Beruflich Dokumente
Kultur Dokumente
of GIS
PART
1
CHAPTER
What is GIS?
Learning outcomes
By the end of this chapter you should be able to:
BOX1.1
have been located and entered into the GIS. The factors are regularly added to the GIS from reports
Happy Valley management team is trying to establish provided by the ski patrols and local weather service.
whether there is any common theme or spatial pattern The warden can use this information and the GIS to
to the accidents. Do accidents of a certain type occur help identify which runs should be opened or closed.
only on specific ski pistes, at certain points on a ski 5 What will the spatial implications be if an organi-
piste such as the lift stations, or at particular times of zation takes certain action? The access road to Happy
day? So far one accident black spot has been identified Valley is now too narrow for the number of skiers
where an advanced ski run cuts across a slope used by visiting the area. A plan is being prepared for widen-
beginners, just below a mountain restaurant. ing the road. However, any road-widening scheme
will have impacts on a local nature reserve as well as
surrounding farm land. The Happy Valley GIS is being
used to establish the amount of land that is likely to
be affected under different road-widening schemes.
If you have a geographical background you may data required and a lack of time and techniques avail-
be asking what is new about these generic questions. able to process these data. The following examples
Are these not the questions that geographers have of GIS applications are used to illustrate the capa-
been contemplating and answering for centuries? In bilities of GIS as a tool for geographical analysis. All
part they are, though in many cases geographers and involve the manipulation of data in ways that would
others using spatial data have been unable to find be difficult or impossible by hand, and each illustrates
answers to their questions because of the volume of different issues associated with the application of GIS.
Introduction 5
(a)
for the identification of suitable radioactive waste searched for sites using these techniques. This was a
disposal sites. Advice on what to do with the UK’s standard approach employed in the siting of a wide
radioactive waste is provided by the Committee on range of activities including shopping centres, roads
Radioactive Waste Management (CoRWM). The and offices (Figure 1.5). The method is time-con-
NDA has the task of interpreting current govern- suming and means that it is impossible to perform
ment radioactive waste policy and siting guidelines the analysis for more than a few different siting
and presenting possible sites at public inquiries. One criteria. The best sites are often missed. GIS tech-
of the problems for NDA is the lack of comprehensive niques offer an alternative approach, allowing quick
and coherent guidelines for the identification of suit- remodelling for slight changes in siting criteria, and
able sites. Another is that radioactive waste is a strong produce results as maps eminently suitable for pres-
political issue because nobody wants a disposal facility entation at public inquiries.
in their neighbourhood and protests against poten- Openshaw et al. (1989) first demonstrated the use
tial sites are common (Figure 1.4). However, NDA of GIS for nuclear waste repository siting, and their
is expected to show that it has followed a rational method is summarized in Figure 1.6. They estab-
procedure for site identification (Department of the lished a number of data layers, each containing data
Environment, 1985). Hydrology, population distribu- for a separate siting criterion (for example, geol-
tion and accessibility are examples of important siting ogy, transport networks, nature conservation areas
factors, but how such factors should be interpreted is and population statistics). These were converted
left up to NDA. Where, therefore, should NDA site a from paper to digital format by digitizing (this tech-
nuclear waste repository? nique is explained in Chapter 5), or acquired from
NIREX (the NDA’s predecessor) used a pen-and- existing digital sources such as the UK Census of
paper approach to sieve through large numbers of Population. These data layers were then processed
paper maps containing data about geology, land use, so that they represented specific siting criteria. The
land ownership, protected areas, population and geology layer was refined so that only those areas
other relevant factors. Areas of interest were traced with suitable geology remained; the transport
from these maps by hand, then the tracings were layer altered so that only those areas close to major
overlaid to identify areas where conditions over- routes were identified; and the nature conservation
lapped. NIREX is not the only organization that has layer processed to show protected areas where no
Introduction 7
development is permitted. The population layer was GIS software was then used to combine these new
analysed so that areas with high population densi- data layers with additional layers of information
ties were removed. The four data layers are shown representing other siting criteria. The final result
in Figure 1.7. was a map showing the locations where all the spec-
ified siting criteria were satisfied and, thus, a number
of locations suitable for the siting of a nuclear waste
repository. The advantage of using the GIS to per-
form this task was that the siting criteria could be
altered and the procedure repeated with relative
ease. Examples for several siting scenarios are shown
in Figure 1.8. These illustrate how changes in siting
criteria influence the geographical distribution of
potential sites.
This example shows how a GIS approach allows
comparative re-evaluation and testing of data and
conditions. In this way the decision maker can eval-
uate options in a detailed and scientific manner. The
(a) Step one
work of Openshaw et al. (1989) also illustrates three
other important issues associated with the use of
GIS: the problem of errors in spatial data sets; the
difficulty in establishing criteria for abstract spatial
concepts; and the potential value of using GIS to
communicate ideas (Box 1.2).
Identify relevant
siting factors
Collect appropriate
data and digitize
Yes
Overlay
Examine No
output
OK?
Yes
Final potential
areas
Figure1.7 Radioactive waste case study: geology, population, transport and conservation criteria maps;
Sources: (a) British Geological Survey. © NERC, IPR/71-39C reproduced by permission; (b) Office for National Statistics;
(c) Ordnance Survey; (d) Joint Nature Conservation Committee, www.jncc.gov.uk
Introduction 9
(a) Near-surface disposal in surface clay (b) Deep disposal in suitable deep geology
geology
Figure1.8 Radioactive waste case study (a and b): results from different siting scenarios
OO Errors in source data, such as those introduced distance (for example, 10, 20 or 30 km) ‘far from’
during the conversion of data to digital form, represents. In some cases rules may be applied
may have a significant effect on the GIS site- to guide this process, but in others the numerical
searching process. Mistakes in capturing areas representation of a criterion may depend upon the
of appropriate geology from paper maps may preferences of the person responsible for choosing
lead to inappropriate waste repository sites being and implementing the criterion.
identified, because areas on the ground will OO GIS output can be used to inform public participa-
have different geological properties from those tion in the decision-making process. A series of
recorded in the GIS. Errors in spatial data sets and maps could be used to illustrate why a particu-
the associated issues of data quality are discussed lar geographical location has been identified as a
in detail in Chapter 10. suitable site for the disposal of radioactive waste.
OO The GIS site-searching process relies on the trans- However, the issues raised above – data quality
lation of abstract (or ‘fuzzy’) concepts such as ‘near and the problems of creating spatial criteria for
to’ and ‘far from’ into precise conditions that can abstract concepts – suggest that output from GIS
be mapped. This can be a problem. How do you should be viewed with caution. Just because a map
create a map that shows all the geographical zones is computer-generated does not mean that the pic-
‘far from’ a centre of population? The only method ture it presents is correct. More on this issue can
is to make an arbitrary decision about what sort of be found in Chapter 8.
10 Chapter1 What is GIS?
Evaluating land use planning In this context GIS permits scientists and man-
agers in Zdarske Vrchy to interact with their data
Virtually every country in the world has areas of and ask questions such as:
natural beauty and conservation value that are
managed and protected in the public interest. Those OO What will be the long-term consequences
managing these areas face the problem of balan- of continuing recreational activity for the
cing human activities (such as farming, industry landscape? (Figure 1.10)
and tourism) with the natural elements of the land- OO Where will damage from acid rain occur if a
scape (such as climate, flora and fauna) in order to particular industrial plant continues to operate?
maintain the special landscape character without OO Where is the best location for the re-introduction
exploitation or stagnation. of certain bird species?
The protected area of Zdarske Vrchy, in the
OO Where should landscape conservation zones be
Bohemian–Moravian highlands of the Czech
established?
Republic (Figure 1.9), is an example of an area that
has suffered as a result of ill-considered state con- (after Downey et al., 1991, Downey et al., 1992 and
trol. Unregulated farming, tourism and industrial Petch et al., 1995).
activities have placed the landscape under severe One application of GIS in Zdarske Vrchy has been
pressure. Czech scientists and environmental to identify areas of the landscape for conservation.
managers have relied on traditional mapping and Traditionally, water storage in the region has relied
statistical techniques to monitor, evaluate and pre- on the use of natural water reservoirs such as peat
dict the consequences of this exploitation. However, wetlands and old river meanders (Petch et al., 1995).
with change in the political administration of the Current land-use practices, in particular forestry
country came scientists and managers who were and farming, are resulting in the gradual disappear-
looking not only into a policy of sustainable devel- ance of these features. The consequences of these
opment for Zdarske Vrchy, but also at GIS as a tool changes have been localized droughts and floods in
to help with policy formulation (Petch et al., 1995). areas further downstream. In turn, these changes
GIS is seen as a tool to bring together disparate have brought about a reduction in plant species and
data and information about the character and activi- wildlife habitats. Managers in the Zdarske Vrchy
ties that take place in the Zdarske Vrchy region. region wanted to identify conservation zones to
Data from maps, aerial photographs, satellite protect the remaining natural water reservoirs as
images, ecological field projects, pollution moni- well as to identify those areas where it may be pos-
toring programmes, socio-economic surveys and sible to restore the water retention character of
tourism studies have been mapped and overlaid to the landscape. To do this, they needed to establish
identify areas of compatibility and conflict. the characteristics of the landscape that determine
whether or not a particular location is likely to
retain water. Specialists in hydrology, geology and
ecology were consulted to identify a range of impor-
tant criteria describing:
OO the type of soil and its water retention ability;
OO the character of the topography (for example,
presence or absence of hollows or hills);
OO the type of land use, as certain agricultural
practices exploit the water retention capacity of
the landscape; and
OO the presence or absence of human-enhanced
Figure1.9 Zdarske Vrchy water drainage channels.
Introduction 11
(b) Relief (20 m contours) (e) Recreational load: this map has been produced by
combining the maps shown in panels a–d with several
other data layers for the region
Identify
relevant
data
Remote Paper
sensing Topographic ecological maps
data maps from field work
Digitize
Extract Digitize
land cover ecological
maps zones
Overlay to
identify sensitive
Extracted from Extracted from
zones
water management nature conservation
database database
Rivers Ecologically Nature
and sensitive conservation
ponds zones sites
Overlay to
identify sensitive zones
for consideration as
conservation zones
Decide No
which zones to
implement
Yes
New
conservation
zone established
GIS professionals were then asked to find appro- These data were acquired and entered into the
priate sources of spatial data that could be used to GIS. The hydrologists, ecologists and geologists
represent these criteria. A range of sources was iden- were then asked how each of the criteria (land use,
tified including: topography, soil type and drainage) might interact
to influence the retention capacity of the landscape
OO paper maps (for soil type and geology); at a particular location. First, the scientists used the
OO contour maps (for topography); GIS to look at the relationship of the criteria they
had identified in areas where natural water reser-
OO ecological field maps (for drainage conditions);
voirs were still in existence. This involved adding
and
more data to the GIS about the location of existing
OO remote sensing (for land use). water retention zones. These data came from the
Introduction 13
regional water authorities as paper maps. The rela- Europe and is being practised by environmen-
tionship between the geographical distribution of tal managers all over the world. In addition, the
the various landscape characteristics at these loca- Zdarske Vrchy project reveals a number of other
tions was then used to develop a model. This model important issues associated with the use of GIS.
allowed the managers of the Zdarske Vrchy region These include the problem of data sources being
to predict which other areas could be restored as in different map projections, the value of GIS as a
natural water reservoirs. The next stage was to modelling tool and the role for GIS as a participa-
check which of these areas were located in exist- tory problem-solving tool (Box 1.3).
ing conservation zones, as it was easier to change
existing conservation regulations than to set up
Finding a new home
new conservation areas. Figure 1.11 summarizes
the method used and shows how the GIS was used At some stage in our lives, most of us will need to
to integrate data from a range of different sources. look for a new home. Perhaps because of a new job,
Figure 1.11 also shows how these data were overlain or a change in family circumstances, our accommo-
with additional data about existing water retention dation requirements will change and we will have
zones and existing conservation areas to identify to look for a new place to live. This can be a time-
potential new conservation sites. consuming and frustrating task. The requirements
The Zdarske Vrchy project shows how GIS can of individual family members need to be considered.
be used to bring together data from a wide variety Do they need to be close to schools, major roads or
of sources to help address a range of environmen- a railway station? Perhaps they would prefer to be in
tal management problems. This use of GIS is not an area where insurance costs are lower. Maybe they
unique to environmental planning in Eastern want to be in an urban area to be close to shops and
OO The greatest problem associated with bringing data OO The Zdarske Vrchy case study also shows how GIS
together for the creation of the Zdarske Vrchy GIS can be used to create models of environmental
was deciding which map projection to adopt as the processes with maps used as the building blocks
common frame of reference. Several different pro- for the model. The topic of modelling and GIS is
jection systems were used by the source maps. A returned to in Chapters 6 and 7.
projection system is the method of transformation OO Bringing people together to search for a solution
of data about the surface of the earth on to a flat to a common problem is often difficult. Different
piece of paper. Because many methods exist which specialists will have different ideas about the prob-
can be used to perform this task, maps drawn for
lem. For example, an ecologist might recommend
different purposes (and maybe even for the same
one approach, an engineer a second and an econo-
purpose but at different points in time) may use dif-
mist a third. The Zdarske Vrchy project showed
ferent projection systems. This does not present
how, through the use of GIS, the common medium
any problems as long as the maps are used inde-
of the map could be used as a tool to help experts
pendently. However, when the user wishes to
overlay the data in a GIS, the result can be confus- from different backgrounds exchange ideas and
ing. Features that exist at the same location on the compare possible solutions. The idea that GIS can
ground may appear to lie at different geographical be used as a participatory problem-solving tool has
positions when viewed on the computer screen. also received considerable attention from the GIS
This problem became apparent in the Zdarske research community (Carver et al., 1997) and often
Vrchy project when the road network, present on involves the use of web-based mapping tools.
two of the maps, was compared. Map projections Chapter 7 considers this topic in more detail.
are explained in more detail in Chapter 2.
