Acknowledgements
Abstract and keywords
Chapter I: Introduction
Chapter II: Background2.1 DefinitionsChapter III: Methodology
2.2 History of GIS
2.3 Components of GIS
2.3.1 Data Sources
2.3.2 Data Structures
2.3.3 Data Analysis
2.4 GIS in Archaeology
2.4.1 History and Methodology
2.4.2 Applications
2.5 GIS pitfalls
2.5.1 About error in spatial data
2.5.2 Data Pitfalls
2.5.3 Analytical Pitfalls
2.5.4 Theoretical Pitfalls
2.5.5 Interpretation Pitfalls3.1 Target PopulationChapter IV: Results
3.2 About the Survey
3.3 Risk Analysis
3.4 Survey Format4.1 About the ParticipantsChapter V: Conclusions
4.2 Familiarity with GIS
4.3 Type of Applications
4.4 Impact on Research
4.5 Awareness of Pitfalls
4.5.1 Data Collection
4.5.2 Metadata
4.5.3 Accuracy tests
4.5.4 Generalizations
4.5.5 Computer errors
4.5.6 Distance Measurements
4.5.7 Overlays
4.5.8 Vector/Raster Conversions
4.5.9 GIS Algorithms
4.5.10 Map Scales
Appendix A. National Map Accuracy Standards
Appendix B. Overview of Spatial Data Transfer Standards
Bibliography
This thesis would not have materialized without the help and guidance of a number of people. I would like to thank my advisors Gregory Johnson for his valuable input, and Chuck Elschlager for his productive comments. David Rapson, from the University of Wyoming has provided me with many great ideas to implement in this thesis. Ian Johnson at the University of Sidney and Kris Hirst from the Mining Company have been instrumental in bringing the survey that the project is based upon to the wide population of archaeologists using GIS. Aydin Baltaci lended his generous help with the CGI script that drove the online survey. Wayne Spivak from SBA consulting kindly hosted the survey on his server. Finally, Monica Greaves has given much of her precious time to the editing of this thesis. Her superior grasp of the English language has made the thesis much easier to read. Thank you Monica.
Other people have not been involved with the thesis directly, but have, nonetheless, been an invaluable influence. I would like to thank my friends, Redwan El Jaouhari for his encouraging words, and Ase Skarpengland for lifting my spirits many times through her cheerful e-mails.
It is my honor to dedicate this thesis to my parents, Hafida and Boutayeb Gourad . Thank you for everything you have given me.
Geographic Information Systems are a recent tool in the field of archeology.
As a complex pattern recognition device that is only useful if correctly
understood, success of any archaeological endeavor utilizing it is directly
linked to the awareness of its limitations. Via an online survey, the project
seeks to highlight the scope of GIS use in the field of archaeology as
well as the archaeologist’s knowledge of GIS pitfalls.
Keywords:
Geographic Information Systems, archaeology, Spatial data certainty.
Geographic Information Systems (GIS) are tools for the input, analysis
and output of spatial data. Geographers initially used these tools for
resource management purposes (Burrough 1986). Over the last decade, GIS
applications have revolutionized many disciplines in many ways (Marble,
1990), though some disciplines adopted them earlier than others. In the
field of archeology, GIS has barely reached the end of the experimental
stage. Although it was used fairly regularly in the early 1980’s, (Kvamme,
1996) its present utilization has dramatically increased. At the time this
paper was written, over 500 archaeologists worldwide were registered GIS
users with the online database "GIS-using archaeologists", developed byPaul
Miller and Ian Johnson in 1995. It is suspected that the actual number
of GIS users in the archaeology circle is substantially higher.
Archeology, as a spatial discipline, has used GIS in a variety of ways.
At the simplest level, GIS has found applications as a database management
for archaeological records, with the added benefit of being able to create
instant maps. It has been implemented in cultural resource management contexts,
where archaeological site locations are predicted using statistical models
based on previously identified site locations. It has also been used to
simulate diachronic changes in past landscapes, and as a tool in intra-site
analysis; although this last application has not enjoyed the same popularity
as the others.
Like any new technology, GIS carries with it a variety of pitfalls that
only the experienced and critical user might be aware of. These pitfalls
fall under several categories. They can be related to unqualified data
sources, analyses mistakes, blurry perception or theoretical glitches.
However, the end- results are the same; decision making is all too often
based on erroneous results. In many disciplines, including Geography, where
GIS was first used, working the error potential of GIS is as established
as the use of the tool itself. In archaeology, such is not the case as
there seems to be more excitement about the tool than caution. Indeed the
archaeological literature is replete with GIS success stories, while caveats
are scarce.
Since GIS is a complex pattern recognition device that is only useful
if correctly understood, the success of any archaeological endeavor utilizing
it is directly linked to the awareness of its limitations. (Kvamme, 1988).
The purpose of this paper is to highlight those limits in general and in
the field of archaeology. This project also seeks to establish a quantitative
approach to the use of GIS in archaeology, and answer questions such as:
How is the tool being used? How are archeologists educating themselves
about GIS? What are the most problematic and most successful applications
of GIS in archaeology? How many archaeologists are aware of the specific
shortcomings of GIS, and what is the impact of GIS on archeological research?
The project makes use of an online questionnaire that contains questions
designed to reach five objectives: 1) Determine who is using GIS in the
archaeological community. 2) Establish the user’s familiarity with the
tool. 3) Highlight the type of archaeological applications and analyses
that are being conducted using GIS. 4) Find out about the impact of GIS
on the user’s research. 5) Establish the user's level of awareness of GIS
pitfalls. It is hoped that the project will provide a better understanding
of the scope of GIS use in archaeology, as well as highlight the limitations
of the tool.
Attempting to isolate one definition of GIS that functions universally
has proven difficult, if not futile (Maguire 1991, Clarke 1997). The following
definitions will demonstrate the nuances of the technology:
"An Information System that is designed to work with data referenced
by spatial or geographic coordinates. In other words, a GIS is both a database
system with specific capabilities for spatially-referenced data, as well
as a set of operations for working with the data" (Star and Estes, 1990,
p.2)
"A geographic information system is a special case of information systems
where the database consists of observations on spatially distributed features,
activities or events, which are definable in space as points, lines, or
areas. A geographic information system manipulates data about these points,
lines and areas to retrieve data for ad hoc queries and analyses" (Duecker,
1979, p 106)
Tracing GIS ultimately leads us to the first use of the thematic map
to graphically depict one spatial aspect of the real world. Cutting features
from certain thematic maps and adding them to others turned out to be a
very useful concept, and was employed as early as 1922. In 1950 Jacqueline
Tyrwhitt in Britain combined four data layers of land elevation, surface
geology, hydrology, and farmland into a single map thereby establishing
herself as the inventor of map overlay, the heart of GIS operations today.
Two planners at the Massachusetts Institute of Technology took her work
further by introducing weighting techniques to map overlay by photographically
manipulating map layers (Clark 1997, p. 7).
Thus far, map overlay was a manual task that involved physically cutting, aligning and superimposing map layers. The computer revolutionized cartography by automating these tasks. As early as 1959, a graduate student by the name of Waldo Tobler introduced a computer technique that allowed to digitally superimpose maps (Tobler, 1959). His techniques consisted of three stages, which he called MIMO (map in-map out): map input, map manipulation, and map output. This was to be the first of the many computer-mapping programs that took advantage of the advent of modular programming languages.
The software packages were not very powerful in terms of map analysis, but did simplify the map overlay process. This was crucial because it gave impetus to the introduction of digital spatial data to government agencies. Hence, the introduction of CIA’s World Data Bank and the US Census Bureau’s DIME (Dual Independent Map Encoding), a breakthrough encoding system in digital mapping that is still in use today. (Clark, 1997, p.8). More digital data necessitated the creation of more powerful systems to analyze it. One of the first fully developed GIS programs was The Canada Geographical Information System (CGIS), which debuted in 1964. Other programs either did not adapt to changes in computer hardware trends, or were simply not developed further.
In the 1980’s, as computer hardware and data storage became more affordable,
GIS systems made the transition from the mainframe to the minicomputer
and finally from the workstation to the personal computer. Academically,
GIS classes were incorporated in curricula and commercially, it became
a multi-billion dollar business.
GIS derives its analytical capabilities from combining a database management
system with a graphical display. The two entities (tabular and graphic)
are linked together in such a way that operations on one are automatically
reflected in the other. The most crucial components of GIS are digital
map data, their representation, and the set of data manipulation functions.
One way to produce GIS data is to manually enter point coordinates of
the area of interest, which could be taken directly from field surveys
using traditional tools or GPS (Global Positioning System) receivers. Data
can also be converted from databases that already have spatial coordinates
embedded in them. Existing paper map data still constitute a large source
of raw data for GIS. The most common methods of transforming that medium
into digital data are scanning and digitizing, either manually or automatically.
These trends are starting to change with the increase in the use of remote
sensing techniques. Narrowly defined, remote sensing is the measurement
of the electromagnetic properties of a surface without making direct contact
with it (Davis and Simonet, 1991). The technology includes Aerial photography
and satellite imagery. Major sources for these media in the United States
are the National Airphoto Program, the Landsat program, and the French
SPOT satellite. Remotely scanned data is available either directly from
the government or through distributors. A large amount of these data is
available via the Internet as a little or no cost download.
