The Nature of Geographic Information Systems
By Charles Convis, ESRI, Sept 28, 1996
Table of Contents Section 2
What is GIS?
Top 10 Reasons Biologists Donít Use GIS
GIS and Antique Cars
Cassiniís Planisphere, a Framework for the World
What is GIS Again?
PARTS OF A GIS
Data Foundations of GIS
Technology Foundations of GIS
Theoretical Foundations of GIS
Human Foundations of GIS
GIS and a New Framework for Conservation Biology
This diagram shows 4 basic parts of a GIS in the context of our relationship
with nature. In practice, a GIS consists of people using technology to
work with data under various methods in order to meet specific human needs.
This has a specific implication of human-centered activity, in contrast
to other technologies, databases and theories. Nature underlies all of
these parts. The human study of nature gives rise to much of the theory
and methods used, and recording of landforms and natural patterns constitutes
much of the available spatial and related data. Human needs for health,
clean air & water, food, solace and wonder are all part of our relationship
with nature. Technology arose historically from our pursuit of dominance
over nature (and over each other in wartime).
We will now look at each of these parts in more detail.
The data that a GIS operates on consists of any data bearing a definable
relationship to space, including any data about things and events that
occur in nature. At one time this consisted of hard-copy data like traditional
cartographic maps, surveyorís logs, demographic statistics, geographic
reports and descriptions from the field. Advances in automating and digitizing
data in the last several decades have allowed more and more such data to
be incorporated into digital form making it accessible to computerized
analyses. Advances in spatial data collection, classification and accuracy
have allowed more and more standard digital basemaps to become available
at different scales.
The technology foundations of GIS have been changing extremely rapidly
so this diagram tries to represent not just the technology elements but
their trends through time. GIS technology is based on computer hardware
and different types of software programs. Extremely rapid development and
advancement is typical of both elements. A rule thumb is that they double
in analytical and processing power every 3 years, unlike any other industry
in the history of humankind. A particular trend in development has been
more powerful ways to link computers together, such as the internet, web
graphic sites, and "data superhighways" allowing wider and wider
distribution of power. This also has the effect of making old-style centralized
computer installations less interesting and instead viewing knowledge and
data operations as decentralized across a wide variety of players or clients,
with an according shift in the central computer to services in support
of those clients. A related trend has been the development and acceptance
of standards for handling higher and higher level functions, like graphics,
allowing these to become standard parts of all computers and freeing software
developers to address higher and higher levels of creative functionality,
ease of use, and software design (like object orientation). Improved networks
and software capability has meant software fields like GIS and Statistics
are able to advance and merge in more powerful ways to create commercial-grade
software in formerly challenging and specialized application areas like
ecological modeling, wildlife management and taxonomic databases
The different analysis, management and application methods used in a
GIS come from a lengthy history of theoretical advances in a broad variety
of fields, including Landscape Architecture, Mathematics, Geography, Systems
Theory, Biology and Political Science. The first figure (above) shows how
these fields provide the conceptual models for GIS capability in five general
areas, space, features, operation/analysis, integration/synthesis, meaning/significance.
The second figure (below) shows an alternate temporal view of how these
fields built upon each other over time to create the body of theory comprising
GIS today. We will trace the development of just a few of the basic ideas
of GIS theory.
Models for describing space: These
represent the different models used to describe space, derived mainly from
concepts in statistics, topology and cartography.
- Space is a plane of continuous variation, which we can sample
at different points. This is a statistical concept of space commonly
found in biological hypotheses. This model emphasizes continuity and continuous
variation. It is usually managed by dividing space into a regular grid
of cells, also called a raster or matrix, where each cell has a value for
the variable in question. Additional variables are put into additional
layers using the same grid. This sets up a multi-dimensional matrix registered
to the ground such as commonly used in multivariate statistical studies
of habitat and niche. The concept of space as a regular plane was first
formulated by Descartes in 1637 in developing a language of coordinate
space for geometric analysis. Gauss and Humbolt carried these ideas further
in working together to map the earthís magnetic field, creating the first
thematic map based on systematic sampling. Gauss was also the first to
publish methodologies of mathematical statistics. R.A. Fisher finally formalized
modern concepts of statistics, sampling on a plane, and multivariate methods
in 1930, by which time gridded research sites were in common use among
biologists. Raster models are very simple to automate and their use was
so widespread among the first GIS programs of the 1960ís and 1970ís that
they can be said to have been independently arrived at numerous times.
- Space is a plane of discrete things, which we can classify according
to spatial relationships and multiple attributes. This concept,
also called a vector model, is mostly topological, and borrows from geometry
the notion of discrete things, which we represent in space a points, lines,
polygons and surfaces. Topology provides the formal language for defining
the invariant relationships between all of these geometric elements regardless
of their shape. This model also underlies traditional cartography, which
is important in GIS since maps are a primary source of data for GIS. Cartography
as an art began with the Renaissance and the first voyages of discovery,
receiving considerable impetus with the introduction of the printing press.
