Scribd Logo
Upload

Welcome to Scribd!

  • Gis

    Uploaded by

    Wonder Lycan
    6 views13 pages

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Report this Document

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content
    6 views13 pages

    Gis

    Uploaded by

    Wonder Lycan

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content
    of 13
    GEOGRAPHICAL INFORMATION SYSTEMS IN ARCHAEOLOGY (7. —— PAE James Conolly OLS Department of Anthropology, Tren Unversity & Deseauae of Archoceiings, maverscy College London Mark Lake Insninate of Archacologs. University College Lomdem oe CAMBRIDGE UNIVERSITY PRESS —_—_—_ oS y PRESS Jbourne, Madrid, Cape Town, Sin; APO. SS0 Pata. Day fe UNIVERSIT! awaeid6 pane York, Me! ‘Cambridge, New ambnidge University Press a inburgh Building, Cambridge CB2 SRU, UK published in the United States of America by Cambridge University Press, New ¥ ork www.cambridge.ore rreratio on thistle: wor cambridge.0rg/9780521797443 © 1. Conolly and M. Lake 2006 ‘Tis publication isin copyright. Subject to statutory exception and w the provisions of relevant collective licensing agreements, fo repeeduction of any part may take place without the written permission of Cambridge University Press. First published 2006 ‘Third printing 2008 Printed in the United Kingdom at the University Press, Cambridge A.catalogue recon for this publication is available from the British Library ISBN 978-0-521-79330-8 hardback ISBN 978-0-521-79744-3 paperback Cambridge Uni URLs ant Prss as no responsibly forthe persistence or acurey and does not guarantee dex arty internet websites referred to in this publication, accu orate, at MY content on such websts is, oF will emai, First principles 2.1 Introduction a ‘The power of GIS, as with other computer programs, can be deceptive: visually impressive but ultimately meaningless results can appear unassailable because of the sophisticated technologies used to produce them (Eiteljorg 2000). The famil- iar adage ‘garbage in, garbage out’ is particularly applicable to GIS, and one of our primary aims throughout this book is to provide guidance on how to use this technology in ways to strengthen and extend our understanding of the human past, rather than to obfuscate it. In this chapter we start by providing an overview of the “first principles’ of GIS: the software and hardware requirements, geodetic and car- tographic principles, and GIS data models. These provide the conceptual building blocks that are essential for understanding what GIS is, how it works, and what its strengths and limitations are. Although some of these “first principles’ may be familiar to readers who are experienced in cartography and computer graphics, we nevertheless provide a thorough review of each as they yield the foundation on which we build in later chapters. 2.2 The basics 2.2.1 GIS functionality ‘What does a GIS do? Simply providing a definition of GIS and referring to its abilities to capture and manipulate spatial data doesn’t provide much insight into its functionality. More informative is to break some of the basic tasks of a GIS into five groups: data acquisition, spatial data management, database management, data visualisation and spatial analysis. Some of the routine tasks performed under these headings are outlined in Fig. 2.1 and described in Box 2.1. While each of these tasks are important in themselves, be considered as both an integrated and as an integrating vides a suite of tools that help people interact and understan It is important to stress that although the origins of GIS digital cartography, GIS is not just about ‘maps’ nor is it the digital manipulation of the sorts of information and mi depicted on maps (cf. Longley et al. 1999). The use of G! Contribution to make in terms of understanding spatial an tionships between natural and anthropogenic phenomena ( itis increasingly common to make the distinction betwee! to process geospatial data (GIS), above all GIS should technology that pro- \d spatial information. are strongly rooted in necessarily only about ethods that are usually IS has a much broader \d even space-time rela- (Couclelis 1999). Indeed, n the software tools used and a geographic information science (‘GlSc") ur art principles “4 spatially related tO ‘al ‘Archaeological sito 4 | phenomenon rela Temporally Aspatial attributes related to 232 Teese spatial and asatal characteristics of archcol Fe 2 rupli sftaly elated to other adcent burials, xp aoe pias of ht gnder or age gTOUP. temporally related to ether buna cea! steep poston ofthe Dy, wl have ae an mrpology, ad possess any numberof