One widely used data source is the Digital Elevation Model (DEM), which
is a digital representation of continuous spatial variation (Burrough,
1986). DEMs are produced through the interpolation of sampled data of the
earth’s surface through remote sensing techniques. Various data can be
derived from DEMs including: Contour maps, line of sight maps, maps of
slope, convexity, concavity and aspect, shaded relief maps, drainage network
and drainage basin delineation. (Burrough, 1996)
Geographic Information Systems combine the power of databases with graphic display. The database component of GIS consists of tabular records that are connected to each other via various means depending on the database, the most popular of which is currently the relational database. The way that tabular data is portrayed graphically falls within two formats, vector or raster. Vector data structures are displayed through points, lines and polygons. The coordinates of every point, as well as the direction of lines, connections, and adjacent polygons are explicitly stored (figure 2.1). In contrast, a raster or grid format displays an image or map through pixels that carry inherent values. Thus, in a raster data structure, a point can be represented by a cell of a certain value. A line is represented by a number of connected cells, and an polygon is formed by grouping cells in such a way as to represent the actual area. Figure 2.2 and 2.3 show a line and a chair displayed in both vector (Side A) and raster format (Side B).
Figure 2.1: Vector representation of spatial data. (from Clarke 1997)
Figure 2.2
Figure 2.3
Difference in raster and vector representations. (from Burrough, 1986)
GIS are tools for spatial analysis that can perform functions as simple
as measuring the distance from point A to point B, to complex modelling
of spatial patterns. The following are common GIS analyses and their potential
archaeological applications.
- Simple Distance, or Euclidean distance: this can be used to determine travel time and proximity to points of interest.
- Vector and raster overlay: Vector overlay falls under the three categories of point in polygon, line in polygon and polygon in polygon overlay. Raster overlay is accomplished by performing mathematical formulas on cells of the layers in the overlay (figure 2.4). The power of overlay is obvious in a discipline that is concerned with cultural layers. One could generate new themes based on different layer combinations. For example, one could overlay a tool layer with a bone layer to identify where tools and bones coincide.
Figure 2.4a: Simple overlay in raster based GIS
Figure 2.4b: simple Overlay in a vector based GIS
From Daniel Karnes (http://www.dartmouth.edu/~geog48/lectures/Lec07.html)
- Polygon shape, length, perimeter, area and edginess: These analyses highlight the shape of spatial entities, which can reveal a great deal about their function. For example, long thin polygons tend to have less surface area than circular polygons, which can be measure of defense, and thus are a factor in settlement patterns. Perforated areas within the polygon, or spatial integrity, can also be crucial for statistical models that consider spatial continuity as an independent parameter.
- Functional Distance: this involves adding impediments to Euclidean distance to simulate real life situations, which is crucial to archaeological modelling because based on such distances, least Cost Surfaces can be generated and least cost paths established.
- Neighborhood Functions: This includes the much-used function in archeology, nearest neighbor analysis, which can be applied to one or more layers.
- Directionality, connectivity and network complexities: these are often used in network analysis. An examples of which in archaeology is the hydrological network, which can signal the presence of site locations (Allen, 1990b; Zubrow, 1990a).
- Density of distribution, quadrat analysis and Thiessen polygons: these are common functions used to find a break in a certain pattern, such as biological distributions.
- Gravity Models: these are extensions of cost surfaces. Layers can be weighted and combined according to defined parameters to create a locational model.
- Slope and Aspect: Usually derived from Digital Elevation Models (DEMs), these are some of the most commonly used parameters in predictive modelling.
Interpolation: Where surface data is not available, GIS can be used to create continuous spatial data based on a sample of points. Many Interpolation algorithms are built in current GIS software packages. The nature of the terrain determines the best algorithm to be used. One of the most commonly used data sources, themselves a result of interpolation, are Digital Elevation Models or DEMs.
Archaeology has made use of many technologies in the past. As early
as 1922, archaeological sites were being photographed using World War I
surplus planes (Olsen, 1985). Such uses of remote sensing are now common
in the discipline and can be combined with GIS for even more powerful analyses.
Remote sensing techniques brought many benefits to archaeology. It is the
most efficient way to survey large areas of land for structural changes
in the soil, vegetation or terrain that might be a result of human behavior.
It is also the fastest way to survey unsafe terrain or sites where archaeologists
are not allowed entrance for political or other reasons. Similarly, GIS
brings many benefits to archaeology. Data updates and corrections which
were once slow and costly can be conducted quickly and efficiently. Using
the technology, measurements can be completed in a matter of seconds. Building
complex simulations that are only limited by imagination becomes possible
in a reasonable amount of time. (Kvamme, 1989).
The first regular use of the word GIS in archaeology started between
1983 and 1985 (Kvamme, 1996). The application that is credited with making
GIS a mainstream tool in the field of archaeology is predictive modelling.
During The 1970’s cultural resource managers were eager for tools that
would help survey the vast amount of federally controlled land. In addition,
as archaeology shifted its focus from the site to the landscape (Warren
1990b), data was more readily available for immediate analysis and there
was more impetus to make use of automated predictive modelling. Computers
became established as a favorite tool to create the statistical models
that became mainstream in archaeology (Alderderfer, 1987).
Predictive modelling seeks to establish a causal relationship between
certain environmental parameters and known archaeological site locations,
build a statistical model based on that relationship, and apply the model
to unsurveyed land. The time and labor saved is phenomenal, and GIS was
the most capable tool for the task. Predictive modelling gained tremendous
popularity during the latter half of the 1980’s as statistical methods
grew better and computer hardware and software became more powerful, and
financially manageable. GIS use expanded not only in the archaeological
community, but in the social sciences as well. Unlike many technologies,
GIS was seen as more than a tool. Many considered it to be a revolution
in spatial thinking and a catalyst for a change in the way human spatial
behavior is studied (Marble, 1990). In many ways, the technology has been
equated to the advent of relational databases. As opposed to the other
types of databases, network and hierarchical, when relational databases
were introduced to archaeology, they affected the discipline by changing
the way data was collected, the nature of questions asked about it, and
the overall conceptual design of the archaeological dig. (Harris and Lock,
1996 p.352)
Since predictive modelling was such an influential factor in the development of GIS in archaeology, it is appropriate to define "modelling" and explain its role in archaeology.
As archaeology moved from the descriptive to explanatory stage, predictive
modelling became crucial (Sebastian & Judge, 1988).Voorips defines
archaeological models as "Partial representations of a theory and are formulated
in a manner which enables the archaeologists to test the theory by means
of empirical data." (Voorips, 1986) Warren describes predictions as being
crucial to the scientific method. When new patterns are found in a certain
context, hypotheses are formed to explain them and models are built in
other contexts to test the hypothesis (Warren, 1990a).
The statistical models used in archaeological predictive modelling are
a variation of regression analysis and its many categories. Bivariate linear
regression looks at the relationship between two variables, an independent
variable, such as proximity to rivers and an independent variable, such
as the presence of an archaeological site. Multiple regression studies
the relationship between one dependent variable and two or more independent
ones allowing the independent variable to be proximity to rivers as well
as slope. Logistical regression models, on the other hand, trace relationships
among multiple independent and dependent variables. In this case, the dependent
variables could be the presence as well as absence of archaeological sites.
The independent variables can include as many parameters as the archaeologist
deems important to archaeological site location (Savage 1990). For this
reason, this technique is best suited to archaeological applications, and
is the one that is used the most. Stepwise logistic regression is an extension
of a standard logistical regression and is different only in that it uses
pre-selected significance levels at which variables are either added or
deleted. (Christopherson et al., 1996)
The archaeological literature abounds with examples of GIS use in archaeology.
The following paragraph examine applications that fall within four categories:
archaeological predictive modelling, simulation of past changes, intra-site
analysis, and database management.
One example of using GIS to implement an archaeological predictive model
is the Ontario ministry of natural resources’ three-year research project
to predict landforms that are likely to contain archeological sites. The
project used a database of known archaeological sites to establish the
success of their predictive model. 75% of the sites in the model matched
actual archaeological sites in the database. A more detailed description
of this can be found on the World Wide Web, on the following site: (http://modelling.pictographics.com/intro.htm.)
Another
example is the GIS project of the Tell el-Umayri sites, which range from
lower Paleolithic to late Islamic and from urban centers to small camps
(Christopherson et al., 1996). The first part of the project consisted
of creating an environmental probability model using stepwise logistic
regression models in order to find additional sites based on the known
localities. The second part of the project involved building erosion models
to look at Iron Age agricultural intensification patterns. The project
used the Universal Soil Loss Equation (USLE) developed by the USDA Agricultural
Research Service to build a model of erosion potential and its relationship
to the presence or absence of archaeological sites. Likewise, as a part
of an impact assessment for a proposed US Air Force program in Montana,
GIS was used to develop a predictive model of archaeological site distribution
in an 8500 square mile area in north Montana. Logistic regression techniques
were used to build the model (Carmichael, 1990).
