographs and remote sensed imagery. This reflected in a large
part the origins of GIS tools in cartography, while remote sensing
tools - which were correspondingly weak in handling vector data
but handled imagery quite well - originated in the photogrammetry and satellite intelligence communities. By the early 1990’s
the widespread availability of powerful computers using the
familiar mouse-and-menu based ‘desktop metaphor’ meant that
it was practical to have tools capable of integrating GIS and RS
approaches. The first truly comprehensive GIS/RS tools became
available in the mid-1990’s, although there are still degrees of
specialization in products at the advanced level.
GIS researchers steal ideas from other disciplines constantly, and
so it is likely that the future history of GIS will be rich with innovations and spatial analysis tools we can hardly imagine. Five
years ago it would have been hard to imagine a free 3 dimensional
GIS providing global visualization tools across the Internet - but
Google Earth does just that. With the widespread availability of
GIS tools, the vibrancy of the open source development community, and ever more sophisticated spatial sensors, the future of
GIScience and GIS tools is assured.
Google Earth - Three-dimensional GIS visualization, free
on the Internet.
Some Fundamental Concepts
GIScience is jargon-rich, just like other scientific and technologic
fields. This can be a barrier to understanding even the most basic
descriptions of what programs do and how. Here are some fundamental GIS concepts.
The Entity-Attribute Model: In GIS it is common to represent
spatial features (say, a house or a bridge) on a map using graphics,
but to store descriptive information about the feature in a database
as a row in a table. The spatial feature is called an entity, and the
descriptive entries are each called attributes. To understand what
is ‘in’ a spatial dataset, as a result, one must examine both the
spatial and attribute content.
Vector Data Model: The vector model describes the world using points, lines, and polygons. Any one dataset will be of one
type - say a set of lines representing roads - and will have associated attributes stored. The vector model is compact and efficient
The Vector Data Model represents spatial features as
simple geometric primitives. Given enough line segments, realistic spatial representation can be achieved.
for cartography, but is not well suited to phenomena that vary
smoothly in space (like air pollution or wind speed).
Raster Data Model: The raster model represents the world as an
array of cells, much like a checkerboard overlaid on the Earth. As
such it has a constant spatial resolution - all features have to be
generalized to the size of a ‘cell,’ and this might significantly affect what can be stored. Raster datasets are analytically powerful,
useful for phenomena that vary smoothly in space, but are much
larger than typical vector datasets.
The Raster Data Model uses a regular array of cells to
cover geographic space. It is geometrically simple and
powerful for analysis.
Map Projection: The almost-round Earth cannot be simply
squashed onto a flat piece of paper (or, equivalently, a flat computer screen) without significant distortion. Map projections take
spherical coordinates (latitudes, longitudes) and transform them
mathematically into Cartesian coordinates (with orthogonal x,y
axes) so that world features can be viewed on plane. This always
involves, but seeks to minimize, distortion of shape, area, and
angular relationships.
GPS: Physical devices that, coupled with a system of satellites
orbiting the Earth, provide very accurate (metre-scale) surface positioning. Field data collection devices (such as enhanced PDA’s
and Tablet computers) typically link to GPS’ to get locations onthe-fly during data capture.
Spatial Analysis: These techniques use spatial relationships within
and between GIS datasets to answer questions such as: what
is near what; what overlaps what; how do attribute vary across
space; what is the density of a phenomena, and so on. There are
many such approaches, ranging from some that are strongly related to logic to those that are profoundly statistical.
What is near what? Distance ‘buffers’ around roads and
interpolation of population allow siting of hospitals to be
guided by spatial analysis.
The references listed at the end provide much more information
on these and related topics in GIScience.
Background to GIScience
As mentioned above, GIScientists are always on the lookout for
techniques to steal from other areas of study. GIScientists return
the favor by placing topics that have traditionally ignored spatial
relationships in a spatial and temporal context and thus support
advances across many subject areas. For example, GIScience took
much from ecology, but many modern ecological approaches are
based on GIScience ideas about spatial analysis and would be
impractical without the automation of GIS tools. A few areas that
GIScience has built upon are:
Geography: Of course, GIScience is profoundly about geography.
