Geographic Information Management in Local Government - Chapter 4

CHAPTER 4 Spatial Data KEY QUESTIONS AND ISSUES • What are the main characteristics of spatial data? • What are the main types and sources of spatial data? • What is a data model and how is spatial data modeled? • What methods of data capture are available? • What types of databases are used in GIM and why are they so important? • Why is data quality important and how do we achieve it? • What analyses are typically carried out on spatial data? • How do models of spatial processes help decision making? • What are the main forms of GIS output? 4.1 WHAT ARE THE MAIN CHARACTERISTICS OF SPATIAL DATA? GIS are simplified computer representations of reality. The data they use are typically observations and measurements made from monitoring and recording the world around us. However, capturing the appropriate data can be a daunting and time-consuming task. Although there are many sources, there are basically only two categories: primary data, collected through first-hand observation, and secondary data, collected by another individual or organization. All data typically have three dimensions relating to their location (where they are), their attributes (what they are), and the date when they were collected. GIM places the greatest emphasis on using the locational or spatial element for trans-forming data into information, thereby giving it meaning. As we have seen already, the traditional way of storing, analyzing, and presenting spatial data is the map. Cartographic methods are centuries old, and there are many similarities between their approach and the theoretical framework for GIS. Hence there is a great deal to learn from the cartographer’s approach, not least that the purpose of the map ©2004 by CRC Press LLC decides the features to select and defines the amount of generalization, the spatial referencing system, and the method of representing of the data. During the mapping process the cartographer must: • Establish the purpose the map is to serve • Define the scale at which the map is to be produced • Select the features (spatial entities) from the real world that must be portrayed on the map • Choose a method for the representation of these features • Generalize these features for representation in two dimensions • Adopt a map projection for placing these features onto a flat piece of paper • Apply a spatial referencing system to locate these features relative to one another • Annotate the map with keys, legends, and text to facilitate use of the map (Hey- wood et al., 1998, after Robinson et al., 1995). The scale of the map is determined by the purpose or purposes to be served and represents the ratio of a distance on the map to the corresponding distance on the ground. That is, at a scale of 1:2500, a line of 1 cm on the map represents a line of 2500 cm or 25 m on the ground. Local authorities use a wide range of map scales, but the most common are 1:1250, 12,500, and 1:10,000 for large-scale mapping and 1:50,000 for small-scale mapping. Fundamentally, maps use three basic symbol types to represent real-world fea-tures: points, lines, and areas. The same three basic spatial entities are used in any GIS. Points are used to represent features that are too small to be shown as areas, e.g., lamp posts, manhole covers, and street furniture on large-scale maps. Lines, which are simply an ordered set or string of points, are used for linear features such as roads, pipelines, administrative boundaries, and river networks. Networks are sometimes treated as a separate data type but are really just an extension of the line type. Finally, areas are represented by a closed set of lines and are used to define features such as buildings, fields, and administrative areas. Area entities are frequently referred to as polygons. As with line features, some of these polygons exist on the ground, e.g., buildings, and some are imaginary, e.g., census enumeration districts. Three-dimensional areas are treated as surfaces, which can be used to represent topography or nontopographic features such as pollution levels and population den-sities. Sometimes, surfaces as well as networks are considered as separate entity types. Each spatial entity may have more than one attribute associated with it. Attributes are the nongraphical characteristics of the entity. For example, they can describe the type of building defined by a polygon — a house, a school, or an office — or the class of road represented by two parallel lines. These attributes allow certain GIS operations to be performed, e.g., “where are all the primary schools within a par-ticular ward?” or “which is the shortest route from A to B?” However, in order to answer such questions, the geometric relationships between the spatial entities must be understood. In GIM, topology is the term used to describe the geometric characteristics of spatial entities or objects. In relation to spatial data, topology comprises three elements: adjacency, containment, and connectivity. Objects can be described as adjacent when they share a common boundary, whereas containment describes one ©2004 by CRC Press LLC feature contained within another, e.g., a house within a garden. On the other hand, connectivity is the geometric property used to describe linkages among line features, e.g., roads connected to form a bus network (Heywood et al., 1998). In order to carry out analyses of the basic spatial entities, it is necessary to treat the spherical Earth as a flat two-dimensional surface (a sheet of paper) by using a suitable map projection. This transformation is achieved by approximating the true shape of Earth, thereby introducing errors into the spatial data. These will vary depending upon the projection method chosen from the wide range available. Some will distort distances, others direction, while others will preserve shape but distort areas. Users need to know which map projections are being used, particularly if they wish to combine data from different sources. Otherwise, features that exist at the same location on the ground may appear to lie at different geographic positions when viewed on the map or computer screen. For mapping small areas of the globe, especially those like the U.K. that have only a small extent of latitude, the Transverse Mercator projection is often used. It has the advantage of maintaining scale, shape, area, and bearings over small areas and was chosen as the basis of the OS’s National Grid system. Spatial referencing is used to locate a feature on Earth’s surface or on a map. Several methods of spatial referencing exist, all of which can be grouped into three categories: geographic coordinate systems (latitude and longitude), rectangular coor-dinate systems (e.g., the OS’s National Grid system), and noncoordinate systems (e.g., the U.K. postcode system). Most spatial referencing systems have problems associated with them. Heywood et al. (1998) list three examples: spatial entities may be mobile — e.g., animals, cars, and people can be located only at a particular time; spatial entities may change — e.g., road improvements occur, policy areas are redefined; and the same object may be referenced in different ways — e.g., a building may be represented as both a point and a polygon on maps of different scales. Despite these problems, the ability to link, or “glue” together, disparate datasets using spatial referencing is vital to the management of geographic information, as the following section will show. 4.2 WHAT ARE THE MAIN TYPES AND SOURCES