aspatial tits cy prot and glad qunttave characteristics of he skeleton) fee Wi 28 ical daa «alamo abut spall asocation and interaction that describes spatial nly shi such sath a eresses path 6’, “from settlement p one can see setematg:» ‘Sets l0Dmeatof sh water’ AS we discussed in Chapter 1, some peste, asecaion ae refered 10 as topological, such as when We talk about pat eng comecions. Tpoogcal relationships are also described as orientation indeed Because ely he connective relationships between objects are importa and x c- arextaion sata loation. Orietation-dependent or directional relatoni x thse tat 3 relational directions, such as above, below, in front, behind, ot carina rection ea west south, north Jones 1997. p25). + Temporal relationships that desribe the date and/or associated features in wan tem ie ‘contemporary with, later than, ‘earlier than’ ete. Temporal elesin canbe imporant for ensuring that particular types of analysis, suchas ste gates, usderaken oly on contemporaneous sites. * One or more aspatial attributes that describe the nature of the object. This me «consist ofa biography of a site or object, information about the colour and raw met! ‘of an object, the time of day that a field was fieldwalked, the shape of the cross-#ein ‘of a feature, or the estimated age of a burial. tener ssciteapatial data with spatial objects means that #P0 ipa age spatial characteristics of non-spatial data. For example, & lg ann tat records their spatia provenance and aspatal Possible tw explore i. eight, size, shape, raw material, reduction stab ee cae ‘elationship between their location and their othe © : isan extremely i ‘ti as found applica in ay ly important ability of GIS that has foun " Measofachieological esearch, 2.2 The basles 45 2.2.3 Components of a GIS AAGIS iva computer-dependent technology. In addition o the computer ivelf, there fare a number of other important components to a GIS. The most important ones } : ’ theep:/ 0 ‘Software In order wo qualify aa GIS the wftware must have: i)» spatial datahave that ‘stores and manages spatial ubjecty (i) some mechanism of linking atritmte date to these spatial objects, either as an internal functiom of the GIS package, o by previd- ing functions that enable access to external databave systems; (i) a “gexmprocessing ‘engine’, which permits the manipulation and analysis of the geospatial information stored in the spatial and attribute databases. Nome of the many GS packages cur rently available perform all take equally well The choice of software consequently ‘needs to be made with respect to several factors, including the tasks its needed for, ‘shat operating system it has to run under (eg. UNIX or UNIXtike systems such as MacOS X, Linus, Ini, or Solaris; orcme of the version of Microsoft Windows) and the size of the budget for software, hardware and training cost, A large number of packages are available - too many for us to attempt to review ~ each with their own srengths and weaknesses in terms of ease of use and the range of analytical tools they offer. An afternoon's research on the web will provide a reasonable grounding in the range of software options. If cost isa primary concer, itis worth knowing that one ‘of the more powerful GIS packages, GRASS GIS," is available fee of charge under an open-source licence, Excellent comprehensive commercial GIS packages include Idris? the ArcGIS suite of programs. and Maplnfo.‘ and all may offer discounts for educational users. Hardware In addition to the computer that runs the sofware, which could range from ' small palmtop computer toa large institutional mainframe there are several other hardware components that are essential to making a GIS work. These can be divided ino two groups. The first consists of input devices, which might be limited to the keyboard and mouse supplied withthe computer, but could extend todigitising tablets, flatbed and roll scanners, digital surveying equipment such as global positioning system (GPS) devices and Total Stations, or geophysical sensors, Chapter S discusses, the various methods for acquiring digital data in some detail. The second group consists of the output devices needed for viewing and sharing information. Acomputer ‘monitor isthe basic piece of display hardware but, with the obvious exception of the WWW, itis nota very convenient device for distributing information to other people. ‘Some type of printer, from standard leer devices to larger colour plotters, is needed for producing the maps, graphs and tables that GIS routinely produces. We review ‘map production and spatial data communication in Chapter 12. People