Navigational usage of maps was limited, however, until Alberti published
his treatise on the laws of projective geometry in 1435 which led Mercator
to his famous projection, where navigators could sail by taking angles
directly from the chart. A true understanding of map projections had to
wait until Gauss first formulated the ideas of non-Euclidean geometry in
1800, whereby forms were described not according to their fixed shape,
which could vary depending on how you tilted or twisted space in projecting
earthís sphere onto a flat map, but by the unchanging relationships of
points to lines, and lines to polygons, thus inspiring his student Reimann
to found the formal science of topology in 1851, on which all vector models
are based. The specific linkage of vector topological models to relational
models for description of attributes, now called the geo-relational model,
was developed at ESRI in the 1970ís.
- Locations in space are described using coordinates.
This concept originated with Descartes when he first invented the science
of analytical geometry and provided a numeric method (coordinates) for
describing shapes in space. Coordinates can be thought of as defining a
grid across space.
Models for describing features: (things
in space) Classified features are differentiated using distinct boundaries.
This reflects another concept as old as science. Nature is fuzzy, class
boundaries are not. Applying a class boundary to nature is an act of applying
a disciplineís sharp world view over blurry natural patterns to see if
interesting conclusions, questions, or decisions are produced.
Models for creating meaningful results: Operation
and analysis - Spatial analysis includes questions of adjacency,
containment, exclusion, proximity, subdivision, grouping, orientation,
movement, time. Spatial Analysis is based on the ideas of spatial overlay
and spatial sets which derive from set theory operations, such as intersection
and union, defined by Cantor in 1874. It also draws from developments in
simulation and modeling in the early days of computers. Early GIS developers
coined the term "map algebra" in the 1960ís to describe spatial
operations on gridded data, and the developing field that later came to
be called computational geometry provided early algorithms for spatial
analysis in the 1970ís.
- Integration and synthesis - Spatial data from different
sources can be integrated by restructuring, generalizing and transforming
(Maguire and Dangermond 1991). Restructuring is the process whereby data
from different models is recast into a common model (i.e. raster into vector).
Generalizing is where detailed data is smoothed out and/or aggregated to
make it align better with less detailed data. Transforming is where data
is changed to a new coordinate system, a new scale, or a new map projection
in order for it to match the other data sets. - A basemap standard provides
the common framework against which other spatial data can be processed
to ensure compatibility. Two different maps can be made to match each other
using the methods above, but a more long-term solution is to select a neutral
base map for use as a common foundation across projects. The idea of standard
basemap control dates back to Cassini, and in 1891 a program to develop
a standard international map of the world was begun, which was not concluded
until the US Defense Mapping Agency finished the Digital Chart of the World
- User Needs Analysis - GIS projects must explicitly define
the needs, capacities and expectations of the project users if they are
to be successful. This is based on formal methods for user needs analysis
and project design developed by the very first GIS (Tomlinson 1962), and
drawn in turn from concepts in cybernetics and systems theory. - Human
environments can be designed to be in harmony with natural environments.
This is a fundamental assumption of Landscape Architecture, with served
as the birthplace of the GIS industry at Harvard in the 1960ís and 1970ís.
The ideas were first formulated by Geddes in 1904, leading to the international
development of town and country planning.
Humans have basic needs for food & water, shelter, and health. Through
most of our history these have been directly derived from nature and the
various industries that provide them are current users of GIS technology,
such as transport, forestry, and utility companies. Humans have also often
shown a basic and often selfish desire for power, such as exclusive control
over land and resources. The careless pursuit of these "basic"
needs is a fundamental cause of the current extinction crisis, however,
the use of GIS as a planning and management technology among some of these
industries and a growing interest in sustainability holds hope that enough
bridges can be built between opposite camps, organizations and disciplines
to slow and stop these causes. Around these basic needs is a set of less
self-centered needs regarding a personís place among others and in the
world. The need to belong is very strong and expresses itself at many different
levels from family to church to community to humanity as a whole. This
also includes the equally strong need to belong to a place in the natural
world, ranging from landscapes of childhood memories and dreams to parks
and bioregions. The need for beauty and awe is an important part of the
character of the west and the human connection with wilderness. It is also
why the arts are such a fundamental part of the human experience, and why
humans are so profoundly moved by proximity to things that touch the deepest
levels of their subconscience such as wilderness, wildlife and great art.
The desire to protect wilderness and wild nature is common in the computer
industry as a whole and the GIS industry in particular. The millions paid
to acquire the notebooks of the greatest artist in history Leondardo, came
not from the arts but from the pioneer of personal computing. Humans also
have a need to explore, to engage their curiosity in trying to comprehend
the awesome, beautiful universe around them. We also have a corollary need
to communicate and teach. This need is most evident in the sciences, which
have the additional distinction of trying to understand universal truths
about what the world is and how it works, and trying to do it objectively
so that others can understand it as well through independent verification.
Finally, humans have the need to believe in something beyond themselves,
which commonly expresses itself as religious faith. The details of that
faith donít matter so much as the ability of faith itself, when exercised
in love, to inspire humans to work selflessly in service to others, especially
those in need. Stewardship of a piece of wild land differs little from
leading a congregation in this regard, the conscious act of placing the
needs of strangers before the needs of oneself is uniquely human.