Automated,
computer-assisted and near-manual cartography focus on the visual map,
which has traditionally been the principal product of cartographic
activities. A new focus of activity within digital cartography is
given prominence in Guptill and Starr's (1984) definition of
cartography as "an information transfer process that is centred about
a spatial database which can be considered, in itself, a multifaceted
model of geographic reality. Such a spatial database then serves as
the central core of an entire sequence of cartographic processes,
receiving various data inputs and dispensing various types of
information products". This definition of the subject makes the
spatial database the focus of the subject and regards the traditional
focus of activity, the visual map, as one amongst a range of
information products as illustrated in Figure 2.
Both in theory
and in practice the map database has become the ultimate reference map
- the digital map, which is a commercial product in its own right. The
archives of national mapping agencies, including the United States
Geological Survey (Starr, 1986) and the Ordnance Survey of Great
Britain will in future be held in digital rather than analog form and,
in Britain, users will be able to purchase on demand and to their own
specification Customer Plots from Current Data (CPCD) from high street
retail outlets (Haywood, 1987).
A number of applications need to visualize these data. Visualization
is especially necessary when cartographic/topographic data are
minimally structured and held as raster- or video-scanned images as in
the Domesday Project (Openshaw and Mounsey, 1986), British Telecom's
VMIS system, or in the map-based information system adopted by English
Heritage (Clubb and Hilder, 1989). Trade literature suggest that the
previous map sheet based database is being replaced by spatially
continuous maps to allow users to zoom and pan smoothly over an
extensive area. Users own data are usually added as 'layers', which
are superimposed on top of the base maps on display for visual
analysis and comparison. Users can randomly retrieve and display such
digital and analogue maps stored on optical disc. The age of the
electronic atlas, anticipated by Monmonier (1981), is already here.
Although, the user interface allows users to take measurements off
these maps, as in the BBC Domesday system, this can be extremely
inconvenient as found by Maling (1989). Similarly, so-called spaghetti
digitised vectors, which have not been topologically structured, tend
to be used as background maps. Until relatively recently Ordnance
Survey large-scale digital maps were not topologically structured
(Haywood, 1987; Smith, 1987). Also map sheets can overlap and,
even where they do not, automatic edge-matching of digitised sheets is
not a trivial task. The Ordnance Survey of Northern Ireland has been
able to supply topologically structured and edge-matched background
data since 1987 (Brand, 1988).
Limited amounts of data may be captured in a topologically structured
form, using one of many de-facto formats, such as the widely used
GIMMS segment format (Waugh and McCalden, 1983). These stipulate that
lines should not
cross except at junction points, called nodes. The user explicitly has
to input these nodes and the codes for the hierarchy of areal objects
to the left and right of boundary lines. This places an unnecessary
burden on the
user in order to ease computer processing. It is error prone and
contradicts the requirement for user friendliness in human-computer
interaction. Moreover, it greatly increases the cost of bulk data
capture.
Consequently, large scale analog to digital conversion of maps rely
either on raster or video scanning or the digitising of only the
point, line and text features in a relatively free format, which need
not conform to topologic standards. This approach to vector digitising
produces a feature-, rather than an object-, based data model. In an
object recording model all information pertaining to some object of
interest, such as an individual address or highway, may be retrieved
in a systematic and complete manner. For example, it would be possible
to retrieve by name the complete
set of boundaries of an areal object, with all the detached parts
and/or holes within it (Mitchell et al, 1977; Wade et al, 1986; Kirby
et al, 1989).
In a
feature-based model this information is associated with fragmented
features, which are scattered within the database, and may remain
implicit. Fragmented features tend to be displayed in an ad-hoc order;
this not only
detracts from the quality of the map user interface in human-computer
interaction, but it also has performance implications at the levels of
both software (for example, the number of polyline calls) and hardware
(increase
in pen-up and pen-down movements). Also, area fill and the accurate
calculation of distances, areas and volumes require the explicit
recording of line and area topology within the database. Experience
has shown that an
object-recording database expedites the update of both geometric and
attribute data and the management of so-called change (historical)
data and their chronology.
Although proprietary software has been developed for geometrically
adjusting digitised vectors and establishing line connectivity (Ericksen,
1984; ESRI, 1985), the results are not entirely satisfactory
(Visvalingam and Sekouris,
1989). But, once the vectors have been transformed into a link and
node connected structure, it is possible to extract the area topology
by geometric processing (Kirby et al, 1989) and infer automatically a
great deal of implicit information about mapped entities and make this
explicit (de Simone, 1986; Wade et al, 1986).
Durham (1989, p 27) described raster-to-vector conversion and
autotracing, which derives an internal structured model of a drawing
from the display image, as "an innovation in human-computer
interaction, a first step towards
the intelligent sketchpad which has some basic knowledge of how the
human eye sees shapes". Wade (1986) and de Simone (1986) demonstrated
how the computer could automatically recognise administrative
hierarchies and user-perceived objects, such as houses, minor buildings, roads and
pavements respectively.
Digital models of reality are more
useful when they represent topographic, rather than cartographic,
entities. The Ordnance Survey 1:50 000 Small Scales Demonstrator (SSD)
Database of Sheffield and Doncaster (Sheet 111)
was captured from analog maps. Cartographic databases tend to be
incomplete because of the limitations imposed by the paper medium. To
avoid clutter, only high priority features and objects, such as
highways and administrative
boundaries, are recorded in their entirety. Other features, such as
river networks, remain incomplete, especially in built-up areas. Also,
often on a map, the direction of flow of rivers is implied by contours
and line width and the process of digitising does not make such
information explicit. Consequently, the Institute of Hydrology
is using the Digital Terrain Model, produced by the Military Survey
from height information captured from 1:50 000 maps, to fill in by
simulation such missing information and extract a complete
flow-directed network of rivers and their catchment boundaries (Moore,
personal communication). Peucker and Douglas (1975), Mark (1984) and
Marks et al (1984) have also used DTMs for identifying drainage basin
catchment boundaries and river networks and have described the problem
posed by 'pits' in the DTM.
The use of digital maps within the rapidly developing and commercially
important area of Geographical Information Systems (GIS) is greatly
facilitated by the recording of user-perceived objects. The objects
recorded in the SSD Database were manually chained. Clearly, this is
not practicable for the identification of all objects in Great
Britain. Manual processes are also error prone and tend to gloss over
any existing errors. Visvalingam and Sekouris (1989) suggested that
area-objects in the SSD database could be automatically recognised if
candidate seeds for the
primitive regions (Kirby et al, 1990) could be extracted by scanning
the colour separations used in printing. Structured digital map data
were used by Mason et al (1988) for segmentation and interpretation of
remotely sensed
images. Early automated, computer-assisted and near-manual cartography
were mainly concerned with computerising map generation for the human
information processing system. Advances in digital mapping have
already stimulated
research into computer interpretation of maps based on heuristics so
as to make explicit the required topologic and semantic information in
a cost-effective way; digital mapping has launched automated
cartography into a new
era.
A major difference between Digital Cartography and CAD lies in the
greater variety, complexity and sheer quantity of cartographic and
thematic data, the varying constraints on their interpretation and
graphic display and in
the variety of conceptual and pragmatic spatial data models in use.
The relational database model is increasingly used for managing
thematic attribute data but the topological data has been handled by
specialised, often ad-hoc, sub-systems. For example, in ARC/INFO (ESRI,
1985), the attribute data are handled by INFO, a relational DBMS,
whilst the topological data are handled by the ARC system. ARC is
based on the GIRAS model (Mitchell et al, 1977) and the DLG-3 data
format (Allder and Elassal, 1984) used by the United States Geological
Survey. These were in turn based
on early work on data structures by Peucker and Chrisman (1975) and on
topology by Corbett (1979). The United States Bureau of the Census,
whose Dual Independent Map Encoding (DIME) model is widely cited for
its
pioneering role, has now switched to the TIGER model (Marx, 1986).