14 Chapter1 What is GIS?
their place of work. To find a new home acceptable There are examples of software developed to per-
to all the family a decision support system (software form similar tasks, including Wigwam in the UK
to help you make a decision) may be appropriate. (Anon., no date) and GeoData in the USA (Esri,
GIS can act in this role. 1995). These systems have been designed to make
Much of the data needed to help answer the ques- it possible for a home buyer to visit an estate agent,
tions posed above can be gathered and converted explain the type of house and neighbourhood
into a format for integration in GIS. Heywood et al. they prefer, and come away with a map showing
(1995) have successfully completed this exercise and the locations of houses for sale which meet their
created a house-hunting decision support system. To requirements. These products bring help to the
find a suitable home, participants were first required home buyer deciding where to look for a new home
to decide which of a series of factors (insurance costs, in an unfamiliar area. GIS in this context is a deci-
proximity to schools, railways and roads, urban sion support system. Online applications such as
areas) were important in their decision making. upmystreet.com and Google Street View also allow
These factors were allocated weights and scores users to search for properties and explore the char-
reflecting their importance. Constraints, areas where acteristics of the area in which they are found.
a new home would not be suitable under any con- The house-hunting example shows how GIS
ditions, were also identified. Constraints excluded can be used to link databases with similar types of
certain areas from the analysis altogether; for ex-
data. This improves the speed and efficiency with
ample, participants could decide that they did not
which an appropriate location can be found. In this
wish to live within 500 m of a major road.
respect, it is a similar application to the NDA case
Once the weighting process had been completed,
study discussed earlier. However, it differs from
the data selected were combined in a GIS using a
the NDA example in that a large part of the search
multi-criteria modelling technique. This technique
process may be carried out using the attributes
will be explored in more detail in Chapter 7, but, in
brief, the data layers were combined using weight- associated with a spatial feature. In the case of the
ings, so that the layer with the highest weight had house-hunting example, these may be the number
the most influence on the result. The resulting of bedrooms a property has or its price. Heywood
maps were used to help target the house-hunting et al. (1995) raise other issues associated with the
process. The method is summarized in Figures 1.12 use of GIS as a decision support tool. These are the
and 1.13. problem of different GIS software products giving
Locations of houses that were for sale were plot- different results, the problem of defining search
ted over the top of the suitable areas identified, and criteria and the human constraints on the decision-
ranked according to the number of criteria which making process (Box 1.4).
they meet. If further details of these houses were The examples above are not enough to illustrate
available in computerized form, they were accessed the range of applications and problems that GIS can
by pointing at the map to find out: for instance, how be used to address. Even when a problem cannot be
many bedrooms they had, or whether they had a solved entirely using GIS, there may be the poten-
garden. To achieve this, the information on the map tial for some GIS input to aid the decision-making
(locations of properties) was linked with a database process. Table 1.1 offers additional pointers to
of house features. This type of data is often referred further examples of GIS use by local government,
to as ‘attribute’ data. Attribute data and database defence agencies, utility companies, commerce
concepts are considered in detail in Chapter 4. and business.
Introduction 15
House hunting
criteria
identified
Would like to
Would like to
Must be close Must live near live in an area
live close to
to a school a main road with a low crim e
an urban area
rate
Map showing
Ma p Ma p Map showing
insurance zones
showing school showing main location of
as a surrogate
locations roads urban centre
measure for
crime rates
GIS procedure
Map showing Map showing Map showing
used to extract
proximity to proximity to proximity to
areas of low
school main road urban centre
insurance
Decide
to look at houses No
in these
areas?
Yes
View houses
(a) Railway constraint (b) Countryside constraint (c) Proximity to roads (d) Proximity to school
(e) Combination of railway constraint and countryside (f) Combination of proximity and constraint maps
constraint [(c), (d) and (e)] with proximity to road used as the
most important factor
LEGEND
Motorway Most suitable
U-road Acceptable
Estuary
Urban Area
STUDY
CASE
house-hunting case study
OO Heywood et al. (1995) compared the results they but in reality it may be that you wish to be in
obtained using two GIS software products to iden- close proximity to a swimming pool and a medi-
tify appropriate neighbourhoods for one house cal centre. If data are not available for these
buyer. The results were different. The differences factors, they cannot be included in the analysis.
were in part explained by small differences in the So, defining the problem, and identifying all rel-
search methods used by each GIS. Different GIS evant criteria, are crucial steps in the design of GIS
will implement similar methods in slightly differ- projects. The result you obtain will be influenced
ent ways, and there will also be variations due to by the questions you ask. If you do not ask the
the way data are stored in the GIS. Therefore, a right questions, you will not get the right answer.
clear understanding of the way GIS software works Therefore, good project design is an essential
is crucial if you are to be able to understand and component of using GIS. In Chapter 12 we provide
explain your results. Chapters 2 to 8 expand on you with a methodology to help you plan your own
these issues. GIS project.
OO Heywood et al. (1995) also considered that there OO Human factors such as awareness and training also
could be difficulties with results due to the way influence the effectiveness of GIS as a decision
the problem was defined at the outset. For exam- support system as they will help the user formu-
ple, the limited selection of data suggested above late appropriate questions. Chapter 11 looks at
may be available to help with the site selection, these issues in more detail.
TABLE1.1 ApplicationareasforGIS
REFLECTIONBOX
OO Make a note of all the questions you have asked GIS to help you find a new home? Explain the rea-
or heard recently that have a spatial component. sons for your answer.
Can you classify them into questions about loca- OO Think about the impact of GIS on the case stud-
tion, patterns, trends, conditions and implications? ies. Has the use of GIS been beneficial? Have there
OO In the house-hunting example GIS has been used been any problems? To what extent might GIS be
to try to improve the method we use when search- regarded as a unifying technology which spans
ing for a new home. Would you consider using a many different disciplines in each instance?
of the shorter definitions give an idea of what a GIS OO search for particular characteristics or features
elements (the computer system, data or processing At the other extreme, the components of a GIS
tools) will function as a GIS in isolation, so all might include: the computer system (hardware and operat-
be considered of equal importance. However, it is ing system), the software, spatial data, data management and
perhaps the nature of the data used, and the atten- analysis procedures and the people to operate the GIS.
tion given to the processing and interpretation of In addition, a GIS cannot operate in isolation from
these data, that should lie at the centre of any defi- an application area, which has its own tradition of
nition of GIS. ideas and procedures. It is this more comprehensive
GIS draws on concepts and ideas from many perspective that is adopted here.
different disciplines. The term ‘Geographic
Information Science’ has been adopted to refer
to the science behind the systems. Geographic Computer systems and software
Information Science draws on disciplines as diverse GIS run on the whole spectrum of computer sys-
as cartography, cognitive science, computer science, tems ranging from portable personal computers
engineering, environmental sciences, geodesy, land- (PCs) to multi-user supercomputers, and are pro-
scape architecture, law, photogrammetry, public grammed in a wide variety of software languages.
policy, remote sensing, statistics and surveying. Systems are available that use dedicated and expen-
Geographic Information Science involves the study sive workstations, with monitors and digitizing
of the fundamental issues arising from the creation, tables built in; that run on bottom-of-the-range PCs
handling, storage and use of geographic informa- or notebooks; and that run on portable Personal
tion (Longley et al., 2005), but it also examines the
Data Assistants (PDAs), tablet PCs or handheld GIS/
impacts of GIS on individuals and society and the
GPS devices (Figure 1.14). In all cases, there are a
influences of society on GIS (Goodchild, 1997). Mark
number of elements that are essential for effective
(2003) considers the definition and development of
Geographic Information Science in detail. GIS operation. These include (after Burrough, 1986):
Goodchild (1997) offers a useful summary of key OO the presence of a processor with sufficient power
concepts that help with the definition of GIS: to run the software;
OO Geographical information is information about OO sufficient memory for the storage of large volumes
places on the Earth’s surface. of data;
OO Geographic information technologies include OO a good quality, high-resolution colour graphics
global positioning systems (GPS), remote sensing screen; and
and geographic information systems. OO data input and output devices (for example,
OO Geographical information systems are both digitizers, scanners, keyboard, printers and
computer systems and software. plotters).
OO GIS can have many different manifestations.
Likewise, there are a number of essential soft-
OO GIS is used for a great variety of applications. ware elements that must allow the user to input,
OO Geographic Information Science is the science store, manage, transform, analyse and output data.
behind GIS technology. Discussion of these issues follows in Chapters 4
to 8. However, although GIS generally fit all these
requirements, their on-screen appearance (user
OO COMPONENTS OF A GIS interface) may be very different. Some systems still
require instructions to be typed at a command line,
There is almost as much debate over the compo- while others have ‘point and click’ menus operated
nents of a GIS as there is about its definition. At using a mouse. Examples of popular GIS interfaces
the simplest level, a GIS can be viewed as a software are shown in Figure 1.15. The type of interface indi-
package, the components being the various tools vidual users find easier to operate is largely a matter
used to enter, manipulate, analyse and output data. of personal preference and experience.
20 Chapter 1 What is GIS?
Spatial data
All GIS software has been designed to handle spa-
tial data (also referred to as geographical data).
Spatial data are characterized by information about
position, connections with other features and
details of non-spatial characteristics (Burrough,
1986; Department of the Environment, 1987). For
example, spatial data about one of Happy Valley’s
weather stations (Figure 1.16a) may include:
OO latitude and longitude as a geographical
reference. This reference can be used to deduce
relationships with nearby features of interest. If
the latitude and longitude of a weather station
are known, the relative position of other weather
stations can be deduced, along with proximity to
ski slopes and avalanche areas;
(a) Command line
OO connection details such as which service roads,
lifts and ski trails would allow the meteorologist
access to the weather station;
OO non-spatial (or attribute) data: for instance,
details of the amount of snowfall, temperature,
wind speed and direction.
In a similar way spatial data about a ski piste
(Figure 1.16b) may include:
OO a series of spatial references to describe position;
OO details of other runs that cross or join the ski piste;
OO attribute data such as the number of skiers using
the piste and its standard of difficulty.