GIS operators are the most crucial part ofthe system as they are responsible for the design and analysis of spatial datasets. A GIS is never a fully objective process ~ data and questions ean rarely be simply “fd toa GIS and useful results retumed ~s0 itis essential that the specialists responsible for digitising, processing and analysing data are closely integrated with both project design and data collection. Ths i less ‘ofan issue when one researcher is conducting both the project design, data collection ‘and analysis, but in large research projects or commercial archaeological units iis. iteit. Mwe.clarklabs.org jeori.com, — twaw.napinfo.com. xcs mays led eat gage, acter Hees pe gy, Senay cal Way Ci aay gl Fh ms 0 a PGES Fhe yo ARGS aay ess ny and IS fe visualisation, management and analy. ta maps. tis consequently importan yy 0 Fay or digital, simplify the world and presen, a ie i nes Naps ce divided into two basic typos. May oslo ie generinfrmation sou the physical sua Prtoarmade features like roads ers, sete sta varity of diferent formats and scales, cach samen 80, tes enh frist are compiled tases ot te st ad emphsse ography sctements reswiege tS inte UK, te Ornuce Survey produce a variety of atest Pott Sr att pid dia feats) showing levation natural ange aes Gnsgachedngl nd iso sts nd ma sand igs tar ual for arange of differen aplications Te yee ten peal aps rhe USA and mnt counts eo xmas, Cat's Cente for Topographic Information, and Ge stab . em tp pve information bout singe feu of nd, out nforaton abut single subject. When e dearer tata ph pets common depay them on wantin nage telson poof constant numeric value, such as clevaion oct tpt thems), pcipiation (sohyets) or even frequencies ot hata (dl ter ees ae mor likey to be displayed on chop oy che sing or symbols to display average values of information nie nx, chs vegetation, geology or numbers of artefacts collected ina survey un pone 1 Maes ee isis Toenphase te ciferences between traditional paper maps and the dynamic! ine hat GS offers itis worth noting some of the constraints of the former Loney eal. 1958, p. 6). Paper maps differ from GIS because they are: Suc The yranic spacetime interactions between objects cannot easily be dpi! {4 angsinpopalaon and settlement patterns, or environmental change AGP Mat arg of enabling exploration ofthe dynamics of temporal pts caclen Yo Sydney Archacological Computing Lab's TimeMap prs! ® into this form of dynamic mapping. al Maliisesionality cannot be easily depicted on paper Muli a ina Aine ete dimensonal representation of topography ect Ceci gy 9Fdpay avaiable in GIS (e.g, Portugal an S08 TT 2.3 Cartographic principles 7 ss the Bah ino eprsing 8 curved te dino suis, such a he Pinon often snocessinfican dstrtion patel etsurementseblow) ar ones ates oping en een mt ale SEpesonotimse, nyc torts fo xn ve ‘ftw cal oxo) Whe i rma tien rset sora eescatonnseewe moe ost or woking wih Shy etc tame npr mapa pita truant Oreste pyr map ed and enoly ewe somacings ne mop. wheres 4p ap ay te ote connces iarattne cso waionl ps Dia neato on-state ates ft bjtson ai Mivtobccoe! sn fers nfomatncanonly bund by ference tne, IXCiS hassel amazes oven digit eye witreprdsto abut dat {npuelar» GS mrecomprcenhe ata ttre ee of pate 29 ‘My eno da poten mre ul compar with ts pope coupe ‘A further major advantage of a GIS over traditional mapping is that a GIS permits organisation of different components of the same map into different thematic players (and thus often referred to as shematic mapping), which is the basic way t spatial data are organised within a GIS environment. In practice this means at in one GIS digital display many different elements may be combined, each f which can be individually turned on or off, queried, modified, reclassified and ited, Many analytical functions, such as spatial queries, can operate across one more layers depending on the need of the GIS analyst. Map layers, or subsets individual layers, can also be combined to produce new maps at will, providing, tential insight into relationships between elements on different themes. 