More recent studies have shown that static, topologically structured
archives may be managed using the relational model ( Kirby et al,
1987;Smith and Groom, 1987; van Roessel, 1987). Reference maps,
whether analog or digital, do not attempt to make all geometric,
topologic and semantic relationships explicit; a great deal of
information remains implicit and is derived by analytical and
interpretative processes. Commercially available DBMS need to be
extended, at the very least, to undertake these processes in an
efficient and user-friendly manner (Frank, 1982; van Roessel and
Fosnight, 1985; Guptill, 1986). Responding to queries about complex
spatial objects in a large database is an inherently difficult
computational task.
Although the relational model has some known and lamented weaknesses
for managing topological data, the scope for developing systems on
modest system configurations and later porting them onto dedicated
database machines is
appealing. The mapping of static data does not require the same level
of complexity as expected within a system for interactively editing
and merging clients' data and the base map. The abstraction of minimal
conceptual models and functionally appropriate pragmatic models of
both static and dynamic 2D spatial data is still the subject of
research (Visvalingam and Sekouris, 1989). Loudon (1986) gives a
flavour of the problems inherent in modelling complex geological data
in 3D.
Given the size of integrated spatial
databases, the efficiency of storage, retrieval and performance are
major considerations. Knowledge-based techniques are being
investigated to guide search (Smith et al, 1987). A
great deal of research has been focused on spatial queries - not just
for windowing but also for supporting relational operations on
topological data and for polygon handling (see Burrough, 1986a;
Guptill, 1986), as undertaken
within ARC/INFO (ESRI, 1985; Green et al, 1985). The problems of
polygon intersection and spatial indexing are not unique to digital
cartography. Bounding rectangles, or envelopes, have been widely
used in range and
quadtree-based searches (Abel and Smith, 1983; Samet, 1988). Ballard
(1981) and Shneier (1981) have also investigated strip trees and edge
quadtrees respectively for representation of linear features.
Research effort is being directed at developing more efficient storage
and search strategies particularly since there is now considerable
interest in the integrated management of vector and raster data, for
example in the
conjoint use of topographic data and satellite images. This is impeded
largely by the limitations of existing designs for vector data. Early
solutions involved the conversion or discretisation of vector data
into a common raster form so as to facilitate comparisons based on
proximity and boolean algebra (Tomlin et al, 1983). Such an approach
facilitates logical and statistical analysis but impedes structural
analysis since the topology of the data becomes blurred. Even the
widely cited Peuquet's vaster encoding is subject to this limitation
since it represents vectors as Freeman chain codes with a resolution
determined by the raster data (Peuquet, 1984).
More recently, others have investigated variants of two basic schemes
for linking vector and raster data whilst preserving the geometric
quality of vector data. Volatile spatial databases are modelled using
pointer-based
region quadtrees as proposed by Samet and co-workers (see for example,
Ibbs and Stevens, 1988) but static databases are represented more
compactly and indexed using pointerless linear quadtrees as proposed
by Gargantini (1982). Linear quadtrees offer considerable
advantages for compact encoding and efficient processing of image data
and, for spatial indexing, are superior to the more compact
two-dimensional run-length encoding (2DRE) proposed by
Lauzon et al (1985). The elegance of the quadtree structure has
captured the imagination of a number of researchers, who seek to fit
vectors into this scheme (see Diaz and Bell, 1986) or adaptations of
it (Kleiner and Brassel, 1986; Abraham, 1989; Newell and Theriault,
personal communication). But, quadtree indexing of vectors
involves exhorbitant overheads (see for example the statistics quoted
by Ibbs and Stevens, 1988) and there is insufficient published
evidence to commend the use of quadtrees with vector data at present.
Clearly there is a need for further research in this area.
The problem of integrated use of vector and raster data structures is
not unique to digital cartography and GIS. They are equally important
within other applications of computer graphics. The demand for direct
manipulation in human-computer interaction is directing research
effort towards improving the integrity and performance of screen
update when displaying vector data on low-cost raster refresh
terminals (Slater, 1986; Slater et al, 1989). Unless the requirements
of these various tasks are considered together, separate techniques
for essentially similar and related tasks will lead to unnecessary
complexity and inefficiency within digital cartography and GIS.
Thus, the digital map is not just a computer-readable file of map
data. Visvalingam (1989, p 30) proposed that the term implies "a
compact, structured, integrated and elegant representation of spatial
data and their aspatial attributes in a manner that facilitates rapid
inference and retrieval and speedy but error-free update of data. This
implies pre-processing and substantial restructuring of input data so that the
digital post-processing system may infer spatial forms, relationships
and patterns in a way which matches, and if possible surpasses, human
information processing capabilities". This definition excludes
uninterpreted raster and video-scanned images and spaghetti vectors,
despite their value and use as visual maps.
Much progress has been made towards the establishment of a common
vocabulary and standards for the exchange of topographic data at a
national level, even if not an international one (Ordnance Survey,
1989; DCDSTF, 1988). Also,
many applications need to relate thematic and statistical data from
different sources. The problems involved in linking data were outlined
in the Chorley Report (Department of Environment, 1987), in which a
number of useful recommendations were made to minimise these problems.
The quality of data or its fitness for purpose is another important
issue. One of the roles of the recently formed Association of
Geographic Information is in
influencing the establishment of standards.
3.3 Non-graphic products
based on Digital Maps
Graphic visualization of the data is
necessary in many, but not all, map-based applications. When a user is
confident of the quality and appropriateness of the data and of the
procedures for operating on them, highly structured databases enable
many routine applications to function without the previous requirement
for visual maps. Data need not be visualized as maps in order to
schedule, time and cost the cutting of grass verges, the cleaning of
streets or the salting of roads; to route vehicles using in-car
navigation systems or to check the mileage on travel expense claims; or, for computing cut-and-fill from terrain models in highway
engineering.
The digital map has become the ultimate reference map for 'map
reading', cartometric, other analytic, inferential and
cross-referencing purposes. In many local authority applications
the paper map was used as a means of
cross-referencing and accessing spatially related textual and numeric
information. Within a computerised environment, the base map need not
appear within the user interface. For example, development staff in a
local planning authority would ideally like to input a site address
and pull out the planning constraints which apply to it for input into
other pre-existing software. These constraints are normally recorded
'on', i.e. related by overlays to, a large-scale base map. As
expressed by Weights (1989, p9), "if the map is a necessary but
intermediate step in answering an enquiry,
then it makes for greater efficiency to the map users if that
intermediate step is hidden, and the results of the geographical
search displayed on a low cost conventional terminal". The lack of
need for visual maps in some
applications has led some to conclude that such activity falls within
the field of GIS rather than digital cartography. But, as explained by
Weights, the inter-linking of data is reliant upon use of fully
structured digital maps. Digital cartography provides the backbone of
many GIS.
3.4 Some Developments in Automated
and Computer-Assisted Cartography
There has been notable progress in automated cartography both with
respect to automatic interpretation, as described earlier, and in
visual map production. Many of the processes involved in the
production of the digital
version of the Ordnance Survey 1:625 000 scale Route Planner map were
automated (Hadley, 1987). Research into automatic name placement (for
example by Cook, 1986) had not progressed far enough in time to avoid
interactive positioning and scaling of names. The Ordnance Survey is
also using its home-grown system for in-house Automated Map Production
(AMP) and for producing maps to customer requirements. These systems
are not fully automated at present since manual checks and edits are
still necessary (Clayton, 1989).
The move towards automation has necessitated some re-design,
elimination of some content (Hadley, 1987) and modifications to the
graphic (Smith and Turner, 1987) to achieve results in the short term.