Data management and analysis procedures This process should include verification procedures
The functions that a GIS should be able to perform to check that the data are correct and transforma-
include data input, storage, management, transfor- tion procedures to allow data from different sources
mation, analysis and output. Data input is the process to be used. GIS need to handle two types of data –
of converting data from its existing form to one graphical data and non-spatial attribute data. The
that can be used by the GIS (Aronoff, 1991). It is the graphical data describe the spatial characteristics
procedure of encoding data into a computer-read- of the real-world feature being modelled. For ex-
able form and writing the data to the GIS database. ample, the hotels in Happy Valley may be described
24 Chapter1 What is GIS?
(a) Raster digital elevation model (b) Raster land cover data (c) Satellite image
(d) Digital colour aerial photograph (e) Vector contours and roads
(f) Vector soil polygons (g) Vector census polygon (h) Vector land cover polygons
boundaries
Figure1.18 Examples of raster (a–d) and vector GIS data layers (e–h)
Components of a GIS 25
by a series of points. In some cases, particularly (DBMS). A DBMS is a set of computer programs for
when area and line features are used to model real- organizing information, at the core of which will be
world features, the graphical data may include a database. Database applications that have no GIS
information about the linkages between them. For component include management of payrolls, bibli-
example, if the boundary of an area feature such as ographies, and airline booking systems. In the same
a car park is also a snow fence that prevents skiers way that DBMS organize these different types of
from overshooting the nursery slopes, this infor- data they can be used to handle both the graphical
mation may be stored with the graphical data. and non-graphical elements of spatial data. An ideal
Non-spatial attribute data describe what the features GIS DBMS should provide support for multiple
represent. They tell the computer what a particular users and multiple databases, allow efficient updat-
set of entities represents (for instance, a set of points ing, minimize repeated (or redundant) information
and allow data independence, security and integrity
may represent hotels). In addition, further non-spa-
(Smith et al., 1987). Relational databases, flat files
tial attribute data may be stored which provide extra
and other database models used by GIS will be dis-
information about the hotels (standard, number of
cussed in more detail in Chapter 4.
rooms and restaurant facilities).
It is the ability of GIS to transform spatial data – for
Data input and updating are frequently the example, from one entity type (points, lines and
most expensive and time-consuming part of any areas) to another – and to perform spatial analysis,
GIS project and their importance and complexity that distinguishes GIS from other types of informa-
should never be underestimated. Approximately tion systems.
80 per cent of the duration of many large-scale Transformation is the process of changing the
GIS projects is concerned with data input and representation of a single entity, or a whole set of
management. Aronoff (1991) estimates that the con- data. In GIS, transformation may involve changing
struction of a large database could cost five to ten the projection of a map layer or the correction of
times more than the GIS software and hardware. systematic errors resulting from digitizing. In addi-
Data input methods are discussed in Chapter 5. tion, it may be necessary to convert data held as
The data management functions necessary in any rasters to vectors or vice versa.
GIS facilitate the storage, organization and retrieval Aronoff (1991) classifies GIS analysis procedures
of data using a database management system into three types (Figure 1.19):
(a) Soil map display (b) Reselection of specific soil (c) Simple soil erosion model
types predictions
1 Those used for storage and retrieval. For example, 3 Modelling procedures, or functions, for the
presentation capabilities may allow the display of prediction of what data might be at a different
a soil map of the area of interest. time and place. Predictions could be made
2 Constrained queries that allow the user to look about which soils would be highly vulnerable to
at patterns in their data. Using queries, only erosion in high winds or during flooding, or the
erodible soils could be selected for viewing or type of soil present in an unmapped area.
further analysis.
Transformation and analysis procedures can also photographed, stored digitally or plotted to produce
be classified based on the amount of data analysed. permanent hard copy. Most GIS provide the ability
Data in GIS are normally held in a series of layers. to design screen formats and forms for plotting and
For instance, a 1:50,000 topographic map might be these help to ensure that all maps have titles, keys,
digitized to create a series of layers – one layer for north arrows and scales, just as with traditional carto-
road data, one for buildings, one for recreational graphic output. The facilities available for map design
interest (parking, picnic sites and youth hostels) and can be very extensive, incorporating myriad of dif-
additional layers for soils and population data. Data ferent colours, symbols and line styles, and it is often
layers normally contain data of only one entity type: possible to design additional symbols for your own
that is, point, or line, or area data. This is illustrated use. This can make map design very time-consuming,
in Figure 1.20. Analysis can be carried out either but effective. The audience for any GIS product is an
on one layer at a time, or on two or more layers in important consideration when designing output, but
combination. The techniques available in GIS for generally it is best to keep products clear and simple.
manipulating and correcting data in preparation for Options for data output and communication of
analysis, and the analysis methods available, are dis- results from GIS are discussed further in Chapter 8.
cussed in greater depth in Chapters 5, 6 and 7.
The form of data output used will depend on cost People and GIS
constraints, the audience to whom the results are
directed and the output facilities available. A local Most definitions of GIS focus on the hardware,
government agency may produce simple tables, software, data and analysis components. However,
graphs and maps for the communication of impor- no GIS exists in isolation from the organizational
tant points to politicians, whilst professional map context, and there must always be people to plan,
makers may produce detailed plots for publication. implement and operate the system as well as make
In other cases, data may be output in digital form decisions based on the output. GIS projects range
for transfer to another software package for statisti- from small research applications where one user
cal analysis, desktop publishing or further analysis. is responsible for design and implementation and
However, most GIS output is in the form of maps. output, to international corporate distributed
These may be displayed on-screen for immedi- systems, where teams of staff interact with the
ate communication to individuals or small groups, GIS in many different ways (Figure 1.21). In most
REFLECTIONBOX
OO Look again at Table 1.1. Can you add some applica- match the definitions and component descriptions
tions for GIS in areas and activities in which you are given above? How could it be improved?
interested or involved? OO Try to summarize the components of a GIS in a
OO If you have a GIS of your own, or access to one at work diagram or table.
or college, make a list of all its components. Does it
CONCLUSIONS
GIS technology is now well established and, as nologies used in surveying and field data collection,
we will see in Chapter 9, has been in use since the visualization and database management will also
1960s. Some of the work cited in this and subsequent influence the development of GIS. Further comments
chapters may have been written several decades on the future of GIS can be found in Chapter 13.
ago – this is an indication of the maturity of GIS. The There have been some notable failures in GIS.
growth in application areas and products in recent Sometimes data difficulties or other technical prob-
years has helped GIS to become an accepted tool lems have set back system developments and
for the management and analysis of spatial data. applications; however, there are also human and
This trend is set to continue as computer technology organizational problems at the root of GIS failures.
continues to improve with smaller, faster and more Before we can begin to appreciate these fully, to
powerful devices, and as more data become avail- ensure that our GIS applications are successful, it is
able in digital formats directly compatible with GIS. important to have a good understanding of what a GIS
In addition, the striking advances in related tech- can do and the data it works with (Chapter 2).
Further study 29
Spatial data
Learning outcomes
By the end of this chapter you should be able to:
OO Give examples of map projections and explain why they are important
OO Define topology
PRACTICE
and secondary data used in
Happy Valley
PRIMARY DATA SOURCES OO The number of skiers using a specific lift on a
particular day. The automatic turnstiles at the
OO Daily snow pack data collected by the Happy Valley entry points to all the lifts collect these data. They
ski patrols. These data are used to help make are used to monitor lift usage.
decisions about which runs to open and which to OO The number of avalanches recorded by the ski
close. patrols. These data are used to help predict
OO Number of lift passes purchased each day. These avalanche risk.
data are used to monitor the demand for skiing on
different days of the week.
BOX 2.1
SECONDARY DATA SOURCES OO National and regional lifestyle data derived from
market research surveys are used to estimate
OO Published meteorological maps for the Happy the demand for skiing and target the marketing of
Valley area. These data are used to assist with Happy Valley.
avalanche forecasting. OO Datasets and activities relating to the Happy Valley
OO Local topographic maps. Back-country ski trail Case Study can be found online at www.pearsoned.
maps are prepared using local topographic maps. co.uk/heywood.
avalanche incident that took place in Three Pines to those who are familiar with the area. However,
Valley on 14 February 2002, the three modes are: because GIS have no ‘local knowledge’, all spatial
data used in GIS must be given a mathematical spa-
OO temporal – 15:30 hrs 14 February 2002;
tial reference. One of the most common is a map
OO thematic – wet slab avalanche triggered by two co-ordinate. Here, a co-ordinate pair (x,y) is used to
off-piste skiers; and locate the position of a feature on a uniform grid
OO spatial – Three Pines Valley, south-facing slope. placed on a map. Spatial referencing is considered in
The temporal dimension provides a record of more detail later in this chapter.
when the data were collected and the thematic It is common to find the term ‘temporal data’
dimension describes the character of the real-world used to describe data organized and analysed accord-
feature to which the data refer. Additional thematic ing to time, thematic data used for data organized
data for the avalanche incident might relate to the and analysed by theme, and spatial data for data
size and consequences of the avalanche. In GIS the organized and analysed by location. However, even
thematic data are often referred to as non-spatial or though one dimension may be used to organize data,
attribute data. These are illustrated in the avalanche the other dimensions will still be present.
incident report map in Figure 2.3. GIS places great emphasis on the use of the spatial
The spatial dimension of data can be regarded dimension for turning data into information, which,
as the values, character strings or symbols that in turn, assists our understanding of geographic
convey to the user information about the location phenomena. Therefore, we consider next the charac-
of the feature being observed. In the case of the teristics of spatial data in detail and examine how the
avalanche on 14 February 2002 we know that the map metaphor has shaped these characteristics. This
incident occurred on a south-facing slope in Three is followed by a review of the thematic dimension of
Pines Valley. In this case, the spatial reference used spatial data and discussion of a range of sources of
is a textual description that would only be of use spatial data, including surveys, aerial photographs,
satellite images and field data sources.
Examples of ratio scales are 1:5000 and 1:5,000,000. 1:10,000 or 1:25,000) cover small areas and contain
At a scale of 1:5000 a 1 mm line on the map repre- large amounts of detail. With some data used in
sents a 5000 mm line on the ground. In the same GIS, such as aerial photographs or satellite imagery,
fashion a line of 1 m on the map represents a line the scale is not immediately obvious and may have
of 5000 m on the ground; the units do not matter to be calculated by the user. Scale is also important
as long as they are the same. A verbal scale would when using spatial entities (points, lines and areas)
express the scale in words, for example ‘1 cm repre- to represent generalized two-dimensional versions
sents 50 m’. Finally, a graphic scale (or scale bar) is of real-world features.
usually drawn on the map to illustrate the distances
represented visually. Graphic scales are frequently
Spatial entities
used on computer maps. They are useful where
changes to the scale are implemented quickly and Traditionally, maps have used symbols to repre-
interactively by the user. In such cases, recalculat- sent real-world features. Examination of a map will
ing scale could be time-consuming, and the ratios reveal three basic symbol types: points, lines and
produced (which may not be whole numbers) may areas (Monmonier, 1996). These were introduced
be difficult to interpret. Redrawing a graphic scale in Chapter 1 (Figure 1.17) and are the basic spatial
in proportion to the map is relatively straight- entities. Each is a simple two-dimensional model
forward and simple to understand. It is often that can be used to represent a feature in the real
possible in GIS to specify the scale at which you world. These simple models have been developed
require your maps using a ratio representation. by cartographers to allow them to portray three-
Standard topographic maps contain examples dimensional features in two dimensions on a piece
of verbal, ratio and graphical scales. It should be of paper (Laurini and Thompson, 1992; Martin,
remembered that small-scale maps (for example, 1995). Box 2.2 provides more details on the types
1:250,000 or 1:1,000,000) are those that cover large of features that points, lines and areas can be used
areas. Conversely, large-scale maps (for example, to represent.