Just as GIS can be used to model the location of archaeological sites,
it can also be implemented to simulate past activity. Kathleen M. Allen
used GIS to model trade networks between Native American s and Europeans
in the early historic period (ca. AD 1550-1750). The data for the project
consisted of hydrology maps, the location of the native populations and
the locations of European forts, and trading posts (Allen, 1990b). The
analyses were conducted using ArcInfo’s network module. Zubrow also used
a hydrology coverage as data to model the spread of European populations
through the state of New York between 1608 and 1810. The study created
different migration models according to various initial situations and
different impedance levels on the capacity of the hydrological networks
as well as the populations. The models allowed the author to shed more
light on the process involved in settling a frontier and to further identify
the limiting factors of migration (Zubrow, 1990a). Finally, a study of
the changing pattern of settlement and land use over 2000 years, ranging
from the Celtic Iron Age to the modern period in France made use of GIS’s
modelling capabilities. The project was undertaken in the Arroux River
Valley in Burgundy, and used SPOT modern land use data together with earlier
Landsat data from 1972 to conduct diachronic studies on the landscape.
The GIS package GRASS was used to conduct analyses including: line of sight,
optimum routings and three-dimensional image displays (Madry & Crumley,
1990)
GIS use in intrasite analysis has not enjoyed the popularity of other applications because most GIS packages present difficulties in handling multi-dimensional data (Harris & Lock, 1996). Nonetheless, the third part of the Tell el-Umayri project used GIS to examine ceramic distributions within the sites (Christopherson et al., 1996).
GIS has also been used as a multi-site management device. Because of
the ease of updating and producing hybrid maps, GIS has been the tool of
choice of many archaeologists to keep track of the archaeological record.
Such use is exemplified in investigations at Olorgesailie, a Pleistocene
hominid site in Kenya. The tool was helpful in shifting the focus from
the site to the nearby landscape (Potts et al., 1996).
The above examples are some of the early applications of GIS in archaeology.
The literature is now replete with more diverse examples making use of
the ever more powerful capabilities of new software.
2.5.1 About Errors in Spatial Data
The most accurate spatial databases will not correctly represent the
real world and should instead be thought of as abstractions of reality,
which means that any GIS data will have accuracy and precision problems.
It is now a well-accepted fact that error should be seen as an inherent
dimension in digital data and not a mere inconvenience (Chrisman, 1991).
Instead of trying to solve all accuracy problems, an impossible task, it
is more realistic to factor them into the analysis, and establish an error
cap on the data and the operations to be conducted on them that is dependent
on how exact the results need to be.
The two issues of concern when dealing with digital data are accuracy
and precision. Accuracy is a measure of deviation from the truth (Dutton,
1989), or a standard of what is considered a true value (Chrisman 1991).
Precision refers to the degree of detail in which an entity is recorded.
Aberrations in either can lead to spatial error, categorized either as
substantive error, which is likely to affect the conclusions reached in
an analysis, or trivial errors, which are harmless (Fotheringham, 1989).
Finally, there are uncertainties, which are caused by a lack of parameters
to judge the accuracy of spatial data. (Dutton, 1989).
Errors can be classified as follows: 1) Data Collection and recording
errors, which can be a result of careless collectors and the use of inadequate
methods and tools. Following the collection phase, errors can occur when
data is classified and entered in a digital database. 2) Data manipulation
and analysis errors, which can present the most problems and are usually
propagated by several levels of abstractions making them difficult to trace.
3) Theoretical errors, which are discipline-specific and are usually the
result of faulty or simplistic theories. 4) Data Interpretation errors,
which lie in the perception of the user and are generally the result of
erroneous generalizations or misuse of the tool.
It is very easy to be naïve about error in GIS for multiple reasons.
Documented case studies of GIS flaws are not as plentiful as GIS success
stories (Aangeenbrug, 1991). Map data analysis has a heavily laden legacy
of using more traditional, less precise tools that are not as complicated
as GIS. The increase of precision has not been accompanied by an improvement
in accuracy standards. There are no studies that empirically show the dimension
of GIS errors. Finally, there are no methodologies in any commercial GIS
package that can compensate for or at least alert the GIS user to the potential
propagation of errors in spatial databases.
Results of GIS analyses are only as good as the data used to produce them. Certain types of spatial entities are easier to map than others. Features of a discreet nature such as soil or vegetation data are inherently difficult to map due to their fluid distribution. Instead, the maps pass through an intellectual filter that places arbitrary boundaries on the data. (Demers, 1991). Accuracy issues are present in virtually every GIS data source.
Digitizing is subject to human error as well as many other factors,
including the age of the map, the type of map media (Bolstad et al., 1990;
Dunn et al., 1990) as well as the quality of the original map. One study
documented manual digitizing precision of approximately 0.0025 inch for
well defined points, which corresponds to 4.8 feet at a scale of 1:24,000,
and 50 feet at a scale of 1:250,000 (Bolstad & Smith, 1992). There
is also the process of coordinate registration, which can potentially cause
more error in the digitizing process. Coordinate registration involves
converting from digitizer coordinates to the coordinate system that is
defined by the source map’s projection. (Bolstad et al., 1990).
Scanning map data spares the user a lot of time and labor. However,
the accuracy of image produced is a direct function of the scanner used.
(Litton, 1996). The resulting data is also still subject to human editing,
another way of introducing more errors. Aerial photography can affect data
accuracy through variation in the vertical movement of the aircraft bearing
the camera, which can produce less than reliable photographs, especially
around the margins. Satellite imagery is usually thought of as having a
high standard of accuracy, but one should not neglect the fact that the
data derived from it is usually a result of classification techniques that
are ultimately human-driven. Since the data is pixel based, the size of
the pixel itself is a level of abstraction that can affect analyses on
the data.
Another popular source of data is the Digital Elevation Model, which
are readily available and span the entire United States. Archaeologists
in particular have been avid consumers of low quality DEMs owing to their
low cost. DEMs are ultimately a result of interpolation algorithms on a
number of points that have x, y, and elevation values. The points are either
sampled or systematically spaced (Theobald, 1989). It is important to know
that when point data is converted to continuous data, error is likely to
occur, regardless of how good the interpolation algorithms are. As an example,
DEM’s derived from digitized 1:250,000 scale USGS maps have a horizontal
error of ± 127 m; therefore, the unwary
use of DEM data can produce significantly erroneous results (Walsh, 1990).
In a landmark paper, Kvamme warned about the naïve use of DEMs (Kvamme,
1990). The author compared two data sets. The US Army’s defense Mapping
Agency’s DEM based on elevation contours of 1:250,000 scale maps and the
USGS 7.5 minute series with a scale of 1:24,000. The results showed that
a substantial amount of detail was lost in the low resolution DEMs (1:
250,000 scale). Major landforms were smoothed, and minor drainage, ridges,
and hills were absent from the data. This is of concern because it is precisely
these features that archaeologists use most often as environmental parameters
when creating archaeological site prediction models. In addition DEMs are
themselves sources of many other derivative data (slope, aspect…etc). In
another paper, hydrological networks were derived from DEM data in which
small controlled errors were added. The results showed that even errors
of small magnitude significantly affect the quality of DEM extracted data.
(Lee, 1996).
Some of GIS most powerful analytical capabilities are also the source
of the most serious errors. From simple distance measurements to complex,
multi-layer models, the tool is replete with shortcomings that are not
necessarily obvious to the user. The following paragraphs discuss threats
to the accuracy of GIS analyses:
GIS packages process spatial data with a much higher resolution than
the data itself. Most GIS systems use at least 8 decimal digits to allocate
coordinates. 8 decimal digits applied to the entire planet could map spatial
entities up to the nearest 10-cm, which is much more precise than any available
data set. Such precision can lead to unwanted objects (Gopal & Goodchild,
1989).
Measuring distance in a vector-based system is straightforward. In a raster system, however, where the data is based on pixels of a certain size, measuring a distance between point A and point B is a matter of adding up the pixels between the points (Demers, 1997, p.205). The problem with this method is that for lines that are curvy, the diagonal distance of pixels needs to be measured, a capability not available in all GIS packages. As a result, any function that uses distance measurements, including buffering, least cost analyses and many others are subject to errors (Griffith, 1989).
Three of GIS most appreciated capabilities are overlaying, changing
scales interactively, and combining data from different sources. These
capabilities often lend themselves to abuse by GIS users (Dutton, 1992).
Currently, there is no GIS package that warns the user when two maps with
different scales or different projections are overlaid. This is a mistake
that even the first overlay procedures did not allow for. Combining maps
of two different projections is a clear mistake; however, changing map
scales in a GIS is not so obvious. In any GIS package, one can very easily
change a map of a small scale (1:1000,000) to a large-scale map (1:1000)
and print out maps of the latter scale. On the first map, a line that is
one 1 mm in width covers 100 meters, or the length of a football field.
On the second map, the same line will continue to cover 1 mm, and if the
output results utilize the second map, they can give the false impression
of precision (Demers, 1997: 56). In addition, combining data from different
sources can make errors difficult, if not impossible to trace (Maffini
et al., 1989).