For example, the
conventional hachures have been replaced by special area fill
symbolism. Whilst in general, traditional cartographers accept
and endorse the semi-automated map products of the Ordnance Survey,
they are highly critical and
dismissive of the type of output produced by some others. However,
many applications do not need the quality of displays achieved by
professional cartographers even though some minimum standards are
expected. Some system
developers do, on the other hand, have the naive belief that map-based
information systems can be constructed by simply fronting a database
management system with software for computer graphics or CAD. Essinger
(1986) indicated the type of functions which should be included within
a CAD workstation for map design; and, map design is just one
requirement within digital cartography.
It is equally naive to assume that
poor map design is solely due to a lack of cartographic training.
Research into use of Bertin's (1983) cartographic theory within expert
systems (Mackinlay, 1986) is still in its infancy and even when
further advanced will be inadequate for fully automating design since
this would also require automatic symbol and name placement and
generalisation of displays (see earlier discussion on maps as
transformed representations and generalisations of reality). Research
on sub-tasks, such as line simplification, has yielded a number of
competing algorithms (McMaster, 1987a), including those based on the
idea of self-similarity (Muller, 1987), but no substantive theory
(despite the title of the paper by Peucker, 1975) to guide their use.
Objective mathematical evaluations of
some of these algorithms by McMaster (1987b) and Muller (1987) have
also been misleading (see critique by Visvalingam and Whyatt, 1989).
Whereas simplification of information and graphic design are often
undertaken simultaneously in manual generalisation, in digital
cartography there is a need to distinguish between information
generalisation and display generalisation; the former contributes to
digital mapping, the latter to visual mapping. Both types of
transformations are intellectually
demanding and involve a variety of subtasks and processes. It is
widely accepted that the long-term national objective of scale-free
mapping, recommended by both the Serpell (1979) and Chorley
Committees (DoE, 1987),
is not feasible given the current state of knowledge. Many
applications use maps of the same entities compiled at a variety of
scales with different content and symbolism. Route planning and the
modelling of the hydrological
cycle are mainly interested in the topologically connected network of
centre lines of linear objects. Many land information systems, on the
other hand, focus on the land parcel and are concerned with the areal
extent of such
linear objects. A number of users, such as in local government, need
both detailed and simplified representations of the same objects. At
present, scale-related requirements are met by multiple definitions of
the same areas. Within the Hydrographic Department, important points
at each scale are manually identified and transferred onto larger
scales so
that only the largest scale map is vector digitised. Each point is
thus only digitised once. The scale of display appropriate for each
feature is also recorded within a scale-integrated database (Drinkwater
and Fielding, 1986). Systems based on raster/video images zoom by
retrieving images captured at the next larger scale. Other researchers
are manually integrating road networks captured at different scales
for route planning (Chapman, personal communication) and are
investigating database designs for supporting scale-free mapping
(Abraham, 1989). Brassel and Weibel (1988) refer to earlier reviews
and describe more recent work on the theme of automatic
generalisation.
4. SOME EMERGING TRENDS
Whereas established trends tend to concentrate on the generation
of maps and map-related products, the emerging trends focus attention
on the interactive use of maps within both routine and creative
applications.
4.1 Scientific Visualisation - Realism versus Symbolism
Advances in spatial data processing and in computer graphics have
facilitated the visualisation of three dimensional geographic
phenomena, such as surface terrain and cover (Davis and McCullagh,
1975; Burrough, 1986a; Maver, 1987). Some applications were outlined
by Petrie and Kennie (1987) and others are described briefly in papers
in Blakemore (1986).
Some of these applications will benefit directly from advances in
visualisation technology (McCormick et al, 1987; Herr and Zaritsky,
1988); examples include visual impact analysis of proposed
developments, such as installation of electricity transmission lines
on the environment (Turnbull and Gourlay, 1987) or the construction of
a dam (see colour plates byy International Imaging Systems in Burrough
(1986a)). A computer-generated video movie of a simulated flight over
a model of Los Angeles, constructed using a DTM and remote-sensed
data, indicated the level of realism which could be achieved using
LANDSAT satellite images with 30m pixel resolution (Herr and Zaritsky,
1988). A pixel resolution of 10m is available with SPOT imagery.
It is important to stress that the current thrust towards detailed
realism in computer graphics does not further all cartographic
applications. The realistic visualization of terrain models is
consistent with the scientific paradigm which guides the production of
reference maps. The DTM is a reference map from which measurements may
be made and reference maps aim to produce a faithful representation of
reality subject to prevailing constraints (see Figure 1). But, not all
three dimensional representations aspire towards realism; indeed,
cartography is often used to communicate entities which exist by
virtue of dictum, such as boundaries, and to explore and understand
concepts and phenomena which are not directly observable; i.e. to see
the unseeable. Concept refinement often relies on symbolic
representation of mutivariate proxy data (see section on interactive
visual computing).
Depth cues are already used in manual cartography. They include aspect
relief mapping, systematic hachuring, relief shading, panaromic views
and block diagrams (illustrated in Kraak, 1988) amongst other oblique
views of land and townscapes. These hand-crafted oblique views were
genuine maps since the representations were simplified and symbolic,
rather than realistic. Indeed, the process of adding calculated
realism to abstract descriptions of objects and scenes is contrary to
the main aims of cartography. The latter, as noted earlier, employs a
variety of
transformational processes to derive abstractions about reality and
communicate these through graphic symbolism. Cartography is more
concerned with discovering and communicating the quintessence of
reality rather than
with achieving a photographic record of the real thing. A cartographic
map is thus a graphic precis of reality. Sasada's (1987) work on the
"realism of drawing" therefore is highly relevant.
Geographers have been using DTMs for generation of orthogonal shaded
relief maps for some time now (Burrough, 1986). There could well be a
revival in use of symbolic oblique views of terrain within interactive
applications, where the speed of human information processing in real
time is as important as computer performance. As noted earlier,
cartographic processes evolved to facilitate rapid and accurate
information processing by the user. The
research on advanced cockpit design within British Aerospace is
concerned with the generation and use of symbolic 3D maps based on
DTMs, satellite remote-sensed and other data (Wills, personal
communication). The DTM has
become the base 'map' in 3D thematic mapping. Like the base map, the
DTM has to be generalised in order to communicate the message
effectively.
Automatic generalisation is one of
the outstanding challenges in cartography and the appropriate
generalisation of oblique views of DTMs for specific purposes opens up
new directions of research. There are other developments within
visualization that are relevant to cartography. The SIGGRAPH video
(Herr and Zaritsky, 1988, p1.5) includes an
animation sequence produced by the Human Factors Research Division at
NASA's Ames Research Centre. The researchers have combined the Data
Glove with a speech recognition device and head position sensors to
control a head-mounted display system with two liquid crystal display
units presented to each eye. The virtual (mental) image created by the
stereo pair appears to surround the user in 3D space. The operator is
able to view this image from
multiple viewpoints and explore and interact with this environment as
if it was real. The video only shows use of wireframe displays.
The use of virtual maps is not new in cartography and photogrammetry.
Kraak (1988) listed the visual expedients which have been used in
cartography for enhancing depth perception when using two images.
These include the optical stereoscope, anaglyphs and polarization.
Kraak used computer displays of pairs of oblique views to establish
whether viewing the map images in stereo led to improved performance.
In his experiments Kraak not only used DTMs
and urban scenes but also thematic 3D point symbol and prism maps,
which are already used as mono images in cartography. He concluded
that, provided that they were relatively simple in design, use of
stereo pairs resulted in
faster, but not necessarily better, processing of the map. Gugan and
Dowman (1986) described the design requirements and problems involved
in developing a digital stereo plotter for extracting topographic map
data from SPOT satellite imagery. Their paper included brief comments
on various methods for digital stereo viewing and on some tentative
solutions to the problem of Z-control when using the tracker ball as
an input device to the host image
processing system. Donoho et al (1981) used kinematic displays to
enhance the perception of point-clouds rotating in three-dimensional
space. Statisticians are also using cartographic techniques
within scatter plots (point distributions) of high dimensional data
(Chambers et al, 1983). Thus, advances in scientific
visualisation will benefit a number of cartographic applications.