THEORY
POINTS LINES
Points are used to represent features that are too Lines are used to represent features that are linear in
small to be represented as areas at the scale of nature: for example, roads, powerlines or rivers (see
mapping being used. Examples are a postbox, a tree Figure 2.8). It can be difficult for a GIS user to decide
or a lamp post (see Figure 2.7). The data stored for a when a feature should be represented by a line.
postbox will include geographic location and details Should a road be represented by a single line along
of what the feature is. Latitude and longitude, or a its centre, or are two lines required, one for each side
co-ordinate reference, could be given together with of the road?
details that explain that this is a postbox in current A line is simply an ordered set of points. It is a
use. Of course, features that are represented by string of (x,y) co-ordinates joined together in order
points are not fully described by a two-dimensional and usually connected with straight lines. Lines
geographical reference. There is always a height may be isolated, such as geological fault lines,
component since the postbox is located at some or connected together in networks, such as road,
height above sea level. If three dimensions are pipeline or river networks. Networks are sometimes
important to a GIS application this may also be regarded as a separate data type but are really an
recorded, usually by adding a z value representing extension of the line type. More will be said about
height to give an (x,y,z) co-ordinate. networks and their analysis in later chapters. Like
Maps and their influence on the character of spatial data 39
BOX 2.2
points, lines are in reality three-dimensional. For some of these polygons exist on the ground,
instance, a hydro-geologist may be interested in whilst others are imaginary. They are often used
underground as well as surface drainage. Adding to represent area features that do not exist as
a z co-ordinate (representing depth or height) physical features, such as school catchment zones
to the points making up the line representing a or administrative areas.
stream allows an accurate three-dimensional Two types of polygons can be identified: island
representation of the feature. polygons and adjacent polygons. Island polygons
occur in a variety of situations, not just in the case
AREAS of real islands. For example, a woodland area may
Areas are represented by a closed set of lines and appear as an island within a field, or an industrial
are used to define features such as fields, buildings estate as an island within the boundary of an urban
or lakes (see Figure 2.9). Area entities are often area. A special type of island polygon, often
referred to as ‘polygons’. As with line features, referred to as a ‘nested’ polygon, is created
40 Chapter 2 Spatial data
BOX 2.2
(c) Lake
Figure 2.9 Real-world objects commonly represented as an area
by contour lines. If you imagine a small conical hill A three-dimensional area is a surface. Surfaces
represented by contour lines, this will be can be used to represent topography or non-
represented in polygon form as a set of concentric topographical variables such as pollutant levels or
rings. Adjacent polygons are more common. Here, population densities. Some authors (for example
boundaries are shared between adjacent areas. Laurini and Thompson, 1992; Martin, 1995) consider
Examples include fields, postcode areas and surfaces to be a separate fourth entity type. This
property boundaries. issue will be considered in more detail in Chapter 3.
The representation of real-world features using would be the most appropriate method of repre-
the point, line and area entity types appears rela- sentation, given the number of cities to be included.
tively straightforward. However, the method chosen However, at national and regional scales a point
to represent a spatial feature will depend on the could provide an oversimplified view of the extent of
scale used. Consider the way cities are represented the geographical area covered by a city. A point used
on maps of different scales. On a world map a point here would tell us nothing about the relative size of
Maps and their influence on the character of spatial data 41
(a) OS ‘Route’ map 1:625,000 (b) OS ‘Travel’ map 1:250,000 (c) OS ‘Tour’ map 1:250,000
(d) OS ‘Landranger’ map 1:50,000 (e) OS ‘Explorer’ map 1:25,000 (f) OS ‘Landplan’ map 1:10,000
cities, so it is more likely that the cartographer would for the representation of features such as telephone
choose to represent the cities using areas. At the local boxes, areas for residential blocks and parks, and lines
scale even the area spatial entity may be considered for road networks. This is illustrated in Figure 2.10
too simplistic and the cartographer may choose to for the city of London and shows how choosing the
build up a representation of the city using a mixture appropriate entity to represent real world features is
of point, line and area entities. Points may be used often surprisingly difficult.
42 Chapter 2 Spatial data
1: 20,000,000 1: 5,000,000
Ohio
Ar
ka
ns
as
Mississippi
Re
d
ippis
Missis
1: 250,000 1: 50,000
20m
30m
river will have been smoothed to create a simple, river is 500 m wide. In reality it may be only 50 m
easy to understand map. At larger scales (for ex- wide. If the width of line features drawn on the map
ample, 1:5,000,000) it is possible to show something of were determined rigidly by the map scale, then most
the meandering nature of this great river and its net- features on small-scale maps could not be seen with
work of tributaries: the Red, Arkansas, Missouri and the naked eye. For example, a 50 m wide river on a
Ohio. At even larger scales (for example, 1:250,000) 1:1,000,000 scale map would have to be drawn using
it becomes possible to indicate width, river banks, a line only 0.05 mm wide. Similarly, if a road run-
small bends and meander cut-offs. At larger scales ning along the banks of the river were to be depicted
still (for example, 1:50,000) it may be possible to accurately in this fashion, it would need to be drawn
indicate depth and the positions of sandbanks and on top of the river on the map. In order to make the
shoals that might be important for navigation pur- road distinguishable from the river, the cartogra-
poses. As the scale increases, the cartographer has pher has to displace the road to leave a gap between
greater scope for including more detail. The rela- it and the river. To cope with these and other prob-
tionship between scale and detail is referred to as lems relating to the necessary generalization of map
scale-related generalization and is illustrated in Figure 2.11. features, cartographers have adopted a broad code
Decisions regarding what features to include on of practice relating to selection, simplification, dis-
the final map and which to leave out also need to be placement and smoothing. This is summarized in
made by the cartographer. If the cartographer were Box 2.3. Remember that most maps are just com-
to include every single tributary of the Mississippi munication devices; a way of storing geographical
river network on the 1:20,000,000 scale map, the map information on paper and passing this information
would be covered by dense blue line work and impos- on to others. Maps have a long history, going back
sible to read. For the sake of clarity, the cartographer as far as the ancient civilizations and most were
has to be selective about drawing map features. never originally intended to be used as a data source
Another problem facing the cartographer is how for GIS. A good understanding of the processes of
to depict features in proportion to their size on the cartographic generalization is therefore important if
ground. If a river is drawn as a line 0.5 mm thick on data from paper maps and other spatial data sources
a 1:1,000,000 scale map, this would imply that the are to be used effectively within GIS.
generalization: code of
practice
1 Selection. First, the map feature for generaliza- top of one another, the cartographer may choose to
tion is selected. If more than one source is available displace them by a small degree so that they are both
to the cartographer this may involve choosing the visible on the map image. This may have the effect of
most appropriate representation of the feature or a displacing a feature several hundred metres depend-
blending of the two. ing on the map scale used.
2 Simplification. Next, a decision will be taken to 4 Smoothing and enhancement. If the source data
simplify the feature. For the example of the river this from which a cartographer is working are very angu-
may involve the removal of some minor bends. The lar, because they have been collected from a series
aim of generalization will usually be to simplify the of sampling points, a smoothing technique may be
image but maintain the overall trend and impression used to apply shape and form to the feature. This will
of the feature. give a better representation.
3 Displacement. If there are features that are
located side by side in the real world, or that lie on Source: Adapted from Robinson et al. (1995)
44 Chapter 2 Spatial data
Figure 2.12 The Earth from space and some commonly used global
map projections
Sources: (a) NASA; (b, d, e) Illinois State University Microcam website, www.
ilstu.edu/microcam/map_projections/Conic/Lambert_Conformal_conic.pdf
by permission of Dr Paul B. Anderson; (c) From John Savard’s homepage,
(e) Mercator www.members.shaw.ca, by permission of John Savard
of the Earth. The pattern of land masses and oceans distorted. However, at the top or bottom of the wall
are laid out in a familiar way. The process of trans- the location of the countries is distorted, with the dis-
ferring the spherical Earth onto a two-dimensional tance between countries increased. Our view of the
surface introduces errors into spatial data, the poles will be very distorted, or missing altogether. In
character of which will vary depending on the fact, if the poles are included on our projected map,
projection method chosen. Some projections will the points representing the north and south poles
cause distance between spatial entities to be pre- become so distorted as to be projected as a straight
served whilst direction is distorted. In other cases, line equal in length to that of the equator. Of course,
shape may be preserved at the expense of accurate we also need to ‘cut open’ the image projected onto
area estimates. Figure 2.12 illustrates how differ- the walls so we can unroll it and lay it flat and this
ent parts of the Earth are distorted to enable a fit introduces the problem of discontinuity. When
onto a flat sheet of paper. One way to visualize the we ‘step off’ the map at its vertical edges we will re-
problem of representing a spherical world in two appear at the same latitude on the opposite edge.
dimensions is to imagine a plastic beach ball over- In a square room with flat walls, only a part of
printed with a map of the world showing lines of the Earth’s surface will be visible on any one wall.
latitude and longitude. The inflated ball is a globe, This will depend on the position of the light. The
with the countries in their correct locations, and view will be of half the globe or less. Distortion will
shown as area entities with correct relative shapes be similar to that in the circular room except that,
and sizes. Imagine that you have to deflate the ball since the single wall is straight and not curved, the
and lay it flat on a table whilst still displaying all image of the world will be distorted at all four edges
the countries. The only way to do this is to cut the of the wall and not just the top and bottom.
beach ball into pieces. In doing this you would find If the room is shaped like a tepee (with circu-
that the distances between countries will be altered lar walls tapering towards a point at the apex of
and their shape distorted. The principle is the same the structure) the line of true scale is no longer the
with map projections. equator, as in the circular room, but some line of
If you imagine the beach ball has a hole at the latitude nearer the north pole that lies at a tangent
‘north pole’ large enough for a light bulb to be to the sloping walls of the tepee. The exact line of
inserted, it is transformed into a light fitting. When latitude will depend on the angle of the tepee walls:
the light is switched on, an image of the surface the steeper the walls, the lower the latitude (nearer
is projected onto the walls of the room. Careful the equator); the shallower the walls, the higher the
examination of the images on the walls reveals that latitude (nearer the pole). Not all projections of this
the centre of the image reflects the globe most accu- kind have the north pole uppermost; projections can
rately. It is on this simple concept that the whole have the south pole uppermost or may have the apex
range of map projections is based. of the tepee centred over a different point altogether.
Today there are a wide range of map projections The circular room is equivalent to the family of
in use, and there were even more used in the past. cylindrical projections (which includes the Mercator
Different map projections are used in different parts projection), where the surface of the Earth is pro-
of the world for mapping different sized areas and jected onto a cylinder that encompasses the globe
for different applications. Think again of the globe (Figure 2.13a). This projection is very suitable for
as a light fitting. The picture of the Earth from our making maps of an area that have only a small
‘globe-light’ will vary depending upon the shape of extent in longitude. It has been chosen as the basic
the room in which the light is placed. projection for use by the Ordnance Survey to map
In a circular room, assuming our globe is hang- the UK. The transverse Mercator projection has
ing from the north pole, there will be a continuous the advantage of maintaining scale, shape, area and
picture of the Earth. Countries nearest the equator bearings for small areas. This explains why it has
will appear in their true relative geographical posi- become a popular projection for mapping small
tions. The equator is the line of latitude nearest the areas of the globe. The single wall illustrates the azi-
wall and so represents the line of true scale, along which muthal family of projections (Figure 2.13b) and the
distances (and consequently the map scale) are not tepee is equivalent to the conic family (Figure 2.13c).
46 Chapter 2 Spatial data
(b) Azimuthal projection (light in a square room with flat walls analogy)
• Area is distorted
• Distance is very distorted towards
the bottom of the image
• Scale for the most part is
preserved
Many of the map-based spatial data sources be slight, but at small scales (covering large areas)
used in GIS have a projection associated with the effects can be substantial. Finally, since one of
them. To undertake meaningful analysis it is ne- the functions of a GIS application is to allow the
cessary to know something about the projections integration of data from different sources, the
being used. The results of analyses will be affected ability to alter projections is a fundamental abil-
in different ways by different map projections. If ity of many GIS. There are hundreds of different
a GIS application requires the accurate calcula- map projections and some GIS seem to offer the
tion of areas, then using a projection that distorts capability to re-project data for most of these. Only
areas is obviously not suitable. When using data at the most common projections have been consid-
large scales (covering small areas) the effects may ered above.
Maps and their influence on the character of spatial data 47
Spatial referencing longitude are widest apart at the equator and closest
together at the poles. The relative distance between
A referencing system is used to locate a feature on the
lines of longitude where they intersect lines of lati-
Earth’s surface or a two-dimensional representation
tude (or parallels) is always equal. However, the
of this surface such as a map. There are a number of
real distance will vary depending on the line of lati-
characteristics that a referencing system should have.
tude that is intersected. For example, the distance
These include stability, the ability to show points,
between the lines of longitude intersecting the same
lines and areas, and the ability to measure length, size
parallel will increase towards the equator, with the
(area) and shape (Dale and McLaughlin, 1988). Several
maximum distance existing at the equator itself.
methods of spatial referencing exist, all of which can
Lines of latitude lie at right angles to lines of lon-
be grouped into three categories:
gitude and run parallel to one another. Each line of
OO geographic co-ordinate systems; latitude represents a circle running round the globe.