Most GIS ‘schools’ are within schools of Geography.
Mathematics and Statistics: Spatial analysis and the fundamentals of vector datasets are built upon historical work in geometry,
graph theory, and statistics. Most GIScientists are thoroughly
grounded in statistics and in particular, spatial statistics.
Computer Science: Most GIScience techniques are implemented
on computers as GIS programs or toolkits. Ideas in automata
theory, heuristics, artificial intelligence, and computer graphics
profoundly influence GIS tools and techniques.
Cognitive Science: Ultimately people must use the tools and
results of analysis. GIScientists are strongly connected to psychology and neuroscience through a need to understand spatial perception, cognitive limitations, and how to balance human-machine
symbiosis.
Philosophy and Linguistics: Attributes ultimately describe the
world in numeric or code-word form. What do these refer to?
What are the limits to knowledge? How can classification be
made more intuitive and more accurate? What does that mean?
Many advanced landscape visualizations such as this combine ideas from computer graphics with GIS techniques to
build virtual sets for movies, possible worlds for training,
and so on.
Information and Library Science: Ultimately many GIScientists
are spatial librarians, keeping track of what is where, when. Many
ideas from library science have guided GIScience work on data
collections and indexing just as they have profoundly influenced
the development of the World Wide Web. Remember, Google is a
library science application!
GIS in Use
Where do GIS tools get used? A few examples illustrate the range
of applications:
Siting Studies: Where should the next store in a chain be located?
Where should the base of operations for a military operation be
sited? Where do observed phenomena coincide? Even the placement of individual stores within shopping malls can be studied
through on-site surveys, spatial statistics, and agent-based simulation.
Urban Infrastructure Monitoring: Building a three-dimensional
model of urban infrastructure is vital to national security in critical zones, to infrastructure maintenance, and to the construction of
urban sets for movies such as King Kong! Cutting edge, Canadian
techniques such as terrestrial LIDAR scanning (shown at right)
are revolutionizing the process of ‘capturing’ a city into a GIS.
LIDAR model of a donut shop in Quebec, showing the
combined 3d location and texture mapping capacity of terrestrial LIDAR scanners. Courtesy Paul Mrstik, Terrapoint
Inc.
Line of Sight Analysis: GIS tools can show, given a 3 dimensional
model of the world, what can see what. Line of sight analysis is
useful for military operations (sniper scenarios, observation point
placement), public safety (design of crosswalks and signage) and
public land use planning (aesthetic judgements about zoning).
Related Subjects
GIScience is strongly related to a number of fields that either
extend it in new directions or are strongly related in focus. These
include:
Visualization: Scientific Visualization concerns the display of
scientific data. Landscape visualization is a subfield of GIS that
focuses on 3 dimensional features such as mountainsides.
Even the simplest 3 dimensional landscape model
significantly increases the difficulty of GIS due to the
vastly increased number of spatial relationships that are
possible.
Urban Planning: Urban planning makes extensive use of GIS
techniques, and many GIS techniques are adaptations of paperbased methods from the history of urban planning.
CAD: Computer aided drafting tools were the precursors to GIS
tools. There are many CAD tools that incorporate significant GIS
functionality - the two fields overlap extensively.
References
There are many excellent GIScience texts. A couple worth mention are:
Chang, K.-t. 2006 Introduction to Geographic Information Systems, 3ed. McGraw Hill, Toronto, 432pp.
Lo, C.P., Yeung, A.K.W. 2002 Concepts and Techniques of Geographic Information Systems. Prentice Hall, Upper Saddle River,
New Jersey, 492pp.
There are a number of excellent research and trade journals about
GIScience. Of special note are:
International Journal of Geographical Information Science.
Geoworld Magazine. Online at www.geoplace.com.
Special purpose extensions to commercial GIS tools,
such as Arc Military Analyst shown here, broaden the
use of GIS in specific communities.