4.2 Interactive Visual
Computing and Cartographic Exploration
The Report of the Panel on "Graphics, Image Processing and
Workstations", sponsored by the Division of Advanced Scientific
Computing (DASF) of the U S National Science Foundation (NSF),
advocated the view that "the purpose of
[scientific] computing is insight, not numbers" (McCormick et al,
1987, p3). It believed that "The most exciting potential of
wide-spread availability of visualization tools is not the entrancing
movies produced, but the insight
gained and the mistakes understood by spotting visual anomalies while
computing" (McCormick et al, p6). Unfortunately, the accompanying
SIGGRAPH videotape (Herr and Zaritsky, 1988), produced to illustrate
the pioneering
efforts in visualization, is mainly concerned with workstation trends,
expansion boards, input/output peripherals, lighting, animation and
parallel processing. The illustrations of scientific visualization,
medical imaging
and volumetrics and discussion of future trends tend to emphasize
illuminations of synthetic reality. A student could be forgiven for
concluding that interactive visual computing is mainly about changing
one's viewpoint within an illuminated geometric model of reality.
As described earlier, advances in 3D visualization do benefit some
applications of cartography. However, much scientific and sociological
data are multidimensional and the step up from 2D to 3D is not
sufficient to overcome the visualization problems. Under scientific
and engineering research opportunities, McCormick et al (1987, p A-8)
stated that "With visualization techniques an analyst can work with
many different, related data sets on a display screen; for example,
one can use computer generated markers to show exactly how different
images correspond".
Visvalingam and Kirby (1984) and Visvalingam (1985) had previously
expressed the view that that developments in workstation technology,
concurrency and windowing systems offered some prospects for
overcoming the problems of
hyper-dimensionality and provided a detailed account of how
cross-referencing elements in a set of displays led to insights about
socio-spatial data. In geographic research, understanding has been
furthered by comparing distributions of both individual and aggregate
elements in a set of maps in conjunction with use of tables of raw and
processed statistics.
Semi-manual cross-referencing and exploration of the information
content of multivariate data was laborious even when the spatial
statistics related to a relatively small set of data collection units.
Computer technology has
facilitated the collection, statistical analysis and display of data
for a very large number of units of much higher resolution; but
systems for easy visual cross-referencing and exploration are not yet
available, partly
because of the limitations of past technology.
State-of-the-art GIS, such as ARC/INFO, do not provide the necessary
facility and the SIGGRAPH video, representing the state-of-the-art in
scientific visualization, does not refer even to this requirement.
Visvalingam (1987) suggested that there was a need for a re-think on
systems design for interactive graphical information systems since
many advanced GIS were not developed for multi-process operating
systems. An early prototype, on a 1 Mbyte ICL Perq 1 computer running
an extended version of UNIX, for illustrating the use of maps as
two-way human-computer communication devices was limited by the
available technology and the limitations of the Graphical Kernel
System, GKS (Visvalingam, 1987). There is a lack of adequate support
for such investigative work in Britain, where the funding policy is
more supportive of 'near market' research and development. In
contrast, the National Centre for Geographic Information and Analysis
(NCGIA), funded by NSF, has identified "visualization research
pertaining to the display and
use of spatial data" as one of five priority areas for research (Fortheringham,
1989).
The emphasis to-date in visual mapping (as in computer graphics in
general) has been on picture generation and, to a lesser extent, on
picture interpretation. Even developments in graphic human-computer
interaction
have been oriented largely towards picture manipulation in the context
of picture generation with intelligent sketchpads. After an extended
initial preoccupation with enabling WIMP environments for graphic
interaction, there
is now a keenness to go beyond iconic interfaces and use the language
of graphics more fully within the user interface. Icons act as
pointers or represent information of a nominal kind. Researchers in
human-computer interaction (Mackinlay, 1986; Baecker and Buxton, 1987;
Shu, 1988) have more recently appreciated the need to take advantage
of the wealth of knowledge within cartography and graphic design.
Visual mapping has focused on the communicative role of graphics. But,
the language of graphics is not just a means for communication. Like
mathematics and the natural languages, the visual language provides a
means for exploring and crystallising thought. Bertin (1983) compared
and contrasted graphics with other languages.
4.3 Geographical
Information Systems and Digital Cartography
There is considerable confusion, at present, over the meaning of
the term GIS (Walford et al, 1989). Even prominent researchers in this
field have expressed different opinions. Mark (1986, p 157) defined a
GIS as "a computerized, spatially-referenced data base organized in
such a way that spatial data input, analysis, and output may be
accomplished"; i.e. he equated it with the digital map. Others, such
as Smith et al (1987) regard this as the simpler form of a GIS. Their
description of the field of GIS tended to emphasize the development of
deterministic, probabilistic and other heuristic procedures for
increasing the efficiency of storage, retrieval, analysis and display
of multi-source spatial data; i.e. their definition of GIS is very
similar to Guptill and Starr's (1984) definition of the new
cartography. They too saw the spatial database as the focus of
attention. Coppock and Anderson (1987, p 3) described GIS as "a
rapidly developing field lying at the intersection of many disciplines
- among them cartography, computing, geography, photogrammetry, remote
sensing, statistics, surveying and other disciplines concerned with
handling and
analysing spatially referenced data". Just as the super-discipline of
scientific visualization is seen as an integration of computer
graphics, image processing, computer vision, computer-aided design,
signal processing and user interface studies, GIS is regarded by some
as the super-science concerned with geographically-referenced data;
the name of the Economic and Social Research Council/Natural
Environment Research Council (ESRC/NERC) GIS Science Steering
Committee contributes to this image making. Consequently, others have
begun to relabel their activities as GIS, adding to the confusion. The
following description of the relationships between GIS and digital
cartography must therefore be read as a tentative personal
interpretation.
Even if we ignore the confusion resulting from over-zealous
promotional activities, the term GIS will mean different things to
different people. This is not surprising given the variety of
geographic data and the diversity of users and uses as catalogued in
the Chorley Report (DoE, 1987). At the British Computer Society
GIS Specialist Group's meeting in May 1989 on "What do I expect from
my GIS?", the invited speaker from English Heritage expressed in his
introduction that he was unsure as to whether their map-based
information system was a GIS. Many of the requirements outlined by
Dale at the same meeting also related to the acquisition, management,
manipulation and display of map and map-related data. Indeed, to
many map users, a GIS is a digital and/or visual map based
information system. Consequently, digital cartography provides
the infra-structure for many GIS applications. Maps are central to
many land and property based information systems, which are concerned
with factual information about locations and places; the basic spatial
unit (BSU) in these applications tends to be an appropriately defined
link or a land parcel (DoE, 1987; Dale, 1988; Visvalingam 1988 a & b).
The set of BSUs on a map may be regarded as forming one spatial
coverage. The map and associated gazeteer are used to cross reference
and retrieve data for
administrative and operational purposes within management information
systems (ICL, 1989). As previously described, some applications do not
see visual maps as essential but others, such as the utilities
(Mahoney, 1986) and the Land Registry (Smith, 1988), regard them as
vital to their functions. The recording and routine selective
retrieval and display of data on a single coverage is the least
onerous application of digital cartography. Even here, proprietary
DBMS and CAD software are insufficient although inventory systems are
being launched with raster or video-scanned
data. There is some disagreement (see later) as to whether such
inventory and recording systems fall within the remit of GIS.