OO rectangular co-ordinate systems; and Each circle will have a different circumference and
area depending on where it lies relative to the two
OO non-co-ordinate systems. poles. The circle with the greatest circumference is
The only true geographic co-ordinates are latitude and the equator (or central parallel) and lies equidistant
longitude. The location of any point on the Earth’s from the two poles. At the two poles the lines of lat-
surface can be defined by a reference using latitude itude are represented by a single point – the pole.
and longitude. Lines of longitude (also known as Using lines of latitude and longitude any point
meridians) start at one pole and radiate outwards on the Earth’s surface can be located by a reference
until they converge at the opposite pole (Figure given in degrees and minutes. For example, the
2.14). Conceptually they can be thought of as semi- city of Moscow represented as a point can be given
circles. If you slice a globe along two opposing lines a geographical co-ordinate reference using latitude
of longitude you will always cut the globe in half. and longitude of 55 degrees 45 minutes north and
The arbitrary choice for a central line of longitude 36 degrees 0 minutes east (55° 45’N 36° 0’E). The first
is that which runs through the Royal Observatory set of numbers, 55° 45’N, represents latitude. The N
in Greenwich in England, and is hence known as informs us that Moscow can be found north of the
the Greenwich meridian or the prime meridian. Lines of equator. The second set of numbers, 36° 0’E, tells us
that Moscow lies to the east of the prime meridian.
Therefore, the N and E together give the quarter of
the globe in which Moscow is located (Figure 2.15a).
North Pole
The line of latitude on which Moscow lies is given
(lines of longitude converge)
by the degrees and minutes of this latitude away
80 from the equator (Figure 2.15b). Finally, the line
60
of longitude on which Moscow lies must be identi-
40 fied. Figure 2.15c shows how this angle is calculated
based on relative distance from the prime meridian.
Prime Meridian
20
Lines of
Adopting this approach, all features on the surface
60 Equator 60
40 20 0 20 40 latitude of the Earth can be located relative to one another
and the distance between them calculated. The
shortest distance between two points on the Earth’s
surface is known as the great circle distance.
The latitude and longitude referencing
system assumes that the Earth is a perfect sphere.
South Pole
Unfortunately this is not correct. The Earth is actu-
Lines of longitude
ally an oblate spheroid somewhat like an orange
with flatter poles and outward bulges in equatorial
regions. To complicate matters further, the surface
Figure 2.14 Latitude and longitude of the Earth is far from smooth and regular, as you
48 Chapter 2 Spatial data
PRACTICE
BOX 2.4 Ordnance Survey
National Grid system
The Ordnance Survey National Grid is a rectangular Shetland. This is divided into 500 km squares, which
grid system based on the transverse Mercator are then divided into twenty-five 100 km squares.
projection (Figure 2.16). The grid is 700 × 1300 km Each 100 km square is identified by two letters. The
covering all of Great Britain from the Scilly Isles to first refers to the 500 km square and the second to the
Kilometres
Northing
1300 0
HL HM HN HO HP JL 9
(N02) (N12) (N22) (N32) (N42) (N52)
8
1200 7
HQ HR HS HT HU JQ 6
(N01) (N11) (N21) (N31) (N41) (N51)
5
1100
4
HV HW HX HY HZ JV
(N00) (N10) (N20) (N30) (N40) (N50) 3
1000 2
NA NB NC ND NE OA 1
(09) (19) (29) (39) (49) (59)
0
900 0 1 2 3 4 5 6 7 8 9 0
100 km square SE
NF NG NH NJ NK OF
(08) (18) (28) (38) (48) (58) 00
800 99
NL NM NN NO NP OL 98
(07) (17) (27) (37) (47) (57)
97
700 96
NQ NR NS NT NU OQ 95
(06) (16) (26) (36) (46) (56)
94
600
93
NW NX NY NZ OV OW
(15) (25) (35) (45) (55) (65) 92
500 91
SB SC SD SE TA TB 90
(14) (24) (34) (44) (54) (64) 30 31 32 33 34 35 36 37 38 39 40
10 km square SE 39
400
930
SG SH SJ SK TF TG
(13) (23) (33) (43) (53) (63) 929
300 928
SM SN SO SP TL TM 927
(12) (22) (32) (42) (52) (62) 926
200 925
SQ SR SS ST SU TQ TR 924
(01) (11) (21) (31) (41) (51) (61)
P 923
100
922
SV SW SX SY SZ TV
(00) (10) (20) (30) (40) (50) 921
0 920
0 100 200 300 400 500 600 700 360 361 362 363 364 365 366 367 368 369 370
1 km square SE 36 92
False Origin of National Grid Kilometres
P = SE 366 923
Easting
BOX 2.4
100 km square. Each 100 km square is further divided 100 km square. An example could be SE 366 923. Here,
into one hundred 10 km squares (10 km × 10 km), and the ‘SE’ denotes the 100 km square that has its origin
the 10 km squares are divided into one hundred 1 km 400 km east and 400 km north of the origin of the grid.
squares (1 km × 1 km). Grid references are commonly The ‘366’ and the ‘923’ are the easting and northing
given as six figures prefixed by the letters denoting the recorded to the nearest 100 m.
coverage of all areas where people reside and work. All spatial referencing systems have problems
Providing that individual codes do not refer to single associated with them. Some are specific to the refer-
addresses, they also provide a degree of confiden- encing system, such as the updating problems with
tiality for data released using this as a referencing postcodes or the difficulties caused by geographi-
system. Box 2.5 provides more details on the UK cal co-ordinates with respect to map projections.
postcode system. However, some of the problems stem from the
In the western United States another non-co- nature of the spatial entities that require referencing:
ordinate referencing system is often used. This is OO Spatial entities may be mobile. Animals, cars and
known as the Public Land Survey System (PLSS).
people move; therefore any spatial reference they
Here, there has been a recursive sub-division of are tagged with will only represent their known
the land into quarter sections. By knowing which location at a particular time.
section you are in, you can reference yourself to
OO Spatial entities may change. Rivers meander,
the Earth’s surface (DeMers, 2005). Other non-co-
roads can be relocated and policy areas redefined.
ordinate referencing systems in use are based on
administrative areas: for example, the units used for OO The same object may be referenced in different
aggregation and presentation of population census ways. A house may be represented and referenced as
data in different countries. For referencing within both a point and an area on maps of different scales.
smaller areas, unique feature references may be An additional problem for the GIS user is the
used: for instance, the property reference numbers large number of different spatial referencing systems
used by a local authority, or the pipeline references in use. Choosing an appropriate referencing system
used by a utility company. can be difficult, and it will frequently be necessary to
PRACTICE
In the UK the postcode system was developed 15 items of mail per day (Raper et al., 1992). The
about 25 years ago by the Royal Mail to help post system is very widely used in application areas
sorting and delivery. Each code has two parts – the such as health, marketing and education because
outward code and the inward code. The postcode of its ease of collection and widespread use
system is hierarchical. The first one or two letters for address-based data. However, as with any
refer to a postcode area; these are followed by other postal code system there are problems:
subsequent numbers and letters subdividing OO For entities without an address, a postcode
this into districts, sectors and unit postcodes system is useless. Entities without addresses
(Department of the Environment, 1987) (Figure include rivers, trees, fields and phone boxes.
2.17). The system is further complicated by the OO The spatial units – postal areas, districts and units
existence of single-user postcodes for business – were designed to help mail delivery and bear no
users and addresses which receive more than relationship to other spatial units commonly used
Maps and their influence on the character of spatial data 51
BOX 2.5
(a) (b)
Figure 2.17 The UK postcode system
Source: (a) Reproduced by permission of Ordnance Survey on behalf of HMSO. © Crown Copyright 2011. All rights reserved.
Ordnance Survey Licence number 100030901
by those handling spatial information. However, in OO Some buildings have more than one postcode.
the UK there is a link to the Ordnance Survey grid Office blocks containing different companies, or
reference and census enumeration districts. blocks of flats where there are separate entrances
OO Changes occur to postcodes. In the UK there is or letter-boxes, may have several postcodes.
a three-month update cycle and approximately OO When comparing and plotting population distri-
18,000 changes are made each year. Changes may bution maps, it should be remembered that unit
be corrections, due to the construction or demo- postcodes, which cover approximately 15 houses,
lition of properties, or to the movement of large will represent very small areas in urban environ-
users (who may be eligible to keep their postcode ments, but may be huge in rural areas.
if they move within the same sector). Sources: Adapted from Dale and McLaughlin (1988); Raper
et al. (1992)
as it relates to spatial data, consists of three ele- with fewer than 20 bedrooms of luxury standard?’
ments: adjacency, containment and connectivity require the analysis of attribute data associated with
(Burrough, 1986). the point entities used to represent the location of
Adjacency and containment describe the geo- hotels (see Chapter 4).
metric relationships that exist between area The character of attribute data themselves can
features. Areas can be described as being ‘adjacent’ influence the utility of data sets in GIS analysis. One
when they share a common boundary. For ex- characteristic which is of considerable importance is
ample, the ski slopes and car parks in Happy Valley the scale of measurement used to record and report
may be adjacent. Containment is an extension of the data. For example, every year the managers of
the adjacency theme and describes area features Happy Valley complete a table for a ski resort guide.
that may be wholly contained within another area For this table they provide the name of the ski area,
feature such as an island within a lake. Connectivity its ranking (1st, 2nd, 3rd, 4th most popular), its
is a geometric property used to describe the linkages average winter temperature and the size of the ski
between line features. Roads are usually connected area. Each item of data uses a different scale of meas-
together to form a road network through which urement. The names given to these scales are
traffic can flow. nominal, ordinal, interval and ratio. Table 2.1 shows
An understanding of the geometric relationships each of these scales in relation to the data collated
between spatial entities is important for analysis and for the ski resort guide. Each scale of measurement
integration in GIS. Without knowledge of how enti- dictates how the data can be used.
ties are geometrically related to each other, it is
impossible to answer questions such as ‘What is the TABLE 2.1 Scales of measurement
shortest route from A to B?’ or ‘How many ski slopes
Data Unit of Scale
lie within or are next to zones of high avalanche risk?’
measurement
REFLECTION BOX
OO Pick a GIS project or application in which you are – for example, in the generalization of features or
involved or one that you have read about. Identify the level of detail shown. Does the representation
all the data sources that are used in the project of any features change (for example from point to
and categorize them into primary and secondary area features) as you change the scale?
sources. Create a summary similar to that pro- OO Visit some of the websites suggested at the end
vided for Happy Valley in Box 2.1. of the chapter which allow you to investigate map
OO Use a GIS or online mapping tool to display a map projections further. Try to identify a projection
for an area of interest and experiment with the which maintains shape, one which maintains dis-
scale functions. You may be able to change the tances and one which maintains area. For what
scale by zooming in and out, and you may be able type of GIS applications might your three examples
to specify a particular scale for display. Look for be useful?
differences between small- and large-scale maps
Arithmetic operations, whilst possible on ordinal from that scale, the GIS is unlikely to indicate
data, will again give meaningless results. when impossible or meaningless operations have
On an interval scale the difference between num- been carried out. To a computer numbers are all
bers is meaningful but the scale does not have a real the same and will be treated in the same ways. So,
origin. Temperatures, in degrees Celsius, are a good ranked scores for city sizes may be added together.
example. On a temperature scale it is possible to say Two different soil types could have a numerical
that there is a 10-degree difference between a ther- code to tag them to the appropriate area in a GIS.
mometer that records a value of 10 degrees and one If clay soils have the value 2 and sandy soils 3 on a
that records a value of 20 degrees. Thus, differences nominal scale, multiplying them together to give
can be calculated. However, it would be incorrect soil class 6 would be a meaningless operation. On
to say that 20 degrees is twice as warm as 10 degrees, the other hand, population and area (both on a
because zero degrees on the Celsius scale is not a ratio scale) can be divided to give population den-
true zero. There is still a temperature when the sity, or elevation at one point (interval scale) may
thermometer reads zero! Negative numbers are also be subtracted from elevation at another point to
possible on an interval scale. give difference in elevation.