Different GIS packages use different algorithms for certain analyses, including interpolations. This fact is not always explicit in the literature accompanying GIS software. Variations in the interpolated areas resulting from the differences are more than explicit, however. Kvamme conducted an experiment in which he used three different interpolation and slope algorithms on a single data set and each of the algorithms produced different results (Kvamme, 1990).
Conversions between raster and vector environments are embedded in most
recent GIS packages. The conversion, however, causes either loss of detail
or the addition of new, and spurious, detail. If the process is one of
rasterizing a vector map, then the points, lines and polygons are replaced
by pixels, which adds new dimensions to the data. A point, which has no
dimensionality on a map, becomes two dimensional in a raster environment.
Vectorizing a raster map on the other hand involves replacing pixels with
points, lines, and areas which, depending on the size of pixels, can add
misleading boundaries.
Other difficulties have more to do with the theoretical design of the analysis than the software itself. In archaeology, GIS has been used most often to predict archaeological site locations. The models created by correlating site location with environmental parameters have been criticized by many researchers including, Ebert & Kohler, 1990; Gaffney & Van Leusen, 1995; Kvamme, 1989 & 1996; Sebastian & Judge, 1988; and Gaffney et al., 1996. In a field such as archeology, where human interaction with the environment is complex and unpredictable, such predictive models can lead to environmental determinism (Gaffney & Van Leusen, 1995). Ebert and Kohler serve to remind GIS users of the difference between a system such as a society and a mechanism such as a clock. The predictive models created through GIS assume that the decision-making process inherent in choosing a place to settle is fixed and mechanical. In reality, early settlement choices are much more complex. In many settlements, there are decisions that are subject to causal relationships with the environment, but others are so spontaneous as to make any correlation faulty (Ebert & Kohler, (get date). This is especially true when societies make the transition from gathering food to producing it; their relationship with the environment becomes less salient and is instead, replaced with politics, technology, and other factors. The authority invested in GIS maps can easily mask these theoretical flaws. While the results may be aesthetically pleasing, it is quite possible that they have no substance. Hvar, a Bronze Age site on than island off the coast to of Central Dalmatia in Croatia, is a case in point. The site underwent two studies that clearly illustrate the potential errors of predictive modelling. The first study, which used GIS, came to the conclusion that there was a direct correlation between settlement patterns and economics, a fact that was clearly demonstrated by overlaying agricultural layers with settlement maps. A second, more detailed study proved the earlier results to be much more simplistic than what actually took place. Complex cultural events had contributed to the settlement pattern, which the first GIS study had side stepped (Gaffney et al., 1996).
Geographic Information Systems were initially designed for geographers
and continue to be designed for use by geographers. Any tool transplanted
from one discipline to another, problems is bound to encounter difficulties
(Zubrow, 1990b; Goodchild, 1991; Gaffney & Van Leusen, 1995). One such
difficulty is semantics. Archaeological jargon is very different from that
of geography or computer science. Archaeologists, then, must be wary of
the misinterpretation of certain words. An example of such a term is object
oriented technology, which refers to a new method of representing data,
where every object contains information about itself, methods of communicating
with other objects and a list of the classes of other objects it can communicate
with. The term may be confusing, because it blends precise concepts with
others less than precise. Other problems are more subtle but can still
be very damaging. When a new technology is first adopted in a discipline,
excitement about its potential can blur its shortcomings. This is especially
true with such tool as GIS, which has not been subject to much criticism
(Aageenbrug, 1991). Examples of GIS failures are rare while success stories
abound in the archaeological literature. As mentioned before, GIS can supercede
the role of a tool and becomes an end in itself. Kvamme discusses the definitiveness
that large computer databases can have (Kvamme, 1989). This is particularly
true in the social sciences, where researches are avid consumers of statistical
analysis. Anything generated by a computer can become indisputable because
the human beings behind the analyses tend to be separated from the machine.
Other difficulties such as the small number problem can be caused by the visualization of data in GIS, because of the ease of generating new maps with every single operation. The problem arises when the percentage of a distribution of a factor on a map erroneously dominates the map (Kennedy, 1989). For instance, if a spatial database of archaeological sites categorized according to the presence of human bones and other archaeological material contains a polygon with only 2 archaeological remains, one of which is a human bone, the percentage of human bone in that specific area will be recorded as 50%, which is obviously an overstated representation. The problem becomes even more serious when further data is derived and mapped from the percentages.
This project targeted archaeologists who use GIS. At first glance, this
might seem straightforward; however, reaching a target population that
best benefits the project proved to be rather challenging. This may stem
from the fact that not all archaeologists approach GIS the same way. Some
contract their work, some hire assistants to help them with the tasks at
hand, and others carry out the entire process, from data entry to results
on their own. As a large part of the project tested the knowledge of the
participant about GIS use and pitfalls, it made sense for the person doing
the analysis to take the survey and not the theoretical planner.
There also arose a need to define what can be deemed an " archaeologist".
Is an archaeologist an individual who holds an archaeology degree? Is an
archaeologist someone who practices archeology or is involved in archaeological
research? In this case it was deemed more important to include people who
generate archeological papers. Thus, the survey was rather fluid about
the title of the person who takes and was available to all persons conducting
GIS work in archaeology.
The web site hosting the survey was launched in April 1998 along with
a commercial site tracker in order to monitor site traffic. Invitations
to take the survey were sent during March and April of 1998. To call attention
to the project, the following steps were taken: 1) Over 2000 e-mails were
sent to archaeologists around the world. The survey was posted on many
Archaeology Internet ListServs including Arch-L, Anthro-L, GIS-L, AIA,
GIS-Arch, SAS, ArchComp …etc. 3) The page was also posted on other frequently
visited archaeology sites. As a result, the survey page received a total
of 660 unique hits during the months of April, March, June and July. (Figure
3.1)
Figure 3.1: Survey Site Statistics
Online surveys are not devoid of risks. Three problems were encountered throughout the project. The first was that the page hosting the survey could theoretically be accessed by anyone. The second was the possibility of computer and human errors, whether intentional or not. One of the most common glitches involved participants not knowing how to select multiple answers for a question. The third difficulty was selective input by participants, who chose to ignore certain answers.
To handle the first risk, I had employed a commercial site tracker. These software applications monitor the traffic flow to the site by providing the site owner with information about the visitors. The information includes:
- The IP address from where the visitor is logging to the site.
The second problem was much more serious. Some of the questions on the survey allowed for only one answer, while others allowed the participants to select multiple answers. Although detailed instructions were provided on how to do this from multiple computer platforms (Windows, Mac, and Unix) some participants still bypassed the multiple questions or selected single answers among the choices given. This was easily observable in the database that contained the results. The participants who had trouble inputting data also mentioned their difficulty in the comment area. This presented a problem because the data could be dangerously skewed. As a result, I had to redesign the survey, and add more conspicuous instructions on how to select answers next to every question that supported multiple input (figure 3.2). Participants were asked to complete the survey an additional time. This time the inputs were a lot more accurate than the first trial, which was reflected in the comments of the participants. The results were also scrutinized for inconsistencies. The final number of participants whose results were accepted for analysis was 140.
Figure 3.2: Instructions on how to fill questions that support multiple input.
The third problem, involving participants who opted not to answer some
of the questions, was taken into consideration in the analysis of the survey.
Statistical data were generated according to the number of people who answered
the question at hand. Initially, a script that forced people to fill a
tab was considered, but it was deemed that the choice of not answering
a question was in itself informative.
The survey consisted of six parts:
The first part gave instructions as to how to and who should complete the survey, and an approximation of how long doing so should take. The second part asked for information about the participant: his or her name, which was optional; geographic location; title, and Degree held. The third part attempted to establish the participants’ familiarity with GIS by asking the following questions:
Which one of the following statements most accurately reflect the impact of your GIS software on you research?
The sixth part of the survey sought to establish the participant’s familiarity with quality related issues by presenting 10 issues and selecting on of the following answers:
a- Not familiar with it
b- Familiar with it, but have not considered its impact on my research
The project ultimately accepted 140 entries. The geographic locations of the participants of the survey showed an expected concentration of GIS users in the United States. The next largest concentration was in Canada, followed by the UK, and finally Australia. (Figure 4.1). Table 4.1 shows the geographic location of the participants. 77% of the participants had either Masters Degrees or Ph.Ds.
Figure 4.1
Table 4.1: Geographic distribution of participants
Figure 4.2
91% of the participants were involved in GIS projects when they completed the survey. More than 72% had more than two years of experience using the tool, which is a reasonable amount of time to familiarize oneself with the technology (Figure 4.3). It was interesting that 33% of the participants had not attended any formal GIS classes, workshops, or seminars (figure 4.4). It is important to note that the length of the classes and their level were not emphasized. The question was intended to establish who had taught themselves GIS and who did not. Frequency of GIS use was high, with 41% of the participants using the tool everyday (figure 4.5). Support groups are a way of sharing news, suggestions and problems with people that share a common interest. Being part of an online support group is now one of the best way to find out about common applications of GIS use as well as common mistakes. 32% of the participants were member of GIS support groups (figure 4.6). Only 6% of the participants attended GIS related conferences on a frequent basis (figure 4.7). Going to a conference about a topic is a good indicator of the level of involvement of the participant in that topic.