OF SPATIAL DATA? Data about local authorities’ areas and activities are produced continuously. Many of their everyday activities produce spatial data automatically, some of which is stored digitally in databases but much of which still remains in analogue form in files, ledgers, and photographs. In addition, local authorities use data from various central government departments as well as aerial photography, satellite imagery, and field surveys. Not only are there now an abundance of spatial datasets available both to local authorities and their citizens, there are a wide variety of sources providing data that differ widely in content, currency, and role. Writing in the AGI Source Book for GIS, 1997, Hugh Buchanan usefully categorized this data into three varieties (see Box 4.1): ©2004 by CRC Press LLC • Application data that gives information of importance for answering a particular question • Parcel data that describes abstract units of area that the world is divided up into • Topographic data that tells you about the physical surroundings Buchanan goes on to explain that, for many purposes, some data of each sort is required: Users often already have some application data, and wish to relate it to some other application data, together providing the facts that are of most direct interest. These facts have to be attached or glued to each other, or alternatively to the real world. This is done by using some parcel data that relates the spatial content of some application data to the spatial content of other application data (for example postcodes to census areas). Additionally, it is usually useful to relate these parcels to the real world in the form of some topographic data, so that the data can be vizualized or inspected. Box 4.1 Data Varieties Application Data (Interest) The term application data covers many things, such as socio-economic, geological or property data. A user will often have their own data (such as customer records), and is often also interested in adding value to their own information by relating it to other sets of data. One major source of data about population is the (decennial) census carried out by the Office for National Statistics in England and Wales, the General Register Office in Scotland and the Census Office for Northern Ireland. In addition to the factual bones of the census, much socio-economic flesh is added by surveys of population and behavior. For other application areas, the required data will be different, such as geological, hydrological and land use data. Parcel Data (Glue) Socio-economic application data is often spatially described using a street address, a postcode, an electoral ward or a census enumeration area, but very rarely by a National Grid (map) co-ordinate. Land-related information is very often described by a National Grid co-ordinate, but may be described by an administrative area, such as a county. There are a variety of data products that relate one set of parcels to another and individual parcel sets to the National Grid. Topographic Data (Real World) Topographic data corresponds to the traditional published map, but is now available in a variety of different forms. The first of these is the vector map, where the co-ordinates of each line, point and piece of text are included. A common alternative to vector maps are raster maps.The raster consist of a fine grid of cells, each of which carries a colour value. By displaying the raster, the user can recreate the type of visual appearance that a paper map would have had. In recent years, a third form of topographic data has become increasingly common.This consists of photography and satellite imagery. In computer readable form, these types of data are raster. They are created from cameras and other sensors carried by aircraft and satellites, and are very good at retaining the overall visual impression of the surface, since (for example) the nature of the ground cover can be seen on the image. The largest supplier of topographic data in the U.K. is the Ordnance Survey, who have a wide range of data products. Other suppliers of such data are land survey firms who will create data to order, and other data publishers such as Bartholomews and the AA. Source: Extracted from Buchanan, H. (1997) Spatial Data: A Guide, in D.R. Green and D. Rix (Eds.), AGI Source Book for Geographic Information Systems 1997, London: AGI. ©2004 by CRC Press LLC In local government, the OS’s digital topographic database provides the bedrock for GIS in the traditional map-using services like planning, highways, and estates. However, for many users aerial photographs are easier to interpret as they provide a real picture of the world at a known point in time. Raw photographs are not as accurate as maps as they contain scale distortions, especially at their edges, and make buildings appear to fall away from the center. This problem, together with errors due to changes in ground relief, can be resolved by a process known as orthorectification. Increasingly available are off-the-shelf products containing aerial photographs that have been scanned, orthorectified, and stored as digital databases. The sources for this data include: • Geoinformation Group, a U.K. company formed from a management buyout of Cities Revealed products, providing 25-cm digital databases corrected to OS mapping focusing on cities or counties in high-demand areas • Getmapping.com (formerly Millennium Mapping Company) originally formed to create a millennium archive of the U.K. at 1/10,000 scale • U.K. Perspective, a joint venture between NRSC and Simmons Aerofilms, pro-viding another millennium archive with the ability to create digital orthophoto-graphs on demand For practical purposes, digital imagery is mainly used in a compressed format due to large storage requirements. For example, with the normal 25-cm resolution, a 1-km2 tile takes approximately 45 MB of disk space. However, commercially available software such as Mr SID enable images to be reduced to about 2 MB without significant loss of clarity, making imagery considerably more manageable (Denniss, 2000). High-resolution imagery is also available from satellites and new digital airborne imagers. This is invaluable not only in the construction of an accurate and compre-hensive GIS database but also in maintaining the database at a reasonable cost. New sources of satellite information that are more affordable and have much improved ground resolutions are becoming available. Often the frequencies used to capture the data are such that they can penetrate cloud cover and the data can be quickly processed to order. Land, property, and highways services often describe their data by National Grid coordinates, but most application data in local government is glued together by an address or the postcode system. As a result, local authorities have found both the OS’s ADDRESS-POINT and the Royal Mail Postcode Address File (PAF) invaluable as a means of linking Great Britain’s 25 million addresses and the unit postcodes to National Grid references. The Gridlink initiative launched at the GIS 2000 con-ference by the OS, the Office for National Statistics (ONS), the Royal Mail, and the General Register Office for Scotland (GROS) has further harmonized and improved the consistency and compatibility of postcode grid referencing. However, it still does not provide a single national infrastructure of definitive addresses and related prop-erty information and mapping. Therefore, in September 2002, four government ©2004 by CRC Press LLC ... - tailieumienphi.vn