On a ratio scale measurements can have an abso-
lute or real zero, and the difference between the
numbers is significant. Snow depth is an example. OO OTHER SOURCES OF SPATIAL DATA
It is impossible to have a negative value for snow
depth. Something is also known about relationships So far in this chapter we have considered the
between data, for example a snow pack that is 3 m characteristics of spatial data and their thematic
deep is twice as deep as one that is 1.5 m deep. dimension. To do this we have drawn heavily on
One of the problems with the scales of meas- the map metaphor. However, there are a number of
urement used for the collection of attribute data other sources of spatial data, including census and
is that the distinction between the various scales survey data, aerial photographs, satellite images and
is not always obvious. Many data used in GIS are global positioning systems, which have additional
nominal or ordinal. It is important to take care special characteristics. These are reviewed below. In
when using these data in an analytical context. addition GIS applications may draw on other busi-
If the scale of measurement is not known, or the ness-specific data sets. Box 2.6 provides brief details
GIS user is unaware of what scale has been used of a GIS application which requires the integration
or what operations can be carried out on data of a wide range of data sets, including video.
54 Chapter 2 Spatial data
STUDY
CASE
with GIS
BOX2.6
public questions as to whether locations of crime had tripled to 660 with additional camera locations being
CCTV coverage, let alone find the actual footage of provided by partner organizations and through the
incidents taking place. Police Community Support Officers approaching local
The MAPS GIS was developed to record the firms with CCTV cameras on their premises.
location of the Council’s CCTV Cameras and their Eighteen partner organizations now have access
‘field of view’ (the areas covered by the camera). to the information provided by MAPS with over
Camera locations were surveyed, and detailed maps eighty users trained to use the web-based system.
produced to show the areas visible from each camera. Additional spatial data, to assist in the prevention
At the start of the project only the location of the and detection of crime, has now been added to MAPS.
240 CCTV cameras, owned by Salford City Council, This includes information on; the location of licensed
were included in MAPS. However, as soon as MAPS premises, property ownership, bus stops, bus routes,
was launched and access provided to partner petrol stations, dispensing chemists, second hand
organisations including; Greater Manchester Police, goods shops; and scrap yards.
the Transport Police, the University of Salford, the Since the rollout of MAPS the system has been
Manchester Fire Service; and the Primary Care Trust, widely used in support of crime prevention and
it was clear that the potential of the system was far detection including the planning of surveillance
greater than initially thought. Within three months operations. The use of MAPS is now an integral part
the number of cameras within the system had almost of the daily workflow for volume crime officers who
can now quickly check
whether the crime
scenes have CCTV
coverage without needing
to visit the location of
the camera. Perhaps
the greatest benefit of
MAPS is that Partners
can now work together
to track ‘rolling’ crimes
which move between the
cameras of the different
agencies, something
which would have been
impossible before MAPS
was established.
FURTHER
INFORMATION
www.cadcorp.com/pdf/
PA-CCTV_Image.pdf
Figure2.18b The MAPS system Sources: Peter Robson,
Sources: MAPS interface courtesy of Computer Aided Development Corporation; Map data Network Rail and Paul
reproduced by permission of Ordinance Survey Coward, Salford GIS Ltd
Administrative, survey and census data social security benefits which are collected
routinely or on a one-off basis. They may
Administrative data, surveys and census data are provide useful data on topics which were not the
collections of related information. They may be spa- primary reason for data collection. For example,
tial in character if each item in the collection has a data on benefits claimants can indicate levels of
spatial reference that allows its location on the sur- unemployment or sickness absence from work.
face of the Earth to be identified. OO Survey data: a national coverage survey may be
OO Administrative data: these include data on births, carried out on a sample of the population and
deaths and marriages or details of those claiming ask respondents a series of detailed questions
56 Chapter 2 Spatial data
focusing on a particular topic: for example, aggregated to the same geographies as are used for
health, occupation or housing. Surveys are the census (see below and Box 2.7). Nationally rep-
also conducted for smaller geographical areas resentative surveys may not include local-level
or for specific business or research purposes: geographies, but these sources often allow regional
for example surveys of shopping habits or food results to be mapped and underpin complex
consumption. statistical models which provide explanations to com-
OO Census data: a modern census collects plement detailed mapped distributions. Compared
demographic and household information about with censuses and administrative data, surveys tend
an entire population, unlike ancient censuses to have only large-area regional geographies but con-
which were taken for taxation purposes or to tain comprehensive personal information.
determine the number of males who could Population census data normally have some
be conscripted into the army. Knowing the element of spatial referencing. Most population
population total may be useful in its own right, censuses use a hierarchical series of spatial units to
but details of how people are distributed across a publish data. Two sets of spatial units are associated
country is of additional interest. with the UK census with the smallest being the enu-
meration districts (EDs) used for data collection and
The utility of administrative, survey and census Output Areas (OAs) used for reporting data (see Box
data in GIS applications depends on the detail of 2.7). In the UK census data are not usually released
spatial referencing in the data source. Table 2.2 in map form, but in tables for the spatial areas you
provides some examples of spatially referenced request. However, since a spatial reference is attached
administrative and survey data collected at a national which links the data to the areal units of collection,
level. Administrative data are often released and the data are immediately useful for spatial studies.
TABLE 2.2 Examples of spatially referenced administrative and survey data collected in the UK
Census data describe the socio-demographic state of Ireland on a particular day once every 10 years
a whole country from national level down to local area. (Openshaw, 1995). The data collected are used by
No other data source provides such comprehensive national government for the allocation of billions of
spatial coverage of the human landscape. For pounds of public expenditure and policy analysis.
example, the UK Census of Population provides a Census data are also very valuable commercially since
‘snapshot’ of the distribution, size, structure and they are essential ingredients in marketing
character of the people of Great Britain and Northern analysis and retail modelling.
Other sources of spatial data 57
BOX 2.7
The UK Census of Population is a relatively simple OO Small area statistical geographies: Output Areas,
questionnaire survey administered across Great Super Output Areas.
Britain and Northern Ireland once every 10 years, and
Most of the UK’s census data are released in tables
an estimated 25 million households were involved
with pre-defined cross-tabulations of variables. Each
in the 2011 census. The census is administered
record in a table will correspond to a particular area
separately in England and Wales, Scotland and
and will include an alphanumeric code which can be
Northern Ireland, but most of the questions asked
used to ‘join’ the data to GIS vector polygons so that
and statistics published are common to all these
choropleth maps of outputs can be displayed.
countries, with just minor variations for country
specific items. In the USA the Census has run since 1790 and has
The UK Census covers a wide range of subjects an additional role. The United States Constitution
that describe the characteristics of the population requires a census every 10 years to determine how
of Britain and Northern Ireland. These include many seats each state will have in the US House of
demography, households, families, housing, Representatives. Census data are used to assist
ethnicity, birthplace, migration, illness, economic the apportionment (i.e. the distribution) of the 435
status, occupation, industry, workplace, car seats in the House of Representatives amongst the
ownership and mode of transport to work states. Census counts are also used to assist the
Most censuses use a hierarchical series of redistricting (the redrawing of political districts)
spatial units to publish data, often based on the within each state after apportionment.
administrative geography existing at the time of the The most recent US Census, undertaken on 1 April
census. In the UK the smallest geographical units 2010, was a short-form-only census asking questions
of data dissemination are Output Areas (OAs), a about sex, age, race and Hispanic origin. A longer
geography designed specially for the census to provide version, which was formerly received by one in every
maximum geographic detail whilst still maintaining six households and covered topics such as education,
respondent anonymity (Martin, 1995). For the 2001 employment, ancestry, disability and type of heating
census OAs in England and Wales were based on fuel, has been replaced by the American Community
postcodes and fitted within the boundaries of 2003 Census (ACS) which will be conducted on an annual
statistical wards and parishes. If a postcode crossed basis to provide more frequent demographic,
an electoral ward/division (or parish) boundary, it was housing, social, and economic data. Online mapping
split between two or more OAs. The recommended tools were used to provide daily updates of the 2010
size for an OA was 125 households to ensure that Census mail participation rates for local areas,
smaller wards and parishes are incorporated into and the National Historical Geographic Information
larger OAs and anonymity maintained. For the 2001 System (www.nhgis.org) provides free access
census data just over 175,000 OAs covered England to aggregated census data and GIS compatible
and Wales (www.statistics.gov.uk/geography/census_ boundary files.
geog.asp). Census data are also made available for a Other countries which have conducted censuses of
variety of other geographical areas and spatial scales. population and housing since 2000 include Australia,
These include: New Zealand, Canada, Hong Kong, India, Singapore
and Sri Lanka.
OO Administrative geographies: Nation, region local
government districts, electoral wards. Sources: Paul Norman, School of Geography, University of
OO Electoral areas: European constituencies, parlia- Leeds; www.geog.leeds.ac.uk/projects/census;
ESRC Census Development Programme www.census.ac.uk/;
mentary constituencies, electoral wards. UK Office of National Statistics www.ons.gov.uk/census/
OO Postal geographies: postal areas, postal districts, index.html; Population Reference Bureau http://www.prb.
postal sectors. org; and U.S. Census Bureau, www.census.gov
contains a mass of data and it is necessary to carry information on land use, vegetation type, moisture
out some form of interpretation to make effective or heat levels or other aspects of the landscape from
use of the information portrayed. Aerial photo- the photograph. Aerial photographs are particu-
graphs may be used in GIS as a background for other larly useful for monitoring change, since repeated
data, to give those data spatial context and to aid photographs of the same area are relatively inexpen-
interpretation. Alternatively, the user may abstract sive. For example, Gunn et al. (1994) have monitored
(a) Infrared vertical aerial photograph (b) Vertical colour aerial photographs showing
archaeological remains
(c) Oblique colour aerial photograph (d) Vertical black and white aerial photograph
changes in land use, particularly peat extraction, in vertical features such as mountains, buildings and
County Fermanagh, Northern Ireland, from a time trees appear to lean away from the centre of the
series of photographs. Interpretation of a sequence image or the ‘nadir’ (that point that is vertically
of photographs may allow the dating of events such beneath the camera). This is particularly noticeable
as major floods which cause changes to the land- in the latest generation of high-resolution digital
scape. Curran (1989) identifies six characteristics aerial photography and especially in urban areas
of aerial photographs that make them of immense where the ‘lean’ of tall buildings can obscure the
value as a data source for GIS: streets below. Second, factors that may influence
interpretation need to be considered. These include
OO wide availability;
time of day and time of year. On photographs taken
OO low cost (compared with other remotely sensed in winter, long shadows may assist the identification
images); of tall buildings and trees, but may obscure other
OO wide area views; features on the image. Conversely, in summer,
OO time-freezing ability; when trees are in full leaf, features that may be vis-
ible from the air in winter will be obscured.
OO high spectral and spatial resolution; and
OO three-dimensional perspective.
Additionally, aerial photographs can be used Height above
ground
to obtain data not available from other second-
ary sources, such as the location and extent of new
housing estates, or the extent of forest fires. One 1500 m
characteristic of aerial photographs that constitutes
a possible disadvantage is the fact that they do not
provide spatially referenced data. Spatial referencing
has to be added to features on the image by refer-
ence to other sources such as paper maps. Several
different types of aerial photographs are available,
from simple black and white, which may be used for 1000 m
a wide variety of purposes, to colour and thermal
infrared for heat identification.
The angle at which the photograph was taken
is important. A photograph is referred to as verti-
cal if taken directly below the aeroplane, and oblique
if taken at an angle. Oblique photographs generally
cover larger areas and are cheaper than vertical pho- 500 m
Aerial photography has been successfully used drought periods may reveal the detail of subsurface
in archaeological surveys. Subtle undulations, indi- features as alternating patterns of green and dry
cating the presence of archaeological features such vegetation caused by variations in soil depth.
as the foundations of old buildings and the outlines Aerial photographs represent a versatile, rela-
of ancient roads and field systems just below the tively inexpensive and detailed data source for
surface, can stand out in sharp contrast on winter many GIS applications. For example, local govern-
or late evening images. Photographs taken during ment bodies may organize aerial coverage of their
districts to monitor changes in the extent of quar- LiDAR (Light Detection and Ranging) is a
rying or building development. At a larger scale, remote sensing system that uses aircraft-mounted
photographs can be used to provide data on drain- lasers to collect topographic data from low
age or vegetation conditions within individual fields altitude. The lasers are capable of recording ele-
or parcels that could not be obtained from conven- vation measurements with a vertical precision
tional topographic maps (Curran, 1989). of 15 cm at spatial resolutions of around 2 m.