Although the utilities and local authorities use common base maps,
they tend to operate with different sets of BSUs on separate coverages.
In Britain, the 1950 Public Utilities Streets Works Act requires that
public utilities exchange information concerning their mains and plant
records. This Act has forced a collaborative approach towards solving
the problem of inter-coverage comparisons within computerised systems
(Ives and Lovett, 1986). The GIS for Northern Ireland was
designed to facilitate data sharing (Brand, 1988). A number of other
GIS applications, including forestry, agriculture and conservation
also need to compare data on different map coverages on demand. This
was traditionally done using a light-table. Often, the aim is to
identify target/critical areas with the required combination of
characteristics. Within computerised systems, so-called overlay
analysis may be undertaken in raster and/or vector formats using
ad-hoc or general purpose systems (White, 1978; Walker and Moore,
1988; Sivertun et al, 1988). Both approaches pose their own
problems but the vector approach is conceptually and computationally
more demanding. GIS tools include facilities for integration of vector
and raster data on different coverages.
Even when polygons describe geographic phenomena with clearly
recognisable boundaries, as in choropleth maps, inherent
generalisation errors in source documents and inevitable errors
introduced during data capture and/or
processing produce spurious polygons on overlay. The automatic
'cleaning up' of spurious polygons is not a trivial or reliable
process. When the data are captured from analog sources compiled and
generalised at different scales, existing facilities for scaling and
intersecting polygons are inadequate. Map compilation requires the
matching of generalised and displaced features and adjustment of the
geometry prior to compilation. Saalfeld (1988) provided a
detailed and informative account of advances in conflation, or
automatic map compilation.
The use of map data involve other problems. Although interpolated
boundaries on isopleth and proximal maps are objectively derived, they
are often hypothetical (because the geographic phenomena may have a
continuous
rather than discrete distribution in space or because the
discontinuity, such as a buried geological fault, may not be directly
observable) and hence arbitrary and/or intrinsically fuzzy (Loudon,
1986; Burrough, 1986b). Manual overlay analysis involves
judgement and takes into account a number of factors, including the
accuracy (spurious or lack of) and reliability of different data, a
knowledge of the nature of underlying phenomena and the purpose and
accuracy requirements of the analysis. Thus, experts are able to
extract appropriate and usable information from available, but often
imperfect, data. Automatic overlay analysis does not, at present,
provide scope for the inclusion of such conceptual and semantic
knowledge and judgement.
Overlay analysis is also used for assumption-based disaggregation of
spatial statistics for source BSUs and their re-aggregation for target
BSUs. The properties of spatial statistics are variable and
considerable caution must
be exercised in cross-coverage comparison of spatial statistics (Flowerdew,
1990). Thus, some GIS applications of digital cartography are not just
concerned with the attributes of places, but they are also concerned
with distributions in space for the manipulation of attribute data. In
general, overlay analysis of input or calculated polygons (for
example, by defining buffer zones around point, line and area objects)
requires positional data
of high quality and cost.
A number of GIS applications, in government and in commerce, rely on
the processing of proxy statistics. The emphasis here is on thematic
mapping and on the accuracy of attribute, rather than boundary, data
for studying
spatial patterns, relationships and anomalies. The data collecting
framework consists of basic population units (BPUs), rather than BSUs,
and the boundaries representing the BPUs may be highly generalised and
even undefined, as in the case of unit postcodes. Indeed, many such
applications made little use of geographic maps in the past. This was
mainly due to the financial costs and time delays involved in
acquiring appropriate spatial descriptions. Area-based targetting of
advertising mail and research on socio-spatial structures have relied
more on sophisticated, but blind, statistical analysis of population
and other data for smaller and smaller areas. When spatial references
were unavailable, data from different sources were related using
non-spatial references, such as hierarchic codes
for administrative areas and unit postcodes. Packages for statistical
analysis such as SPSS [Statistical Package for the Social Sciences],
SAS [Statistical Analysis System] and other ad-hoc software have
formed the primary tools within this class of GIS, although
cartographic diagrams have been used to study distributions in
measurement space. Thus, applications which are concerned primarily
with the analysis of enumerated data, such as population,
agricultural, retail and other censuses and surveys, focus on
statistical populations rather than the containing spaces.
It is now widely recognised that a variety of factors at different
stages in information collection, processing and analysis can
introduce errors and unintended biases, leading to use of
inappropriate information derived from
these proxy statistics. Visualization is seen as a means of validation
and as an instrument for generating insight. Owing to fashion, client
pressure and the increasing availability of spatial descriptions,
cartographic capabilities (for example, SPSS Graphics and SAS/Graph)
are being appended to existing statistical packages; the mapping
capabilities of SAS/Graph are illustrated in Carter (1984).
Smith et al (1987) provided an indication of other types of GIS
functions, which include network analysis. There is no doubt that many
GIS include non-cartographic components. Tomlinson Associates (DoE,
1987, p 154) stressed this in their definition of GIS as a "digital
system for the analysis and manipulation of a full range of
geographical data, with
associated subsystems for digitizing and other forms of input and for
cartography and other forms of display used in the context of decision
making. The emphasis is clearly on the analysis and manipulation
functions, and in the GIS field they provide the primary motivation
for using digital methods; if the intention were not to analyze or
manipulate, there would be no point in converting the geographical
data to digital form in the first place". Their review of the North
American experience forms a valuable reference.
Unfortunately, this expert view alienates many map-based applications
which derive significant benefits from the computerisation of
map-based storage, retrieval and display of information. It also takes
a blinkered view of cartography as a form of display. Although
manipulation and analysis functions tend to be central to some
applications, these tend to be idiosyncratic; given the diversity of
users and uses, they cannot form the core of general purpose GIS
architectures. The definition of GIS, provided by Tomlinson
Associates, is not fundamentally different from the modern definition
of cartography provided by Guptill and Starr (1984) but it places the
emphasis on analytic functions rather than on the spatial database.
There is still a tendency towards promotion of general purpose
state-of-the-art GIS. In the author's opinion, all-singing,
all-dancing universal GIS do not provide the most effective
environment for all users. GIS systems could be made more accessible
and effective by architectures which facilitate product factoring and
the development of configurations which meet the requirements of
specific sets of users. From a systems developer's perspective, GIS
"add application specific modelling and manipulation capabilities to
application oriented configurations of components in digital
cartography" (Visvalingam, 1989, p 29). Loudon (1986) provided a
flavour of the specific requirements of geologists.
Coppock and Anderson (1987, p 4) stated that "Digital cartography
interacts with GIS in three ways: by providing a framework, through
national map coverages, for relating other categories of data; as a
source of spatially referenced data in its own right; and as one of
the methods of presenting the results of analyses of such data". This
paper has emphasized that maps are not just sources of data and forms
of display. They are also devices for exploring and understanding data
and for generating insight. Figure 2 accommodates not only the
established applications of digital cartography but also the emerging
interests in interactive cartography and in visual computing. GIS
regard visual maps as useful but dispensable in many cases; they focus
on the digital map. Although the spatial database is attracting
a great deal of research effort at the present stage of evolution, in
the long-term and from a user-centred perspective, visual and digital
maps will be seen as complimentary and equally important
representations of spatial
reality. CAD systems facilitate 'shallow' interaction with data
models. Graphical information systems require 'deep' interaction with
spatial databases through an interface, which could conceivably
include multiple views controlled by a set of concurrent or parallel
communicating processes.
5. CONCLUSION
Coppock
and Anderson (1987, p 3) also stated that "The collection of
spatially-referenced data is, of course, as old as cartography, and
every atlas is a form of GIS in that it brings together a wide variety
of such data from different sources". Both digital cartography and
GIS are rapidly developing fields with outwardly expanding, fuzzy and
overlapping concerns. Despite the wide and disparate concerns of
the discipline, cartography has a well-defined and unique focus,
namely the exploration, interpretation and communication of spatial
forms, distributions and their relationships through maps. Digital
cartography is the technology concerned with the construction and use
of computer-based systems for the practice of cartography and its
applications.