Figure 4.3
Figure 4.4
Figure 4.5
Figure 4.6
Figure 4.7
Figure 4.8 shows the participants who were involved in the decision
making process to choose the GIS package they use. 63% of the participants
were indeed involved in choosing their system, which makes sense in a discipline
like archaeology, where projects usually involve an intimate number of
decision makers, and a substantial number of projects are one-person operations.
Figure 4.9 and figure 4.10 shows the type of software used by archaeologists
and the platforms they run on. Esri’s Arcview and Arc-Info were the leaders
in software used by archaeologists. This was no surprise as the company
has been a pioneer in developing powerful programs with user-friendly interfaces,
as opposed to some of the other packages that are based on line-commands.
PC’s were used more than any platforms, indicating, again, the small nature
of GIS operations in archaeology.
Figure 4.8
Figure 4.9
Figure 4.10
Answers to the first question in this category emphasized the fact that
site-based analyses using GIS are in their infancy (figure 4.11). Region
based analyses have dominated the use of GIS. Figure 4.12 show the type
of archaeological applications that GIS was focused on. Determining the
type of applications used was rather important because it showed whether
the participants are making the most of the tool. While GIS is undoubtedly
a powerful tool and has functions that cater to very complex modelling
needs, it is still a tool that is driven by market demands (Demers, 1997,
p 181), often of a non-archaeological nature. GIS is fundamentally a set
of incomplete modules that can only be expanded upon and effectively utilized
by adding one’s own algorithms through programming languages. Without this
functionality, the user is left with the default capabilities of the software,
which leaves the software makers in control of the shape of future research.
Only 22 of the 140 participants used their own algorithms, and of these
22 only 14 listed them as one of their most successful applications (figure
4.13 and 4.14).
The scope of the software analyses used was wide, and showed that the participants made use of most of the features available in their GIS packages (figure 4.15). Problematic operations were by far data collection, data conversion, and data compatibility (figure 4.16 and 4.17). Data sources showed a high frequency of manual digitizing and database sources. This could be a result of directly imputing data in a computer database in the field, or exporting an existing database into a GIS readable format. Internet downloads were among the least frequent source of data (figure 4.18).
Figure 4.11
Figure 4.12
Figure 4.13
Figure 4.14
Figure 4.15
Figure 4.16
Figure 4.17
Figure 4.18
4% of the participants thought that the simplicity of GIS software limited their ability to apply their models. 7% stated that GIS complexity reduced their ability to apply their models. 50% believed that GIS opened their mind to more expressive models, while 15% expressed that GIS did not change the way their models were designed. 23% decided that none of the above answers applied to them and finally 1% did not answer the question. (figure 4.19). The results reinforced the idea that GIS is more than a tool. It is certainly not a simple tool because it is not limiting anyone’s imagination in terms of analysis. If it is restrictive in a way, it is the software’s complexity that keeps people from implementing what they have in mind. Nonetheless, it is clear that GIS has significantly impacted the users’ spatial thinking as half of the participants thought GIS opened their mind to more ideas. Such responses are typical of the introduction of a new technology to a discipline.
A= The software's simplicity limited my ability to
apply my models.
B= The software's complexity reduced my ability to
apply my models.
C= The software opened my mind to more expressive
models.
D= The software did not change the way my models were
designed or implemented.
E= None of the above statements apply.
Figure 4.19
Data collection was the aspect of GIS that the participants were most familiar with (figure 4.20). 49% of the participants stated that they had modified their research to account for accuracy issues having to do with how the data was collected. 37% have considered the impact on their research. 10% are familiar with it but have not considered its impact. Finally 4% were not familiar with the problems the issue could present.
1= Not familiar with it
2= Familiar with it, but have not considered its impact
on my research
3= Have considered its impact on my research
4= Have modified my research to account for it.
Figure 4.20
The U.S. Geological Survey introduced the spatial metadata standards in an effort to increase the quality of digital data. The document contains rigorous parameters about information that should accompany any shared spatial data. Only 6% of the participants would alter their research according to whether their data meets USGS metadata standards. 43% were not familiar with the issue. 29% were, but have not considered its impact on their research, and a scant 21% have considered how it would impact their research (figure 4.21).
1= Not familiar with it
2= Familiar with it, but have not considered its impact
on my research
3= Have considered its impact on my research
4= Have modified my research to account for it.
All maps contain errors. There is no ultimate spatial database that can serve as a comparison standard for other digital maps (Dobson, 1994). Although one has to do their best to minimize the introduction of new errors during map analysis, it is wise to test the data for inconsistencies already present (Loberger and Foresman, 1993). Many tests have been devised to determine the accuracy of spatial data. The simplest way is to use statistical models to test for the occurrence of normal distributions. For example, one might correlate the vertical location of points with horizontal coordinates (Bolstad & Smith, 1992). In other cases, GPS has been used to test the accuracy of sample points (Thompson and Smith 1996). Others have added information about the quality of data as part of the attributes (Aspinall at al, 1993; Elmes et al., 1994; Dutton, 1989 Guptill, 1989). Logistical consistency checks and sensitivity analyses are also used, sometimes, by taking stratified random samples and applying inferential statistics to them (Vereign, 1989). 23% of the participants have modified their research to account for this issue. 29% have considered its impact and another 29% are familiar with the issue but have not acted regarding it. The remaining 19% were not familiar with data testing methods (figure 4.22).
1= Not familiar with it
2= Familiar with it, but have not considered its impact
on my research
3= Have considered its impact on my research
4= Have modified my research to account for it.
Figure 4.22
Generalizations are an inherent part of every map and are unavoidable. Similarly, the continuous nature of natural phenomena makes it difficult to map boundaries. Work is currently been done on developing fuzzy representations of spatial data (Wang et al., 1990). However, the concept is still not widely applied and the technology is still being developed. In the meantime, spatial data is still generalized according to subjective parameters. 24% of the participants have altered their research design to account for generalizations. 45% have considered their impact. 20% are familiar with them but were not clear on how they would affect their research. 11% were not familiar with the issues. (figure 4.23)
1= Not familiar with it
2= Familiar with it, but have not considered its impact
on my research
3= Have considered its impact on my research
4= Have modified my research to account for it.
Figure 4.23
Computer errors can be present either at the level of software glitches or in the way computers round off numbers as their precision allows. These errors are now of less concern to the GIS user because processing power has been growing at a phenomenal rate. 21% of the participants had modified their research to account for such errors. 45% considered the errors’ impact on their research. 20% are familiar with the errors but do not perceive an affect on their research. 19% were not familiar with them. (figure 4.24)
1= Not familiar with it
2= Familiar with it, but have not considered its impact
on my research
3= Have considered its impact on my research
4= Have modified my research to account for it.
Figure 4.24
As aforementioned, there are differences in the way raster and vector systems measure distance. Depending on the complexity of the spatial entities to be measured, errors could easily propagate through distance measurements. 20% of the participants had changed their research design to account for such differences. 37% did consider their impact. 26% were familiar with them but did not consider their impact on their research. 16% were not familiar with them.1% did not answer the question (figure 4.25).
1= Not familiar with it
2= Familiar with it, but have not considered its impact
on my research
3= Have considered its impact on my research
4= Have modified my research to account for it.
Figure 4.25
Overlays are relatively straightforward in a raster environment. In a vector based GIS on the other hand, the process is much more complicated. The overlays can produce sliver and faulty polygons that can skew the data. 26% of the participants modified their projects to factor in the difference in overlay results. 42% considered the impact. 25% were familiar with the issue but did not consider its impact on their research. 11% were not familiar with the difference in overlays (figure 4.26).
1= Not familiar with it
2= Familiar with it, but have not considered its impact
on my research
3= Have considered its impact on my research
4= Have modified my research to account for it.
4.5.8 Vector/Raster Conversions
As outlined in the first chapter, conversions between the two formats
can either lead to loss of detail or the addition of erroneous information.
22% of the participants did modify their research to account for this factor.
However, 48% considered its impact. 19% were familiar with it, while 11%
were not (figure 4.27).
1= Not familiar with it
2= Familiar with it, but have not considered its impact
on my research
3= Have considered its impact on my research
4= Have modified my research to account for it.
Only 14% of the participants actually modified their research to account
for the fact that different GIS packages use different algorithms to interpolate.
These results were not surprising because this issue is considered to be
one of the most inconspicuous GIS pitfalls. 30% were aware of its effect
on their research. 39% were, indeed, familiar with it, but did not consider
its impact on their research. 17% were not familiar with it(figure 4.28).
1= Not familiar with it
2= Familiar with it, but have not considered its impact
on my research
3= Have considered its impact on my research
4= Have modified my research to account for it.