Measurements are spatially referenced using high-
Satellite images precision GPS. The technology creates a highly
detailed digital elevation model (DEM) that closely
Satellite images are collected by sensors on board a matches every undulation in the landscape. Even
satellite and then relayed to Earth as a series of elec- change in ground-surface elevation detail caused
tronic signals, which are processed by computer to by buildings and trees can be detected. This
produce an image. These data can be processed in a makes LiDAR extremely useful for large-scale map-
variety of ways, each giving a different digital version ping and engineering applications. The images
of the image. shown in Figure 2.22 are derived from LiDAR data.
There are large numbers of satellites orbiting the
Box 2.8 outlines the use of LiDAR data in a hydro-
Earth continuously, collecting data and return-
logical application.
ing them to ground stations all over the world.
Scanned images are stored as a collection of
Some satellites are stationary with respect to the
pixels, which have a value representing the amount
Earth (geostationary), for example Meteosat, which
produces images centred over Africa along the of radiation received by the sensor from that portion
Greenwich meridian (Curran, 1989). Others orbit of the Earth’s surface (Burrough, 1986). The size of
the Earth to provide full coverage over a period of a the pixels gives a measure of the resolution of the
few days. Some of the well-known satellites, Landsat image. The smaller the pixels the higher the reso-
and SPOT, for example, operate in this way. Landsat lution. The Landsat Thematic Mapper collects data
offers repeat coverage of any area on a 16-day cycle for pixels of size 30 m by 30 m. Much greater resolu-
(Mather, 1991). Figure 2.21 shows examples of tion is possible, say 1 m by 1 m, but this has in the
images from earth observation satellites. past been restricted to military use. Recent changes
Most Earth observation satellites use ‘passive’ in US legislation and the availability of Russian
sensors that detect radiation from the sun that is military satellite data have made access to very high-
reflected from the Earth’s surface. Sensors may resolution data easier such as that provided by the
operate across different parts of the electromag- QuickBird and IKONOS satellites. Resolution is an
netic spectrum, not only those portions visible to important spatial characteristic of remotely sensed
the human eye. The multispectral scanner (MSS) data and determines its practical value. A Landsat
on board Landsat simultaneously detects radia- Thematic Mapper image with a pixel size of 30 m by
tion in four different wavebands: near infrared, red, 30 m would be unsuitable for identifying individual
green and blue (Curran, 1989). After processing, the houses but could be used to establish general pat-
images can be used to detect features not readily terns of urban and rural land use. Box 2.9 provides
apparent to the naked eye, such as subtle changes in further discussion of resolution.
moisture content across a field, sediment dispersal For GIS, remotely sensed data offers many advan-
in a lake or heat escaping from roofs in urban areas. tages. First, images are always available in digital
A smaller number of satellites use ‘active’ sensors form, so transfer to a computer is not a problem.
that have their own on-board energy source and so
However, some processing is usually necessary to
do not rely on detecting radiation reflected from the
ensure integration with other data. Processing may
surface of the Earth. Examples are radar-based sen-
be necessary to reduce data volumes, adjust reso-
sors such SAR (Synthetic Aperture Radar). These
have the advantage of being able to explore wave- lution, change pixel shape or alter the projection
lengths not adequately provided for by the Sun, of the data (Burrough, 1986). Second, there is the
such as microwave, and are both able to work at opportunity to process images or use different wave-
night and penetrate cloud layers. bands for the collection of data to highlight features
62 Chapter 2 Spatial data
(a) Vertical urban image showing residential (b) Vertical urban image showing industrial units,
housing and roadway trees, cuttings and embankments
(c) Oblique image of Kverkfjoll volcano, Iceland (d) Oblique image of Odenwinkelkees glacier, Austria
Figure 2.22 LiDAR imagery: vertical urban (a, b); oblique (c, d)
Sources: (a, b) Precision Terrain Surveys (PTS); (c, d): Reproduced with permission of Dr Jonathan Carrivick
of particular interest: for example, water or vegeta- 1989; Maguire, 1989; Mather, 1991). Trotter (1991)
tion. The repeated coverage of the Earth is a further considers the advantages of remotely sensed data
advantage, allowing the monitoring of change at for GIS applications in the area of natural resource
regular intervals, although for some types of management to be:
imagery, cloud cover whilst the satellite passes over- OO low cost relative to other data sources;
head may prevent a useful image being obtained.
OO currency of images;
Finally, the small scale of images provides data
useful for regional studies, and applications have
OO accuracy;
included mapping remote areas, geological surveys, OO completeness of data; and
land use monitoring and many others (see Curran, OO uniform standards across an area of interest.
Other sources of spatial data 63
STUDY
CASE
data solutions to complex
modelling problems
Stuart Lane and Joseph Holden
Land management can influence landscape
response to rainfall events altering the flow of water
in rivers and leading to changes in flood peaks.
High-resolution data can help us understand these
impacts and enable us to develop management
tools that are spatially focused on key areas of
the landscape to provide maximum benefits from
economic investment. This case study outlines
the use of LiDAR DEMs and accurate GPS surveys
to create better connectivity maps to inform
management decisions in an upland catchment in
northern England.
APPLICATIONS
When the soil is saturated from a given point on a
hillslope all the way to the stream channel it can
be said to be hydrologically well connected. In this
instance it is likely that overland flow will move (c) 64 metre
quickly across the land surface into the stream and Figure 2.23 Topographic Index for (a) 2 metre;
there will be a rapid response to any further rainfall (b) 16 metre; (c) 64 metre resolution DEM
64 Chapter 2 Spatial data
BOX 2.8
inputs. It is also likely that any surface pollutants (e.g. D8 algorithm, see Chapter 7), appropriately detailed
from fertilisers or animal droppings) can be washed data sources such as SAR and LiDAR have become
into the river channel network. However, while a given available more recently.
point on a hillslope might be saturated, overland flow Accurate mapping of hillslopes is also useful for
might not be directly connected to the stream network determining the effects of human modifications to the
because there may be places on the same hillslope drainage network (such as from artificial drainage)
where the soil is not fully saturated and water flowing and in planning flood defences. These techniques have
over the surface from upslope can infiltrate into the been used in Upper Wharfedale, England, to model the
soil at that point. Therefore, any surface pollutants impacts of upland land drains. These drains, locally
may be prevented from reaching the stream channel known as ‘grips’, are shallow ditches that were cut
as the flow is ‘disconnected’. Figure 2.24 shows an diagonally across the contour lines into deep peat
example of a hydrological model prediction based on soils during the twentieth century to aid drainage
the Topographic Index using high-resolution LiDAR and improve grazing (Holden et al., 2007). However,
data for a particular time during a heavy rainfall event. these grips have not been shown to improve the land
While a large proportion of the catchment is saturated for grazing or grouse and are more often associated
(green), the blue areas on Figure 2.24 are those with negative environmental impacts. The grips have
where there is direct connectivity with the stream an impact on the saturation by preventing flow from
channel. It is therefore these areas where careful upslope flowing down the hill to the lower parts of
management may be required to reduce grazing or the slope. This means the lower parts of the slope
fertiliser inputs. This sort of approach means that, below the drains are drier than they would otherwise
be (Figure 2.25). This is important because peats are
instead of having a blanket management policy for a
large stores of carbon, but this carbon store is only
whole catchment, we can adopt spatially distributed
maintained when they are kept close to saturation.
management plans allowing some activities to take
Drained peats are known to release more carbon
place on some parts of the landscape and to reduce
into the atmosphere and into local rivers, causing
or ban these activities from other more sensitive parts
the water to be discoloured brown. High-resolution
of the landscape. Using high-resolution data means
mapping of the grips in the landscape allows determi-
that this approach can be done more fairly as it allows
nation of which grips are having the greatest impact.
individual farmers to have mixed management on a
In Figure 2.25 this is shown by the red colouration
field-by-field basis rather than banning one farmer
which indicates all areas where the peat will have a
from applying fertiliser and letting another farmer
smaller length of slope draining into it because of
use it. This information can also help in the definition
the presence of grips that divert the flow away into
of buffer zones to protect water courses from diffuse
the stream channel. Such maps help land managers
pollution (Lane et al., 2004). Although suitable models
for mapping hydrological connectivity using DEMs
have been in existence for some time (for example, the
Figure 2.24 Map of hydrological connectivity in Figure 2.25 Area of catchment with Topographic
Upper Wharfedale, northern England Index affected by grips
Other sources of spatial data 65
BOX 2.8
THEORY
BOX 2.9 Resolution
Resolution is defined as the size of the smallest particular application so that the data collected
recording unit (Laurini and Thompson, 1992) or are immediately comparable. In order to facilitate
the smallest size of feature that can be mapped or subsequent analysis it is necessary to choose the
measured (Burrough, 1986). In the Zdarske Vrchy smallest unit possible when collecting the data.
case study introduced in Chapter 1, the data are These BSUs can then be built up into any other unit
stored as raster layers, the size of each individual by the process of aggregation. In the UK Census of
cell being 30 m × 30 m. In this case the resolution Population the BSU for publicly accessible data is the
of each image is 30 m, since this is the lowest level enumeration district (ED), containing approximately
to which the data can be described. In the case of 500 residents (150 in rural areas). EDs can easily
mapping census variables or other socio-economic be aggregated to form wards. Wards can then be
data collected within administrative boundaries, aggregated into districts, and districts into counties.
the Department of the Environment (1987) refers Disaggregation (the reverse process to aggregation)
to Basic Spatial Unit (BSU) as the smallest spatial to areas smaller than the original BSU is fraught
entity to which data are encoded. The BSUs with difficulties and based on so many assumptions
should be constant across all the data used for a as to pose serious problems for data quality.
Field data sources: surveying and GPS A technique of field data collection which has
found particular favour with GIS users is the use
There are several methods of collecting raw data in
of satellite navigation systems or GPS (Global
the field for direct input into a GIS. These are most
Positioning Systems). These are portable devices
often used when the required data do not exist in
that use signals from GPS satellites to work out the
any other readily available format such as a map
exact location of the user on the Earth’s surface in
or satellite image. Traditional manual-surveying
terms of (x,y,z) co-ordinates using trigonometry
techniques using chains, plane tables, levels and the-
(Figure 2.27). Position fixes are obtained quickly
odolites are examples of direct field measurement,
but the data collected need to be written down on and accurately at the push of a button. The accu-
paper first. Modern digital equivalents of these racy obtainable from GPS receivers ranges from
manual techniques have been adapted so that the 100 m to as little as a few millimetres depending
data collected are stored in digital format ready for on how they are used. Originally designed for real-
direct input into GIS. Examples include total sta- time navigation purposes, most GPS receivers will
tions (high-precision theodolites with electronic store collected co-ordinates and associated attribute
distance metering (EDM) and a data logger) and information in their internal memory so they
hand-held laser range finders (Figure 2.26). can be downloaded directly into a GIS database.
66 Chapter 2 Spatial data
The ability to walk or drive around collecting co- applications. When coupled with high accuracy dif-
ordinate information at sample points in this ferential GPS, these terrestrial laser scanners (TLS)
manner has obvious appeal for those involved in can be used to collect highly detailed terrain data
field data collection for GIS projects. Box 2.10 gives with sub-metre accuracy that can processed to
further details on GPS. create a DEM and ultimately used in GIS. Apart
In recent years a number of companies have from the obvious advantage of accuracy, multiple
developed high-resolution, automated laser versions of these DEMs can be created over a period
range scanners for industrial and engineering of time ranging from hours to years using repeat
Other sources of spatial data 67
THEORY
BOX 2.10 GPS basics
GPS is a set of satellites and control systems compute their location, then the positional error
that allow a specially designed GPS receiver to recorded will be the same for both receivers. A highly
determine its location anywhere on Earth 24 hours accurate positional fix can therefore be obtained for
a day (Barnard, 1992). Two main systems exist: the the roving receiver by subtracting the positional error
American NAVSTAR and the Russian GLONASS. A calculated for the base station.