GIS is an application of cartography. The confusion surrounding GIS
is mainly due to its lack of unique subject matter. It is
insufficient to define GIS in terms of the functionality provided by
state-of-the-art systems or as lying at the intersection of several
disciplines. No doubt, there are
precedents for establishing a new science by carving out and combining
components of existing disciplines - biochemistry provides an
example. But, such a re-grouping of concerns is usually in response
to the emergence of a unique substantive focus.
The promotion of the science of GIS does not appear to be based on
such substantive criteria or academic tradition; instead, the rhetoric
relies on shifting attention away from such issues onto the
capabilities introduced by the new, now increasingly accessible,
Information Technology. The Chorley Committee (DoE, 1987, p 8)
described GIS as "the biggest step forward in the handling of
geographic information since the invention of the map" and that "such
a system is as significant to spatial analysis as the inventions of
the microscope and telescope were to science". If GIS correspond to
the telescope, then digital cartography and maps provide the lens in
the scopes of all (Geo)graphical
information systems.
But, the invention of the telescope did not spawn a new science - it
launched a worthy technology. (Telescopy
has been defined as the art or practice of constructing or of using
the telescope). In philosophical terms, the elevation of GIS to a
science without first identifying its unique and central concerns,
will merely promote scientism. Let us not
forget the debates over the word science which preceded the renaming
of the Social Science Research Council to the Economic and Social
Research Council.
Some leading researchers and experts
on GIS have proposed at least four candidate concerns as central,
namely the interactive map interface, digital mapping, spatial
analysis and applications. If we play down the impact of the changing
tools of trade and rely on more substantive criteria, then interactive
maps (both visual and digital) fall within the remit of cartography
and computer science and the bulk of the analytical methods are
essentially mathematical. Applications in turn belong to their
respective disciplines - geography, geology, surveying and so on.
This does not negate the immense value and commercial importance of
GIS. It is a unique technology,
drawing on several disciplines for systems development.
Cartography, for its part, not only provides a unifying framework and
data,
it also provides objectives, knowledge, principles and techniques.
Smith et al (1987) emphasized the need for systematic application of
theories and techniques from several sub-fields of computer science
and the integration of techniques developed in computer vision and
image processing in the design and implementation of GIS. This paper
has attempted to indicate to CAD readers that the extraction of useful
information from available spatial data is an art as much as it is a
science. If the data and methods of cartography are divorced from and
manipulated without regard to underlying meanings, principles,
constraints and patterns of usage, then the resulting systems will be
of limited value. Worse still, uncritical and de-skilled use of such
systems could result in disaster. The inflexible design of the
emergency system for the fire brigade, in the case of the Hillsborough
disaster (Computer Weekly, 1989), provides a timely warning.
ACKNOWLEDGEMENTS
Whilst accepting full responsibility for the views expressed in this
paper, I would like to thank the following for their helpful comments
- Drs. Graham Kirby and Mike Turner of the Cartographic Information
Systems Research Group, David Fairbairn (Editor of the Cartographic
Journal), Mike Brand of Ordnance Survey Northern Ireland and the
referees of this paper.
REFERENCES CITED
Abel, D. J. and Smith, J. L. (1983) "A data structure and algorithm
based on a linear key for rectangle retrieval problems", Computer
Vision Graphics and Image Processing 24 (1), 1 - 13.
Abraham, I. M. (1989) "Automated cartographic line generalisation and
scale-independent databases", Unpublished PhD Thesis, C.N.A.A.
Allder, W. R. and Elassal, A.A. (1984) USGS digital cartographic data
standards. digital line graphs from 1 : 24 000 scale maps, U.S.
Geological Survey Circular 895 - C.
Baecker, R. M. and Buxton, W.A.S. (eds., 1987) Readings in
Human-Computer Interaction - a multidisciplinary approach (Morgan
Kaufmann, California), Chapter 7.
Ballard, D. H. (1981) "Strip trees : a hierarchical representation for
curves" CACM 24 (5), 310 - 321.
Bertin, J. (1983) Semiology of Graphics - Diagrams Networks Maps
(University of Wisconsin Press, Wisconsin); original in French
published in 1967.
Bickmore, D. P. (1964) "An automatic system of cartography", Paper
presented to the International Cartographic Association meetings, July
31.
Blakemore, M (1986, ed.) Proceedings Auto Carto London (Auto
Carto London Ltd., London).
Brand, M. J. D. (1988) "The Geographical Information System for
Northern Ireland", Mapping Awareness 2 (5), 18 - 21.
Brassel, K. E. and Weibel, R. (1988) "A review and conceptual
framework for automated map generalisation", Int. J. G. I. S.
2 (3), 229 - 243.
Burrough, P. A. (1986a) Principles of Geographical Information
Systems for Land Resources Assessment (Oxford University Press,
Oxford).
Burrough, P. A. (1986b) "Five reasons why Geographical Information
Systems are not being used efficiently for land resources assessment",
in Blakemore, M. (ed.), op. cit., 139 - 147.
Carter, J. R. (1984) Computer Mapping : Progress in the '80s
(Association of American Geographers, Washington).
Chambers, J.M., Cleveland, W.S., Kleiner, B. and Tukey, P.A. (1983)
Graphical Methods for Data Analysis (Duxbery Press, Boston).
Clayton, P. (1989) "Carto 1 : Automated Map Production (MAP)", OS
News 96, 13 - 15.
Clubb, N. and Hilder, D. (1989) "Digital mapping at English Heritage",
Mapping Awareness 3 (2), 18 - 21.
Collinson, A. (1989) The British Cartographic Society, Newsletter, No
1, p 1.
Computer Weekly (1989) "The need for flexibility", June 22, p 17.
Cook, A. C. (1986) "A method for automatically annotating maps", in M.
Visvalingam (ed.) Research based on Ordnance Survey Small-Scales
Digital Data, Cartographic Information Systems Research Group,
University of Hull, 51 - 60.
Coppock, J. T. and Anderson, E. K. (1987) "Editorial review", Int.
J.G. I. S. 1 (1), 3 - 11.
Corbett, J. P. (1979) "Topological Principles in Cartography", Tech.
Paper 48 (U.S. Bureau of the Census, Washington).
Cuff, D. J. and Mattson, M. T. (1982) Thematic Maps : their design
and production" (Methuen, New York).
Dale, P. F. (1988) "Land information systems and the land parcel",
Mapping Awareness 2 (2), 27 - 29.
Davis, J. C. and McCullagh, M. J. (eds., 1975) Display and Analysis
of Spatial Data (Wiley, New York).
DCDSTF - The Digital Cartographic Data Standards Task Force (1988)
"The proposed standard for digital cartographic data", American
Cartographer 15 (1), special issue.
De Simone, M. D. (1986) "Data structures and feature recognition :
from the graphic map to a digital data base", Unpublished Ph.D.
Thesis,
C.N.A.A.
Diaz, B. M. and Bell, S. B. M. (1986, eds.) Spatial Data Processing
using Tesseral Methods (Natural Environment Research Council,
Swindon).
Dickinson, G. C. (1973) Statistical Mapping and the Presentation of
Statistics(Edward Arnold, London), 2nd edition.
DoE - Department of Environment (1987) Handling Geographic
Information (H.M.S.O., London).
Donoho, D., Huber, P.J. and Thoma, H.M. (1981) "The use of kinematic
displays to represent high dimensional data" in Eddy, W.F. (ed.)
Computer Science and Statistics, Proc. of the 13th Symposium on
the Interface
(Springer-Verlag, New York).