3.2 Map projection systems and geodetic datums basic property of a map is that it is has a spatial context ~ more properly, geo- eferenced - by implicitly or explicitly referring to positions on the Earth's surface. )oviously with many maps a precise and absolute spatial context is not important, Quick sketch of the route to a friend's house serves a purpose even though it may inaccurate and relative, However, when precision and absolute spatial context portant, then an explicit system of measurement is required. As the Earth is a iplex shape this is not a trivial process and the science of geodesy is concerned With the measurement of the morphology of the Earth’s surface, The shape of the _ Earths best approximated by a flattened sphere, referred to as either an ellipsoid or _Beoid, and positions on it can be defined using polar or geographical coordinates ig. 2.3). Geographical coordinate systems define degrees, minutes and seconds orth or south of the equator as latitude and degrees, minutes and seconds east or est from Greenwich Prime Meridian as longitude, is is an elegant and simple solution for locating positions on the planet, It is 8 suitable for representing the surface of the Earth on a two-dimensional plane, for example, on a paper map or computer screen, The name given to a system used iva priiples 6 Fi23 Pcs The clothe sper inthe , -plane isthe equ ang ini. olin tithe mein Ip aan atbitay pat n the sara othe Ean the engl str lente ad the ange deine oy ee (ate Wats 198 p. 103) | ‘0 display areas ofthe Earth's round surface on a flat map is map project imoles a mathematica ransfomatin ofthe unit of longitude and latitude (ie 4 sees ata ple Essentlly, a fat map of alarge area of the Earth's surface caret alted without some form of projection. When mapping areas atthe tau pote slonl ele the tansfomation from thre to two dimensions | fatonl ad se tt and pail eorinpaniculartypes of measurement AL fat surface cane ana lee the distonion arising from projection toa estbshed projected national and state mapping agencies have imising error within their own boundaries. At very m subregional or local, the surface of the Earth can troletccoreton Neh gsc be etablshed and used without reference 0 tiscanbe asoueof confor Cn tems large sale’ and ‘smal scale’ 2s orgeater(eg 1 35000 ie scale generally refers to scales of 1 : 50000) than $0000 eg. 1: ton90 1s alsa to maps with scales smaller 00, etc; Thurston et al. 2003, p. 37). oped for both global and national !PPort many or all of the common different projections). Projection Projection (for a ‘Mathematical di surface cording to how the sphere is iscussion see Hiffe 2000). Each 4 2.4 Cartognyphic principles Fig. 24. A conical projection with two ties of secancy (left) and one tine of tangery ight) The point(s) of contact are also referred to as standard parallels ‘one of more of the parameters of distance, direction, scale, conforma and area, although each projection family attempts to minimise distortio two parameters at the expense of increasing it in others. In addition to the three projection families, there are four projection groups defined on the basis of how this distortion is managed. Conformal or orthomorphic projections preserve the 90° intersection of lines of latitude and longitude to ensure Correct angle measurements between points, but in so doing distort area measure ments. Equal-area projections preserve area calculations, so that the multiplication of the two edges of rectilinear features represented on a map and globe will be iden- tical (but the properties of shape, angle and scale are then distorted). Projections that maintain distances between one or more pairs of points are described as equi- distant projections. Any given equidistant projection will only apply to measure- ‘ments taken in a certain direction: sinusoidal equidistant projections, for example, ‘enforce the measurements parallel to the equator, but distort measurements parallel to the meridian, True-direction projections maintain the correct angle from any line ‘measured from the centre of the projection to any other point on the map. ‘A projection is defined by the combination of a family and then a projection type. For example, a conical projection can be conceptualised as fitting a cone over ‘one of the polar regions as depicted in Fig. 2.4, which is then cut along a meridian as in Fig. 2.5. ‘The result is a map in which the lines of longitude are straight and convergent, ‘and lines of latitude are concentric ares. The line of tangency on conie projections is referred to as the standard parallel and distortion increases the further one moves away from this line. The amount of distortion can, however, be controlled by altering the spacing of the lines of latitude; if evenly spaced then the projection will be equidistant along the north-south axis (equidistant conic projection); ifeompressed at the northern and southern ends, then the projection becomes equal-area (Albers equal-area conic projection). Azimuthal (or planar) projections represent the Earth's surface on a lat plane using a single point of contact rather than a Tine of tangeney (Fig. 2.6). Azimuthal