European system, GALILEO, is also due to come Despite 24-hour global coverage, GPS use can be
into service by 2014. The American system consists hampered by certain factors. These include problems
of 24 satellites orbiting the Earth in high-altitude where the path between the satellite and the receiver
orbits. These satellites have a 12-hour orbit time, is obstructed by buildings, dense tree cover or steep
and pass over control stations so that their orbits can terrain, and in polar regions where favourable satellite
be closely monitored and their positions precisely configurations are not always available. Receivers
identified. Satellites and ground-based receivers capable of using signals from the GPS, GLONASS, and
transmit similarly coded radio signals, so that the shortly, the GALILEO constellations, will, to a certain
time delay between transmission and receipt of the extent, circumvent these problems by making sure
signals gives the distance between the satellite and the receiver is always in view of a minimum number
the receiver. If a receiver can pick up signals from of satellites. GPS signals can also be augmented by
three or four satellites, trigonometry is used to additional signals from geostationary satellites and
calculate the location and height of the receiver. groundstations. This provides the standard GPS system
A GPS user will see a position ‘fix’ displayed on with regional real-time differential corrections for
their receiver. Until May 2000 all readings were suitably enabled receivers giving approximately five-
affected by selective availability (SA), a deliberate fold improvement in the spatial accuracy of position
error added to the signals by the US military. This fixes. The Wide Area Augmentation System (WAAS)
has now been switched off and fixes of far greater provides GPS signal corrections in the USA and similar
accuracy can be obtained. One fix, obtained from a systems are being developed in Asia and Europe.
single receiver, will have an accuracy of about 25 GPS is finding a wide range of applications,
m for 95 per cent of the time. More advanced data varying from navigation (air, sea and land), to
collection methods, including the averaging of fixes geomorphological mapping and urban surveying (see
and the use of two receivers in parallel (differential publications such as GPS Solutions or the Journal of
GPS), can be used to obtain readings down to the Global Positioning Systems for up-to-date examples).
sub-centimetre level. The US military still retain the Developments in portable computing and mobile
ability to switch SA back on – for example, during communications have opened up a whole new area
times of crisis or conflicts such as 9/11 and the Iraq of application for GPS within consumer and personal
war, but the existence of GLONASS and GALILEO electronics. Knowing exactly where you are in
mean this will have a much reduced impact. relation to a GIS database on your portable computer,
Differential GPS techniques require two receivers, or receiving location information via your WAP
one fixed at a known location (the base station) phone, can be of great value and is a highly saleable
and the other at an unknown location (the roving commodity. The use of this location technology as
receiver). If both receivers are set up in exactly a form of GIS ‘output’ is known as Location-based
the same manner and use the same satellites to Services (LBS) and is discussed further in Chapter 5.
functions in GIS software. These conversion func- standards to facilitate exchange of data has been rec-
tions adopt commonly used exchange formats such ognized. Some of the standards in current use are
as DXF and E00. As the range of data sources for GIS listed in Table 2.3. More details of one of these stand-
has increased, the need for widely applicable data ards, BS 7666, are provided in Box 2.11.
British Standard 7666 specifies a nationally Part 3 provides the specification for addresses.
accepted, standard referencing method for land The specification provides a nationally consistent
and property in the UK. The standard was developed means of structuring address-based information.
by a multi-disciplinary working party that included Use of the standard should simplify the exchange and
representatives from, amongst others, local aggregation of address-based and related data.
government, the Ordnance Survey, Her Majesty’s The standard specifies that an address must
Land Registry, the Royal Mail, the Forestry contain sufficient information to ensure uniqueness
Commission and academia. The standard provides within Great Britain. The combination of a primary
a common specification for the key elements of data addressable object name and a secondary
sets of land and property in Great Britain. It assures addressable object name achieves this. An address
the quality of land and property information in terms must also contain the name of at least one or more
of content, accuracy and format. of the street, locality, town and administrative area
BS 7666 has four parts. It includes a specification data, so that it is unique.
for: In addition, a postcode is mandatory for a mailing
OO a street gazetteer (BS 7666 Part 1); address, although a postcode may not exist for non-
OO a land and property gazetteer (BS 7666 Part 2); postal addresses. A postal address is a routing
OO addresses (BS 7666 Part 3); instruction for Royal Mail staff that must contain the
OO a data set for recording Public Rights of Way. minimum information necessary to ensure secure
Other sources of spatial data 69
BOX 2.11
delivery. Its presentation and structure are specified conventions, requiring for example the storage of
in another international standard: ISO 11180. text in upper-case format; no abbreviations except
The specification includes details of how locality for ST (Saint) and KM (Kilometre); and no underlining
name, town name, administrative area and postcode of text. Table 2.4 provides examples of land and
should be specified. There are also detailed text property identifiers acceptable under BS 7666.
The Open Geospatial Consortium (OGC), 2010b). The OGC were responsible for proposing the
formed in 1994, is an international consortium of Geography Markup Language (GML) as a GIS data
almost 400 companies, government agencies and standard. GML and early adopters of this standard
Universities working to advance international are described in Box 2.12.
standards for geospatial interoperability (OGC,
PRACTICE
The Geography Markup Language (GML) is a non- Amongst the organizations adopting GML is
proprietary computer language designed specifically the Ordnance Survey (OS), the national mapping
for the transfer of spatial data over the Internet. agency for the UK. The OS will deliver DNF (Digital
GML is based on XML (eXtensible Markup National Framework) data in GML. DNF is a version
Language), the standard language of the Internet, of the OS’s large-scale topographic database that
and allows the exchange of spatial information and will eventually encompass all types of spatial data
the construction of distributed spatial relationships. and all data scales. In the DNF nearly 230,000 tiles
GML has been proposed by the Open GIS Consortium of large-scale topographic data have been merged
(OGC) as a universal spatial data standard. GML is into a single, seamless topologically structured point,
likely to become very widely used because it is: line and topographic database containing information
on buildings, boundaries, roads, railways, water and
OO Internet friendly;
other topographic features. Each feature in the DNF
OO not tied to any proprietary GIS;
is assigned a unique 16-digit identifier that allows it
OO specifically designed for feature-based spatial data;
to be unambiguously referenced and associated with
OO open to use by anyone; and
other features.
OO compatible with industry-wide IT standards.
By adopting GML, the OS is making the DNF
It is also likely to set the standard for the delivery accessible to more software systems and users
of spatial information content to PDA and WAP than would be possible using any other single data
devices, and so form an important component of standard.
mobile and location-based (LBS) GIS technologies. Sources: Holland (2001); GISNews (2001)
70 Chapter 2 Spatial data
REFLECTION BOX
OO How reliable do you think census data are? Try OO an aerial photograph for an archaeological
to list some of the problems that might be faced application; and
when collecting population census data in your OO a satellite image for an agricultural application.
own country. Give some examples of GIS applica- What sensors have been used for the collection
tions or projects in which census data might be of these images? What are their characteristics?
used. How might they be used in a GIS project?
OO Use the Web to find the following (or think of some OO Think about how GPS could be used in a GIS project
examples of your own): of your own. What data could GPS be used to col-
OO an aerial photograph of Manchester, England; lect? How would the data be collected? What might
OO a satellite image of the Great Wall of China; be the problems of using GPS data?
CONCLUSIONS
In this chapter we have looked at the distinction OO the topological structure used to represent the
between data and information, identified the three relationship between entities.
main dimensions of data (temporal, thematic and
In some data sources, one factor will dominate; in
spatial) and looked in detail at how different spa-
others it will be the interplay of factors that gives the
tial data sources portray the spatial dimension. The
data their character. Appreciating the main charac-
main characteristics of spatial data have been identi-
teristics of spatial data is important because these
fied and a review of how the traditional map-making
characteristics will determine how the data can be
process has shaped these characteristics has been
used in building a GIS model. For example, data
presented. In addition, we have considered a range
collected at different resolutions should only be inte-
of other sources of spatial data. The discussion has
grated and analysed at the resolution of the coarsest
shown that any source of spatial data may be influ-
data set. In the Zdarske Vrchy case study the 30 m by
enced by some, or all, of the following factors:
30 m resolution of the land use map generated from
OO the purpose for which they have been collated; TM satellite data dictated the resolution of the data-
OO the scale at which they have been created; base for analysis.
OO the resolution at which they have been captured; Therefore, GIS models are only as good a repre-
OO the projection which has been used to map them; sentation of the real world as the spatial data used
OO the spatial referencing system used as a to construct them. Understanding the main charac-
locational framework; teristics of spatial data is an important first step in
OO the nature of the spatial entities used to represent evaluating its usefulness for GIS. The next step is to
real-world features; understand how these data can be stored in a form
OO the generality with which these entities have been suitable for use in the computer, as this will also
modelled; and influence the quality of the GIS model.
OO Explain the importance of map projections for cartography from a more conventional viewpoint.
users of GIS. Monmonier’s book How to Lie with Maps (1996) offers
a comprehensive and very readable introduction to
OO Describe the characteristics of three sources of
the potential pitfalls of displaying data in map form.
spatial data.
The discussion is just as applicable to maps on the
OO Using examples, outline the importance of computer screen as those on paper. Subjects such
standards for spatial data. as scale, projections and generalization are covered
in detail. Illiffe and Lott (2008) and Maher (2010)
offer practical texts which include detailed infor-
FURTHER STUDY – ACTIVITIES mation on map projections. A good discussion on
UK spatial referencing can be found in Dale and
OO For a project you are involved in, list your data McLaughlin (1988); DeMers (2008) provides a com-
sources. Review each one and identify any issues parable review for the USA. The Chorley Report
about scale, entity definition, generalization, (Department of the Environment, 1987) provides
projections, spatial referencing and topology that brief details of postcodes and recommendations
you think might be relevant. for use of spatial referencing in the UK. Raper et
al. (1992) discuss the whole issue of UK postcodes
OO Compare large- and small-scale maps of the same
in considerable depth and provide examples of
area. Select a small area and note the differences in
address formats and postcode systems in a number
how common features are represented on the two
of other countries including Austria, Germany, the
maps. How is this controlled by generalization?
Netherlands, Spain, Sweden and the USA. A similar
OO Use a world atlas to compare global and regional discussion on the US ZIP code, though not in the
map projections. same depth, can be found in DeMers (2008).
Comprehensive coverage of the principles and
OO Use the national statistics sites in the list of
applications of remote sensing can be found in
websites below to find an up-to-date population
Curran (1989), Clayton (1995), Gibson and Power
figure for New Zealand and the Netherlands.
(2000) or Campbell (2007). Curran (1989) contains
How is the current population total calculated?
a particularly useful chapter on aerial photography
OO Calculate spatial references for your home or that discusses the characteristics and interpretation
office using latitude and longitude and the local of aerial photographs. A good introduction to GPS
grid co-ordinate system. and its importance for GIS can be found in Kennedy
(1996). Seegar (1999) offers the basic principles of
OO Use the Web to find aerial photographs and
geodesy relevant to GPS. Up-to-date information on
satellite imagery for your home town or an area
GPS can be found in publications such as GPS World.
with which you are familiar. Does the resolution
of the image allow you to discern familiar
Campbell J B (2007) Introduction to Remote Sensing. 3rd
locations or features? Can you notice the effects
edn. Taylor and Francis, London
of edge distortion, time of day/year and so on in
Clayton K (1995) The land from space. In:
your image?
O’Riordan T (ed) Environmental Science for Environmental
OO Use the Web to find and compare satellite images Management. 2nd edn. Longman, London, pp. 198–222
captured by different sensors. Curran P (1989) Principles of Remote Sensing.
Longman, London
Dale P F, McLaughlin J D (1988) Land Information
FURTHER STUDY – READING Management: An Introduction with Special Reference to
Cadastral Problems in Third World Countries. Clarendon
Gatrell (1991) provides a good, thought-provoking Press, Oxford
introduction to the concepts of space and geo- DeMers M N (2008) Fundamentals of Geographic
graphical data that is a good starting point for Information Systems. 4th edn. Wiley, New York
anyone coming to GIS from a non-geographical Department of the Environment (1987) Handling
background. Robinson et al. (1995) and Keates (1982) Geographic Information. Report of the Committee of
provide comprehensive coverage of the subject of Enquiry chaired by Lord Chorley. HMSO, London
72 Chapter 2 Spatial data