This question was found to be ambiguous for some participants and 5% of the participants opted not to answer it. It may have been clearer if it were worded as "changing scales in GIS" or overlaying maps of different scales. Nonetheless, 95% of the participants did know what the question was referring to. 40% have modified their research to account for it. Another 40% considered its impact. 15% were familiar with it, but did not see how it can affect their research. None of the participants were unfamiliar with the issue (figure 4.29).
1= Not familiar with it
2= Familiar with it, but have not considered its impact
on my research
3= Have considered its impact on my research
4= Have modified my research to account for it.
Figure 4.29
Through the use of an online survey, the project sought to determine
the nature, impact, pitfalls, and degree of sophistication of GIS use in
archaeology. More than 70% of the archaeologists who took the survey, mostly
from the United States, the UK and Australia, had more than two years of
experience using GIS. 41% of the participants used the tool on a daily
basis and most of them chose or helped choose the GIS system they use.
Based on these numbers, it is not unreasonable to state that most of the
participants were fairly familiar with the tool. This was further exemplified
in the range of applications used. However, the one application that can
separate the novice from the expert is the use of one’s own algorithms
in GIS analysis. Only 22 participants of the 140 actually used their own
algorithms and only 14 listed algorithms as one of their most successful
applications. These14 were interestingly enough, the same people who made
the most of the software in terms of the scope of analyses conducted. They
also took the most classes and were involved in support groups.
The project also sought to determine the overall impact of GIS on archaeological
research and found there to be a consensus among most of the archaeologists
that GIS was a complex tool that revealed other models that could not otherwise
be implemented.
In assessing the archaeologists’ knowledge of common GIS pitfalls and
responses to them, the survey results demonstrated that many concepts which
directly or indirectly affect the quality of GIS output, are either still
unknown to most archaeologists or are not considered serious enough to
warrant changes in project designs.
GIS should be viewed against the larger historical context of tools
introduced to archaeology. Two earlier examples of such tools are Carbon
dating and statistics, which have reshaped the discipline of archaeology
in many ways. Like GIS, both methods have been subject to misuse and misunderstanding.
It is now common knowledge that C14 is not a precise measure of age as
it was believed to have been following its introduction in the late 1940’s,
but rather an approximation that is subject to a variety of factors, some
of which are still being assessed today (Taylor, 1987; Molto et al., 1997).
Similarly, "the archaeological literature is badly polluted with misusages
and outright abuses of statistical methods and theory" (Thomas, 1978, P:
231). If there is any lesson to be learned from these new tools, it is
that GIS should be approached prudently. GIS use within the discipline
of archaeology will not be complete without adding the dimension of error
as a natural extension of the tool. This can be accomplished through many
means. It is advisable to adhere to the already established standards that
govern the use and sharing of maps. Such standards include the National
Map Accuracy Standards (See Appendix A) and the Spatial Data Transfer standards
(See Appendix B). In addition, project managers need to define the quality
of source data based on the quality of the results needed. Finally, archaeologists
should not be dependent on GIS consultants who have no knowledge of the
subtleties of archaeology and often have different priorities. While the
archaeologist seeks quality results, the GIS consultant could be more concerned
with processing time and storage (Kvamme, 1996). The conflict could be
reasonably addressed if the archaeologists were more familiar with the
details of the tool.
Much like Statistics and C14, GIS complexity can easily lead one to
take the results it generates at face value. If one is not conversant with
its inner workings and pitfalls, one can easily produce and consume faulty
results.
National Map Accuracy Standards from the USGS
Agencies of the Federal Government, including the U.S. Geological Survey started work on map accuracy standards. In 1941, the U.S. Bureau of the Budget issued the "United States National Map Accuracy Standards, "which applied to all Federal agencies that produce maps. The standards were revised several times, and the current version was issued in 1947.
As applied to the U.S. Geological Survey 7.5-minute quadrangle topographic
map, the horizontal accuracy standard requires that the positions of 90
percent of all points tested must be accurate within 1/50th of an inch
(0.05 centimeters) on the map. At 1:24,000 scale, 1/50th of an inch is
40 feet (12.2 meters). The vertical accuracy standard requires that the
elevation of 90 percent of all points tested must be correct within half
of the contour interval. On a map with a contour interval of 10 feet, the
map must correctly show 90 percent of all points tested within 5 feet (1.5
meters) of the actual elevation.
Unavoidable Errors
There are certain kinds of errors in mapmaking that are unavoidable. Names and symbols of features and classification of roads or woodland are among the principal items that are subject to factual error. Mapmakers cannot apply a numerical value to this kind of information; they must rely on local sources for their information. Sometimes the information is wrong. Sometimes names change or new names and features are added in an area. U.S. Geological Survey cartographers and editors check all maps thoroughly and, as a matter of professional pride, attempt to keep factual errors to a practical minimum.
"Errors" resulting from selection, generalization, and displacement are necessary results of mapping complex features at reduced scales. In congested areas, large buildings may be plotted to scale and the smaller buildings may have to be omitted; in showing buildings of irregular shape, small wings, bays, and projections usually are disregarded, and the outline is show in general form. At map scale, it may not be possible to show each of several closely spaced linear features in its correct position. In such cases, one feature, such as a railroad, is positioned in its true location and others, such as parallel roads and rivers, are displaced the minimum amount necessary to make each symbol legible.
In 1958, the Survey began testing the accuracy of its maps systematically. Presently, accuracy testing is performed on 25 percent of the mapping projects at each contour interval as a method of controlling overall quality. It is rare for a 7.5- minute map to fail the test, but this happens on occasion.
In testing a map, U.S. Geological Survey experts select 20 or more well defined points; a typical point would be the intersection of two roads. Positions are established on the test points by field teams using sophisticated surveying techniques or by office personnel using photogrammetric methods to determine positions from aerial photographs. Vertical test are run separately to determine precise elevations. The mapped positions are checked against the field and/or photogrammetrically determined position results. If the map is accurate within the tolerances of the U.S. National Map Accuracy Standards, it is certified and published with the statement that it complies with those standards.
By such rigorous testing of some of its maps, the Survey is able to determine that its procedures for collecting map information are working well enough to assure a
high level of map accuracy.
United States National Map Accuracy Standards
With a view to the utmost economy and expedition in producing maps which fulfill not only the broad needs for standard or principal maps, but also the reasonable
particular needs of individual agencies, standards of accuracy for published
maps are defined as follows:
1.Horizontal accuracy. For maps on publication scales larger than 1:20,000, not more than 10 percent of the points tested shall be in error by more than 1/30 inch, measured on the publication scale; for maps on publication scales of 1:20,000 or smaller, 1/50 inch. These limits accuracy shall apply in all cases to positions of well-defined points only. Well-defined points are those that are easily visible or recoverable on the ground, such as the following: monuments or markers, such as benchmarks, property boundary monuments; intersections of roads, railroads, etc.; corners of large buildings or structures (or center points
of small buildings); etc. In general what is well defined will also
be determined by what is plottable on the scale of the map within 1/100
inch. Thus while the intersection of two road or property lines meeting
at right angles, would come within a sensible interpretation, identification
of the intersection of such lines meeting at an acute angle would obviously
not be practicable within 1/100 inch. Similarly, features not identifiable
upon the ground within close limits are not to be considered as test points
within limits quoted, even though their positions may be scaled closely
upon the map. In this class would come timber lines, soil bound, etc.
2.Vertical accuracy, as applied to contour maps on all publication scales,
shall be such that not more than 10 percent of the elevations tested shall
be in error more than one-half the contour interval. In checking elevations
taken from the map, the apparent vertical error may be decreased by assuming
a horizontal displacement within the permissible horizontal error for a
map of that scale.
3.The accuracy of any map may be tested by comparing the positions of
points whose locations or elevations are shown upon it with corresponding
positions as determined by surveys of a higher accuracy. Tests shall be
made by the producing agency, which shall also determine which of its maps
are to be tested, and the extent of such testing.
4.Published maps meeting these accuracy requirements shall note this
fact in their legends, as follows: "This map complies with National Map
Accuracy Standards."
5.Published maps whose errors exceed those afore-stated shall omit from
their legends all mention of standard accuracy.
6.When a published map is a considerable enlargement of a map drawing
(manuscript) or of a published map, that fact shall be stated in the legend.
For example, "This map is an enlargement of a 1:20,000-scale map drawing,"
or "This map is an enlargement of a 1:24,000 scale published map."
7.To facilitate ready interchange and use of basic information for map
construction among all Federal mapmaking agencies, manuscript and published
maps, wherever economically feasible and consistent with the use to which
the map is to be put, shall conform to latitude and longitude boundaries,
being 15 minutes of latitude and longitude, or 7 1/2 minutes, or 3 3/4
minutes in size.
A copy of this document can be found in:
http://www.usgs.gov/fact-sheets/map-accuracy/map-accuracy.html
standards for digital geospatial metadata
Where to Obtain Copies. Copies of this publication are available
from the Federal Geographic Data Committee Secretariat, in care of the
U.S. Geological Survey, 590 National Center, Reston, Virginia 22092; telephone
(703) 648-5514; facsimile (703) 648-5755; Internet gdc@usgs.gov.