with on line of tangeney (e Fig. 25 Altes cqsharea conical preston with o nEENCY (ef) and Se etter ‘scone acs. Fe 26 Azin rojein wit int of ota the Non Poe, The res ceone atss ls oflngide, and concent line of aitude. Angle ond = se mares lean ela offense ie ee rr i AH if Z i i a é ts in the east-west better than the standard Mercatee “ays of mapping the globe 2.3 Cartographic principles a1 > sd Fig. 2.7 Cylindrical projection with line of angeney corresponding to the equator ‘and a meridian (dashed line). The resulting map is tothe right, showing the lines of| latitude as parallel Lines, Finally, the Universal Transverse Mercator Projection (UTM) is a twentieth- ‘century modification of the TM Projection that divides the world into 60 vertical zones, each of which are 6° of longitude wide. There is a central meridian in each of these 60 zones that minimises measurement distortion in the east-west to approx imately 1 m in every 2500 m (Robinson et al. 1995; DeMers 1997, pp. 63-64). Each zone is divided into rows of 8° latitude (12° in the northernmost section) which equates to a 100000-m wide grid square. The central meridian is given a false easting value of 500000 m to eliminate the need for negative numbers when specifying east-west coordinates. For the same reason the equator is given a north- ing value of 0 m for measurements in the northern hemisphere, and 10000000 m for measurements in the southern hemisphere. Universal Transverse Mercator co- ordinates are given by first specifying the zone and then the easting (with 6 digits for 1 m precision) and northing (with 7 digits for 1 m precision). The UTM projec- tion is very popular in GIS and related geospatial technologies like remote sensing because of its global application, minimal distortion and metric coordinate system, Most GPSs are able to record locations in UTM coordinates, making it an ideal system for spatial data collection when a local grid system is not available. In addition to the projection system used to make the map. it is also important {to be aware of which mathematical approximation of the shape of the Earth was used for the construction of a map. The Earth is not an exact ellipsoid, since the surface is not smooth and the poles are not equidistant from the equator. Polar ‘coordinates of latitude and longitude are therefore calculated using a mathematical approximation of the Earth's shape and its centre, Several different approximations hhave been calculated. often for a specific region of the planet. Clarke's 1866 cal. ulations formed the basis for the 1927 datum of North America (North American Datum 27, or NAD27). NAD27 is being replaced by sstellite-derived measure. ‘ments of an ellipsoid called NADS3 bat many organisations still use measurements ‘and locations using the earlier geadetic datum, The Geodetic Reference System TT First principles 7% saa Geoetc System 81 (WOES) and European Terrestrial (caso, Word Gener rae ecentrecaculations of the ellipsoid usey case Sse ETRSS) es generally se Whichever ellipsoid calcula Nation arr por example, most national mapping in Great Brin, the 1830 Airy ellipsoid, although the Ordnan, the ETRSE9 ellipsoid for more recent mapping derived fms Refe ze transformation and reprojection oa share a projection system (e.g. UTM) but are based on different ellipsoig (eg. WGS#4 vesus NADZ7) are not compatible, nor are maps that use differe projection systems (eg. Transverse Mercator versus State Plane) but share the sam ellipsoid (eg. NAD). For example, the physical distance between two points th: have identical geographical coordinates, but one based on NAD27 and the othe based on NADS3, canbe as much as 100m apart in the USA. For data from multipi ‘map ourcestobe combined, the maps must share a common projection and geodeti, reference system. If this is not the case the projections and/or reference system mus: be altered through a process called secondary transformation ot reprojection. T compute the ansformation, fairly specific information is required about the existing and desired projection systems. Most GIS packages provide tools to transform maps from one geodetic datum to another, and dedicated software tools are also: available why ‘convert between NAD27 and NADB3 (see, for example, the directory of hrasentels National Geodetic Survey's website® and their online conversion ob ees ut the lipid and projection system used in the construction of aad ey a ly printed in the comer or are contained in an associated metadata 'gtal data (see Chapter 13 for further discussion of metadata elements). 