Drinkwater, C.R. and Fielding, A.P. (1986) "The Hydrographic
Department and the requirement for digital hydrographic data", in
Blakemore, M. (ed.) op. cit., 361 - 368.
Durham, T. (1989) "Design programs that draw their own conclusions",
Computing 6 July, 26 - 27.
Ericksen, R. H. (1984) "A polygon chaining algorithm to simplify area
encoding", Comput. Environ. Urban Systems 9 (4), 237 - 245.
ESRI (1985) ARC/INFO Users Manual - Version 3, Environmental
Systems Research Institute, Redlands, California.
Essinger, R. (1986) "The philosophy and requirements of computer-aided
graphic design in cartography", in Blakemore, M. (ed.) op. cit., 189 -
196.
Flowerdew, R. (1990) "Statistical methods for areal interpolation :
predicting count data from a binary variable", Int. J. G. I. S.,
in press.
Fortheringham, S. (1989) "The US National Centre for Geographic
Information and Analysis", Mapping Awareness 3 (2), 50.
Frank, A. (1982) "MapQuery : database query language for retrieval of
geometric data and their graphical representation", Computer
Graphics 16, 199 - 207.
Freeman, H. and Pieroni, G. G. (eds., 1980) Map Data Processing
(Academic Press, New York).
Gargantini, I. (1982) "An effective way to represent quadtrees",
CACM 25 (12), 905 - 910.
Green, N.P.A., Finch, S. and Wiggins, J. (1985) "The
'state-of-the-art' Geographic Information Systems", Area 17 (4),
295 - 301.
Gugan, D.J. and Dowman, I. J. (1986) "Design and implementation of a
digital photogrammetric system", Int. Archives of Photogrammetry
and Remote Sensing 26 (2), 100 - 109.
Guptill, S.C. (1986) "A new design for the U S Geological Surveys
national digital cartographic database", in Blakemore, M. (ed.) op.
cit., 10 - 18.
Guptill, S.C. and Starr, L.E. (1984) "The future of cartography in the
information age" in Morrison, J. L. (ed.) Computer-Assisted
Cartography Research and Development Report (ICA Commission C), 1
- 15.
Hadley, C. (1987) "The Ordnance Survey 1:625 000 database map
production system", in Robinson, G. R. (ed.) Going Digital : Pros
and Cons, British Cartographic Society Special Publication Number
4, 30 - 49.
Harvard Laboratory for Computer Graphics and Spatial Analysis (dates
vary) Harvard Library of Computer Graphics, Mapping Collection 19
volumes, (Harvard Graduate School of Design, Cambridge).
Haywood, P. (1987) "The OS Topographic Database Study" in Haywood, P.
(ed.) SORSA '87 DURHAM (Ordnance Survey, Southampton).
Herr, L. and Zaritsky, R. (1988) Visualization : State of the Art,
ACM/SIGGRAPH video review volume 30.
Howe, M. G. (1970) National Atlas of Disease and Mortality in the
United Kingdom (Nelson, London).
ICL (1989) "ICL", Mapping Awareness" 3(2), 3 - 6.
Ibbs, T.J. and Stevens, A. (1988) "Quadtree storage of vector data",
Int. J. GIS 2 (1), 43 - 56.
International Cartographic Association (1973) Multilingual
Dictionary of Technical Terms in Cartography (Steiner, Wiesbaden).
Ives, M. J. and Lovett, R. (1986) "Exchange of digital records between
public utility digital mapping systems", in Blakemore, M. (ed.), op.
cit., 181 -189.
Keates, J. S. (1982) Understanding Maps (Longman, London).
Kirby, G.H., Visvalingam, M. and Wade, P. (1987) "Storage and
retrieval of topographic data using a relational database management
system", in Haywood, P. (ed.) SORSA '87 DURHAM (Ordnance Survey,
Southampton).
Kirby, G.H., Visvalingam, M. and Wade, P. (1989) "The recognition and
representation of polygons with holes", Computer Journal, in
press.
Kleiner, A. and
Brassel, K. E. (1986) "Hierarchical grid structures for static
geographic data bases", in Blakemore, M. (ed.) op. cit., 485 - 496.
Kraak, M. J. (1988) Computer-assisted Cartographical
Three-dimensional Imaging techniques (Delft University Press,
Delft).
Lane, M. (1988) "The rural areas database", Mapping Awareness 2 (2),
17 - 18.
Lauzon, J. P., Mark, D. M., Kikuchi, L. and Guevara, J. A. (1985)
"Two-dimensional run-encoding for quadtree representation",
Computer Vision, Graphics and Image Processing 30, 56 - 69.
Loudon, T. V. (1986) "Digital spatial models and geological maps", in
Blakemore, M. (ed.) op. cit., 60 - 65.
Mackinlay, J. (1986) "Automating the design of graphical presentations
of relational information", ACM Trans. on Graphics 5 (2), 110 -
141.
Mahoney, R. P. (1986) "An information centre", in Blakemore, M. (ed.),
op. cit. 190 - 199.
Maling, D. H. (1973) Coordinate Systems and Map Projections
(George Philip, London).
Maling, D. H. (1989) Measurements from Maps : principles and
methods of cartometry (Pergamon Press, Oxford).
Mapping Awareness (1989) "Trade Directory - Data Capture and
Conversion Services", Mapping Awareness 3 (5), 19 - 25.
Mark, D. M. (1984) "Automated detection of drainage networks from
digital elevation models", Cartographica 21, 168-178.
Mark, D. M. (1986) "The use of quadtrees in Geographic Information
Systems and spatial data handling", in Blakemore, M. (ed.), op. cit.
517 - 526.
Marks, D., Dozier, J. and Frew, J. (1984) "Automated basin delineation
from digital elevation data", GeoProcessing 2, 299 - 311.
Marx, R. W. (1986) "The TIGER System : automating the geographic
structure of the United States census", Government Publications
Review 13, 181 - 201.
Mason, D.C., Corr, D.G., Cross, A., Hogg, D.C., Lawrence, D. H.,
Petrou, M. and Tailor, A.M. (1988) "The use of digital map data in the
segmentation and classification of remotely-sensed images", Int. J.
G. I. S. 2 (3), 195 - 216.
Masters (1989) "The shepherds warning", Computer Images June 1989,
17 - 21.
Maver, T. W. (1987) "Modelling the cityscape with geometry engines",
CAD, 19 (4) 193 - 196.
McCormick, B. H., DeFanti, T.A. and Brown, M.D. (eds., 1987)
"Visualisation in scientific computing", Computer Graphics 21 (6).
McMaster, R. B. (1987) "Automated line generalization",
Cartographic 24 (2), 74 - 111.
McMaster, R. B. (1987) "The geometric properties of numerical
generalization", Geographical Analysis 19 (4), 330 - 346.
Mitchell, W. B., Guptill, S. C., Anderson, K. E., Fegeas, R. G. and
Hallam, C. A. (1977) "GIRAS : A Geographic Information Retrieval and
Analysis System for handling land use and land cover data", U.S.
Geological
Survey Professional Paper 1059.
Monmonier, M. S. (1981) "Trends in atlas development",
Cartographica 18 (2), 187 - 213.
Monmonier, M. S. (1982) Computer-Assisted Cartography : principles
and prospects (Englewood Cliffs, New Jersey).
Morrison, J. L. (1986) "Cartography : A milestone and its future", in
Blakemore, M. (ed.) op. cit., 1 - 12.
Muller, J-C (1987) "Fractal and automated line generalization",
Cartographic J. 24 (1), 27 - 34.
Nagy, G. and Wagle, S. (1979) "Geographic data processing", ACM
Computing Surveys 11 (2), 139 - 181.
Openshaw, S. and Mounsey, H. (1986) "Geographic information systems
and the BBC's Domesday interactive videodisk", in Blakemore, M. (ed.),
op., cit. 539 - 546.