The text also is available by anonymous File Transfer Protocol (anonymous
FTP) server fgdc.er.usgs.gov.
Overview of the Standard Data Transfer Standards (SDTS)
The SDTS base specification (Parts 1,2 and 3) describes the underlying
conceptual model and the detailed specifications for the content, structure,
and format for exchange of spatial data. Additional parts (4 and up) are
added as profiles each of which defines specific rules and formats for
applying SDTS for the exchange of particular types of data.
PART 1 - Logical Specifications
Part 1 consists of three main sections, which explain the SDTS conceptual
model and SDTS spatial object types, components of a data quality report,
and the layout of all SDTS modules.
PART 2 - Spatial Features
Part 2 contains a catalogue of spatial features and associated attributes.
This part addresses a need for definition of common spatial feature terms
to ensure greater compatibility in data transfers. The current version
of Part 2 is limited to small- and medium-scale spatial features commonly
used on topographic quadrangle maps and hydrographic charts.
PART 3 - ISO 8211 Encoding
This part explains the use of a general-purpose file exchange standard,
ISO 8211, to create SDTS filesets (i.e. transfers).
PART 4 - Topological Vector Profile
The Topological Vector Profile (TVP) is the first of a potential series
of SDTS profiles, each of which defines how the SDTS base specification
(Parts 1, 2, and 3) must be implemented for a particular type of data.
The TVP limits options and identifies specific requirements for SDTS transfers
of data sets consisting of topologically structured area and linear spatial
features.
PART 5 - Raster Profile and Extensions
The Raster Profile is for 2-dimensional image and gridded raster data.
It permits alternate image file formats using the ISO Basic Image Interchange
Format (BIIF) or Georeferenced Tagged information File Format (GeoTIFF).
PART 6 - Point Profile
The Point Profile contains specifications for use with geographic point
data only, with the option to carry high precision coordinates such as
those required for geodetic network control points. This profile is a modification
of Part 4, the Topological Vector Profile, and follows many of the conventions
of that profile.
A copy of this document can be found in:
http://mcmcweb.er.usgs.gov/sdts/standard.html
Aangeenbrug T. R. 1991
"A Critique of GIS." In GIS: Principles and Applications. Edited by
David J Maguire, Michael F. Goodchild and David W. Rhind. Harlow, Essex,
England: Longman Scientific & Technical; New York: Wiley. Pp: 101-107
Aldenderfer, M.S., and Herbert D. G. Maschner 1996.
Anthropology, space, and Geographic Information Systems. Oxford
University Press, New York.
Aldenderfer, M. 1987
"Assessing the Impact of Quantitative Thinking on Archaeological Research:
Historical and Evolutionary Insights" in Quantitative Research in Archaeology:
Progress and Prospects edited by Mark S. Aldenderfer. Newbury Park,
California: Sage Publications.
Allen, K.M.S., S.W. Green & E.B.W. Zubrow 1990
"Interpreting Space." In Allen, K.M.S., S.W. Green & E.B.W. Zubrow
(eds) Interpreting Space: GIS and Archaeology. London: Taylor &
Francis. pp. 383-387.
Allen, Kathleen M. S. 1990
"Manipulating Space: a Commentary on GIS Applications." In Allen, K.M.S.,
S.W. Green & E.B.W. Zubrow (eds) Interpreting Space: GIS and Archaeology.
London: Taylor & Francis. pp. 197-200.
Allen, Kathleen M. S. 1990b
"Modelling Early historic trade in the eastern Great Lakes using GIS."
In Allen, K.M.S., S.W. Green & E.B.W. Zubrow (eds) Interpreting
Space: GIS and Archaeology. London: Taylor & Francis. Pp:319-329
Aspinall, R. J. & Miller, D. R. & Richman, A. R. 1993
"Data Quality and Error Analysis in GIS: Measurement and Use of Metadata
Describing Uncertainty." In Spatial Data Proceedings of the Thirteenth
Annual ESRI User Conference, Vol. 2, pp. 279-290.
Bolstad, Paul V. & Gessler, Paul & Lillesand, Thomas M. 1990
"Positional uncertainty in manually digitized map data." International
Journal of Geographical Information Systems, 4(4), pp. 399-412.
Bolstad, Paul V. & Smith James L. 1992
"Errors in GIS, Assessing Spatial Data Accuracy." Journal of Forestry,
November 1992, Vol. 90. Number 11.
Burrough, P.A. 1986
Principles of Geographical Information Systems. Oxford. Clarendon
Press.
Carmichael, David L. 1990
"GIS predictive modelling of prehistoric site distribution in Central
Montana." In Allen, K.M.S., S.W. Green & E.B.W. Zubrow (eds) Interpreting
Space: GIS and Archaeology. London: Taylor & Francis. (Get page)
Chrisman, Nicholas R. 1989
"Modeling error in Overlaid Categorical Maps." In Goodchild, Michael
& Gopal, Sucharita (eds) The Accuracy of Spatial Databases.New
York: Taylor & Francis. Pp. 21-43.
Chrisman, Nicholas R. 1991
"The Error Component in Spatial Data." In GIS: Principles and Applications.
Edited by David J Maguire, Michael F. Goodchild and David W. Rhind. Harlow,
Essex, England: Longman Scientific & Technical; New York: Wiley.
Christopherson, G.L., D.P. Guertin, M.R. Kunzmann et al., 1993
"Comparison of Interpolation Algorithms for Digital Elevation Model
Generation and Subsequent Viewshed Analysis." In Rowlands, P.G., C. Van
Riper III & M.K. Sogge (eds) Proceedings of the First Biennial Conference
on Research in Colorado Plateau U.S. National Parks. Transactions and
Proceedings, Series 10. Denver: U.S. National Park Service, Natural Resources
Publication Office. pp. 226-233.
Christopherson, Gary L.; Guertin, Phillip D.; Borstad, Karen A. 1996
"GIS and Archaeology: Using Arc/Info to Increase Our Understanding of
Ancient Jordan". Proceedings of 1996 ESRI User Conference.
Clarke, Keith C. 1997
Getting Started with Geographic Information Systems. Prentice Hall Series in Geographic Information Science. New Jersey: Simon & Shuster.
Davis, W. F and Simonett. S. D. 1991
"GIS and Remote Sensing." In GIS: Principles and Applications.
Edited by David J Maguire, Michael F. Goodchild and David W. Rhind. Harlow,
Essex, England: Longman Scientific & Technical; New York: Wiley. Pp:
191-213.
Demers, Michael N. 1997
Fundamentals of Geographic Information Systems. New York: John Wiley & Son.
Dobson, Jerome E. 1994
"Face the ground truth about accuracy assessment." GIS World,
7(11), pp. 32-34.
Duecker, K.J. 1979
"Land resource information systems: a review of fifteen years experience." Geo-processing, vol. 1, no. 2, pp. 105-128.
Dunn, R., Harrison, A. R. & White, J. C. 1990
"positional accuracy and measurement error in digital databases of
land use: an empirical study." International Journal of Geographical
Information Systems, 4(4), pp. 385-398.
Dutton, Geoffrey 1989
"Modeling Locational Uncertainty Via Hierarchical Tesselation." In Goodchild,
Michael & Gopal, Sucharita (eds) The Accuracy of Spatial Databases.
New York: Taylor & Francis. pp. 125-140.
Dutton Geoffrey 1992
"Handling Positional Uncertainty in Spatial Databases." Proceedings
of the 5th International Symposium on Spatial Data Handling, Vol. 2,
pp. 460-469.
Ebert James I. and Kohler Timothy A. 1990
"The Theoretical Basis of Archaeological Predictive Modelling and a
Consideration of Appropriate Data-Collection Methods." In
Allen, K.M.S., S.W. Green & E.B.W. Zubrow (eds) Interpreting Space:
GIS and Archaeology. London: Taylor & Francis. (get page)
Elmes, Gregory A.; Cai, Guoray & Twery, Mark J. 1994
"Estimating Spatial Data Error by Inference." Sixth International
Symposium on Spatial Data Handling, Vol. 1, pp. 254-276.
Fotheringham Stewart A. 1989
"Scale-independent Spatial Analysis." In Goodchild, Michael & Gopal,
Sucharita (eds) The Accuracy of Spatial Databases.New York: Taylor
& Francis. Pp. 221-228.
Gaffney, V. & M. Van Leusen 1995
"GIS and Environmental Determinism: A Parallel Text." In Lock, G.R.
& Z. Stancic (eds) Archaeology and Geographic Information Systems:
A European Perspective. London: Taylor & Francis.
Gaffney, V., Z. Stancic & H. Watson 1996
"The Impact of GIS on Archaeology: a Personal Perspective." In Aldenderfer,
M. & H.D.G. Maschner (eds) Anthropology, Space and Geographical
Information Systems. New York: Oxford University Press. Pp. 210-226.
Goodchild, Michael F. 1989
"Modeling Error in Objects and Fields." In Goodchild, Michael &
Gopal, Sucharita (eds) The Accuracy of Spatial Databases.New York:
Taylor & Francis. Pp. 107-113.
Goodchild, Michael F. 1991
"Spatial Analysis with GIS: Problems and Prospects." GIS/LIS ’91
Proceedings, 28 October-November 1991. The Inforum, Atlanta, Georgia.
Gopal, Sucharita & Goodchild, Michael F. 1989
"Preface", In Goodchild, Michael & Gopal, Sucharita (eds) The
Accuracy of Spatial Databases.New York: Taylor & Francis.(get page)
Griffith, Daniel A. 1989
"Distance Calculations and Errors in Geographic Databases." In Goodchild, Michael & Gopal, Sucharita (eds) The Accuracy of Spatial Databases.New York: Taylor & Francis. Pp. 81-90.
Guptill, Stephen C. 1989
"Inclusion of Accuracy Data in a Feature Based, Object Oriented Data Model." In Goodchild, Michael & Gopal, Sucharita (eds) The Accuracy of Spatial Databases.New York: Taylor & Francis. Pp. 91-97.
Harris, T.M.; Lock G.R 1996
"Toward an Evaluation of GIS in European Archaeology: the Past, Present and Future of Theory and Application." In Aldenderfer, M. & H.D.G. Maschner (eds) Anthropology, Space and Geographical Information Systems. New York: Oxford University Press. pp. 350-365.
Kathleen M.; Allen S. 1990
"Modelling Early Historic Trade in the Eastern Great Lakes Using Geographic
Information Systems." In Allen, K.M.S., S.W. Green & E.B.W. Zubrow
(eds) Interpreting Space: GIS and Archaeology. London: Taylor &
Francis. pp. 317-329.
Kennedy, Susan 1989
"The Small Number Problem and the Accuracy of Spatial Databases." In Goodchild, Michael & Gopal, Sucharita (eds) The Accuracy of Spatial Databases. New York: Taylor & Francis. Pp. 187-196.
Kvamme, K.L. & T.A. Kohler 1988
"Geographic Information Systems: Technical Aids for Data Collection, Analysis and Display." In Judge, W.J. & L. Sebastian (eds) Quantifying the Present and Predicting the Past: Theory, Method and Application of Archaeological Predictive Modelling. Denver: U.S. Department of the Interior, Bureau of Land Management. pp. 493-548.
Kvamme, K.L. 1989
"Geographic Information Systems in Regional Archaeological Research
and Data Management." In Schiffer, M.B. (ed.) Archaeological Method
and Theory 1. Tucson: University of Arizona Press. pp. 139-203.
Kvamme, K.L. 1990
"GIS Algorithms and Their Effects on Regional Archaeological Analysis."
In Allen, K.M.S., S.W. Green & E.B.W. Zubrow (eds) Interpreting
Space: GIS and Archaeology. London: Taylor & Francis. pp. 112-126.
Kvamme, K. L. 1996
"A View From Across the Water: the North American Experience in Archaeological
GIS." In Aldenderfer, M. & H.D.G. Maschner (eds) Anthropology, Space
and Geographical Information Systems. New York: Oxford University Press.
pp. 1-14.
Lee, Jay 1996
"Digital Elevation Models: Issues of Data Accuracy and Applications."
Prodceedings
of the 1996 ESRI Conference, Palm Springs, California.
Litton, Adrien 1996
"Quantifying Scanner Accuracy." Proceedings of 1996 ESRI User Conference.
Loberger, Dale A. & Foresman, Timothy W. 1993
"Managing Error-An Approach to Digital Spatial Accuracy." Proceedings
of the Thirteenth Annual ESRI User Conference, Vol. 2, pp. 291-299.
Lodwick, Weldon A. 1989
"Developing Confidence Limits on Errors of Suitability Analyses in Geographic
Information Systems." In Goodchild, Michael & Gopal, Sucharita (eds)
The
Accuracy of Spatial Databases.New York: Taylor & Francis. pp. 69-78.
Madry, Scott L. H. and Crumley, Carole L. 1990
"An applicaation of remote sensing and GIS in a regional archaeological
settlement pattern analysis: the Arroux River Valley, burgundy, France."
In Allen, K.M.S., S.W. Green & E.B.W. Zubrow (eds) Interpreting
Space: GIS and Archaeology. London: Taylor & Francis. (get page.)
Maffini, Giulio; Arno, Michael & Bitterlich, Wolfgang 1989
"Observations and Comments on the Generation and Treatment of Error
in Digital GIS Data." In Goodchild, Michael & Gopal, Sucharita (eds)
The
Accuracy of Spatial Databases.New York: Taylor & Francis. pp. 55-67.
Marble, D.F. 1990
"The Potential Methodological Impact of Geographic Information Systems
on the Social Sciences." In Allen, K.M.S., S.W. Green & E.B.W. Zubrow
(eds) Interpreting Space: GIS and Archaeology. London: Taylor &
Francis. pp. 9-21.
Maguire, J. D. 1991
"An Overview and Definition of GIS" in GIS: Principles and Applications. Edited by David J Maguire, Michael F. Goodchild and David W. Rhind. Harlow, Essex, England: Longman Scientific & Technical; New York: Wiley.
Molto, Eldon J.; Steward Joe D and Reimer, Paula J. 1997
"Problems in Radiocarbon Dating Human Remains from arid Coastal Areas:
an Example from the Cape Region of Baja, California." American Antiquity,
Vol. 62 (3). pp: 489-507.
Olsen, W. J. 1985
"Application of Space-Born Remote Sensing in Archaeology", the University
of Arisona Remote Sensing Newsletter, Fall-Winter 1985
Potts, Richard; Jorstad, Tom & Cole, Daniel 1996
"The Role of GIS in the Interdisciplinary Investigations at Olorgesailie,
Kenya, a Pleistocene Archaeological Locality". In Aldenderfer, M. &
H.D.G. Maschner (eds) Anthropology, Space and Geographical Information
Systems. New York: Oxford University Press. pp. 202-213.
Savage, Stephen H. 1990
"Modelling the late Archaic Social Landscape." In Allen, K.M.S., S.W.
Green & E.B.W. Zubrow (eds) Interpreting Space: GIS and Archaeology.
London: Taylor & Francis. pp. 330-355.
Sebastian, Lynne and Judge, James W. 1988
"Predicting the past: Correlation, explanation, and the use of archaeological models."
In Quantifying the present and predicting the past: theory, method,
and application of archaeological predictive modeling edited by Jeffrey
H. Altschul et al. For the U.S Department of the Interior, Bureau of Land
Management, Denver, Colorado. Chapter 1.pp 1-17.
Taylor, R. E. 1987
Radiocarbon Dating: an Archaeological Perspective. New York:
Academic Press.
Theobald, David M. 1989
"Accuracy and Bias Issues in Surface Representation." In Goodchild,
Michael & Gopal, Sucharita (eds) The Accuracy of Spatial Databases.New
York: Taylor & Francis. Pp. 99-106.
Thomas, David H. 1978
"The Awful Truth about Statistics in archaeology." American Antiquity,
43 (2): Pp: 231-244.
Thompson , Stephen R. & Smith , Creighton 1996
"Using GPS to Assess the Accuracy of Digital Ortho Photo Files and the
USGS Public Land Survey System Database for a Kansas County." Proceedings
of 1996 ESRI User Conference.
Tobler, W.R. 1959
"Automation and cartography." Geographical Review, vol. 49, pp. 526-534.
Vereign, Howard 1989
"Error Modeling for the map overlay operation." In Goodchild, Michael
& Gopal, Sucharita (eds) The Accuracy of Spatial Databases.New
York: Taylor & Francis. Pp. 3-18.
Walsh, Stephen J 1990
"User Consideration in Landscape Characterization." In Allen, K.M.S.,
S.W. Green & E.B.W. Zubrow (eds) Interpreting Space: GIS and Archaeology.
London: Taylor & Francis. (get page)
Wang, Fangju; Brent, Hall G. & Subaryono 1990
"Fuzzy information representation and processing in conventional GIS
software: database design and application." International Journal of
Geographical Information Systems, 4(3), pp. 261-283.
Warren, R.E. 1990a
"Predictive Modelling in Archaeology: A Primer." In Allen, K.M.S., S.W.
Green & E.B.W. Zubrow (eds) Interpreting Space: GIS and Archaeology.
London: Taylor & Francis. pp. 90-111.
Warren, Robert E. 1990b
"Predictive modeling of Archaeological Site location: a Case Study in
the Midwest." In Allen, K.M.S., S.W. Green & E.B.W. Zubrow (eds) Interpreting
Space: GIS and Archaeology. London: Taylor & Francis. pp. 201-215.
Zubrow, Ezra B. W. 1990a
"Modelling and Prediction with Geographic Information Systems: a Demographic
Example From Prehistoric and Historic New York." In Allen, K.M.S., S.W.
Green & E.B.W. Zubrow (eds) Interpreting Space: GIS and Archaeology.
London: Taylor & Francis. pp. 307-313.
Zubrow, E.B.W. 1990b
"Contemplating Space: A Commentary on Theory." In Allen, K.M.S., S.W. Green & E.B.W. Zubrow (eds) Interpreting Space: GIS and Archaeology. London: Taylor & Francis. pp. 67-72.