Ordnance Survey (1989) The National Transfer Format - Release 1.1,
January 1989.
Petrie, G. and Kennie, T. J. M. (1987) "Terrain modelling in surveying
and civil engineering", CAD 19 (4), 171 - 187.
Peucker, T. K. (1975) "A theory of the cartographic line", Proc. Auto
Carto II, Reston, Virginia, 508 - 518.
Peucker, T. K. and Chrisman, N. (1975) "Cartographic data Structures",
Am. Cartographer 2 (1), 55 - 69.
Peuker, T.K. and Douglas, D.H. (1975) "Detection of surface-specific
points by local parallel processing of discrete terrain elevation
data", Computer Graphics and Image Processing 4, 375 - 87.
Peuquet, D. J. (1984) "A conceptual framework and comparison of
spatial data models", Cartographica 21 (4), 66 - 113.
Rhind, D. W. (1977) "Computer-aided cartography", Trans. I. B. G.
NS 2 (1), 71 - 97.
Robinson, A.H., Sale, R.D., Morrison, J. L. and Meuhrcke, P. C. (1984)
Elements of Cartography (Wiley, New York), 5th edition.
Saalfeld, A. (1988) "Conflation. Automated map compilation", Int.
J. G. I. S. 2 (3), 217 - 228.
Samet, H. (1988) "Hierarchical representations of collections of small
rectangles", ACM Computing Surveys 20 (4), 271 - 309.
Sasada, T. T. (1987) "Drawing natural scenery by computer graphics",
Computer Aided Design 19 (4), 212 - 218.
Serpell, D. (1979) Report of the Ordnance Survey Review Committee
(H.M.S.O., London).
Shneiderman, B. (1987) Designing the User Interface
(Addison-Wesley, Reading, Mass.)
Shneier, M. (1981) "Two hierarchical linear feature representations :
edge pyramids and edge quadtrees", Computer Graphics and Image
Processing 17, 211 - 224.
Shu, N. C. (1988) Visual Programming (Van Nostrand Reinhold,
New York).
Sivertun, A., Reinelt, L. E. and Castensson, R. (1988) "A GIS method
to aid in non-point source critical area analysis", Int. J. G. I.
S. 2 (4), 365 - 378.
Slater, M. (1986) "Segments on bit-mapped graphics displays",
Software-Practice and Experience 16 (11), 965 - 980.
Slater, M., Davidson, A. and Smith, M. (1989) "Liberation from
rectangles: a tiling method for dynamic modification of objects on
raster displays" Computers & Graphics 13 (1), 83 - 89.
Smith, N. and Groom, P. (1987) "Testing of relational databases in
Ordnance Survey research and development", in Haywood, P (ed.) SORSA
'87 DURHAM (Ordnance Survey, Southampton).
Smith, N. and Turner, J. (1987) "The Ordnance Survey topographic
database: implications for the input and output of map data", in
Earnshaw, R. A. (ed.) State of the Art in Computer Cartography
(BCS Displays Group, London),
Smith, P. (1988) "Developments in the use of digital mapping in HM
Land Registry", Mapping Awareness 2 (4), 37 - 41.
Smith, T. R., Menon, S., Star, J. L. and Estes, J. E. (1987)
"Requirements and principles for the implementation and construction
of large-scale geographic information systems", Int. J. G. I. S. 1
(1), 13 - 31.
Starr, L. E. (1986) "Mark II : the next step in digital systems
development in the U. S. Geological Survey", in Blakemore, M. (ed.)
op. cit., 200 - 205.
Taylor, D. R. F. (1974) "The Canadian cartographer and the computer -
present trends and future challenges", Canadian Cart. 11 (1),
35 - 44.
Tomlin, C. D., Berwick, S. H. and Tomlin, S. M. (1983) "Cartographic
analysis of deer habitat utilization", in E. Teicholz and B.J.L. Berry
(eds.) Computer Graphics and Environmental Planning (Prentice
Hall, Englewood Cliffs, New Jersey), 141 - 150.
Tomlinson Associates (1987) "Review of North American experience of
current and potential uses of Geographical Information Systems", in
Department of Environment, op. cit. 153 - 161.
Turnbull, W.M. and Gourlay, I. (1987) "Visual impact analysis : a case
study of a computer-based system", CAD 19 (4), 197 - 202.
van Roessel, J. W. (1987) "Design of a spatial data structure using
the relational normal forms", Int. J. G. I. S. 1 (1), 33 - 50.
van Roessel, J. W. and Fosnight, E. A. (1985) "A relational approach
to vector data structure conversion", Auto Carto 7, 541 - 551.
Visvalingam, M. (1985) "Concept refinement through the graphic
representation of large data sets", in Shackel, B. (ed.) Interact
'84 - Proceedings of the First IFIP Conference on Human-Computer
Interaction (North
Holland, Amsterdam), 281 - 286.
Visvalingam, M. (1987) "Problems in the design and implementation of a
GKS-based user interface for a graphical information system", The
GKS Review, The Proc. of the Eurographics Workshop, September
1987,
ISO/TC97/SC24/N5.
Visvalingam, M. (1988a) "Issues relating to basic spatial units : Part
1", Mapping Awareness and Integrated Spatial Information Systems 2
(3), 40 - 42.
Visvalingam, M. (1988b) "Issues relating to basic spatial units : Part
2", Mapping Awareness and Integrated Spatial Information Systems 2
(4), 42 - 45.
Visvalingam, M. (1989) "Cartography, GIS and maps in perspective",
Cartographic J 26 (1), 26 - 32.
Visvalingam, M. and Sekouris (1989) "The management of digital map
data using a relational database model", Ordnance Survey End-of-grant
Report, 50 pp.
Visvalingam, M. and Kirby, G. H. (1984) "The impact of advances in IT
on the cartographic interface in social planning", Department of
Geography Miscellaneous Series No 27, University of Hull.
Visvalingam, M., Wade, P. and Kirby, G. H. (1986) "Extraction of area
topology from line geometry", in Blakemore, M. (ed.), op. cit., 156 -
165.
Visvalingam, M. and Whyatt, J. D. (1989) "The Douglas-Peucker
algorithm for line simplification : re-evaluation through
visualization", Cartographic Information Systems Research Group
Discussion Paper 6, University of Hull,
35 pp.
Wade, P. (1986) "Derivation of hierarchic area objects from O.S.
feature coded vectors", in Visvalingam, M. (ed.) Research based on
Ordnance Survey Small-Scales Digital Data, Cartographic
Information Systems
Research Group, University of Hull, 61 - 70.
Wade, P., Visvalingam, M. and Kirby, G.H. (1986) "From line geometry
to area topology", CISRG Discussion Paper 1, University of Hull, 48
pp.
Walford. N., Lane, M. and Shearman, J. (1989) "The rural areas
database : a geographical information and mapping system on rural
Britain", Trans. IBG NS 14 (2), 221 - 230.
Walker, P.A. and Moore, D. M. (1988) "SIMPLE : an inductive modelling
and mapping tool for spatially oriented data", Int. J. G. I. S. 2
(4), 347 - 363.
Waugh, T.C. and McCalden, J. (1983) GIMMS Reference Manual (GIMMS
Ltd., Edinburgh EH3 9E4), Edn. 4.5
Weights, B. (1989) "Kingston's GIS pilot set for promotion",
Mapping Awareness 3 (2), 8 - 11.
White, D. (1978) "A design for polygon overlay", in Dutton, G. (ed.)
First International Advanced Study Symposium on Topological Data
Structures for Geographic Information Systems, Harvard Papers on GIS,
vol. 6 (Spatial Algorithms : Efficiency in Theory and Practice), (Lab.
for Computer Graphics and Spatial Analysis, Graduate School of Design,
Harvard University).