233 National and regional grid systems Many GIS programs basis for regional a ‘Beographic coordinates of latitude and longitude as the ae comered deco ad eens decimal degrees where minutes and secons 100508 ofa degre) While desma eam Ple 30 minutes 30 seconds is equal that arable to manage the conser degree systems can work welll in GIS packages on merc units, such. Gon Patil measurement, a Cartesian system hoice because ofthe advantages ron ilitary) grid system, is often 4 280 dimensionl Cane ees oes for calculating distances and ares foray ‘System is used for mapping, east-west mea tedonthe — rtxis and Called eastings, and north-south (© greaterthan 1:4 os and are called northings ( etic planar coounane 20° ™Ostrational or regional mappine nc planar coordinates, ae Alongside, or in place of, latitude and Used in the global UTM projection, the beter ? wee.208 no, ‘%¥/¢G5-bin/madcon.pe) CC 2.3 Cartographic principles 23 50 y-axis (northings) 00 “x-axis (eastings) 100 Fig. 28 A two-dimensional Canesian coordinate system. Point pis located by reference tots distance from a 0,0 datum in the x- and y-planes (respectively refered to asthe ‘eastngs" and “nothings. 50 3 (67.3) : ee fF 3 2 zg acana) 09 ‘rani castings) 700 Fig. 29 To calculate the linear distance between points p and q (c), Pythagoras’ theorem is used: ¢ = Va? +62. Asa and bare known (a = Xp —Xy = 33, b= yp — Yq = Ih the calculation ise = VIF + 1S = 355. US State Plane system, British National Grid and in most other national grids. National grid systems, such as the US State Plane system, are often better choices for regional mapping projects because the ellipsoid is often selected to maximise spatial accuracy for the specific area covered by that particular system. In parts of the world where national or military grids are unavailable, then UTM is an excellent choice. We must emphasise again our warning from the previous section regarding the inevitable and significant spatial errors that will result from combining data derived from maps with different projections and/or ellipsoids. Metric planar systems have the important and crucial advantage of allowing the easy calculation of distance and area. For example: linear distance measurements can be calculated using Pythagoras’ theorem (Fig. 2.9); polygon areas can be ag Ftp’ px ie hp it SE apeia ang id the geometric ay the total area; anc centre coud to derive Oe aking the mean of the coordi i sass found by taking he Cordinates of | foraltematives, see Jones 1997, p. 66, Burs ial phenoment f GIS originate wi of spail phen sil data? The roots o iginate with the deve How ds 21 rere inthe mle ofthe ast Century. Inthe late 195 ato apa algonts for handling geographic information yy sa linge prices digital arogrephy, at about the same ge oa ad developed o incorporate computer graphics (e.g. Tobler 1959 se esnada Geogopic information System, developed in 1963 to manage natur, aa crss anand ontcone of bese developments and qualifies as the fe GIS vas followed soon afterby the development of other systems that were caps be ofaxtonated mapping (Foresman 1998; Tomlinson 1998). Automated mappin, coed considerable time savings over traditional paper methods by providing faste andere accra facilis forthe management and updating of spatial data, Thes early systems relied on point, line and polygon ‘geographic primitives’, which stip foxm th buldng blocks of modem vector-based GIS. AGIS works by manipulating the digital representations of real world entities However aGiS only hasafinitset of resources with which to replicate the infinite, complex word and, as a consequence, the digital representations used by GIS. an aecesaily schematic and generalised, The representation of elements of sna fe fas daa ode In ars data models tend to be ve sof ality, though as we shall se in later chapters, simple model ‘ay become the building blocks for more complex models that are desi dt ‘nant retonsips betwen diferent emi ht AS we saw in Caper 1 GS seient we ¢atity and continuous field data reality ir ry simpk these two data struct ; Bram te at lures Were virtually mutually dein stow 1 on ther one or the other, forcing uiers and vector data, oa a Today most GIS permit the mixing o! 1 the most ietiplavess Gee eee Progam ™ ; . giving users the f POE Te tevig ee samt’ without necessarily using a differen ata ay past provide son Millerences between the vector and Tedby each fom, "* MPLS of ways that different sors o 2.4 Data models and data structures 25 Points Lines Polygons i} i cee \ Fig. 210 The tee vector geographic primitives" of pints lines and polygons. 2.4.1 The vector data structure ‘A vector is a mathematical term that refers to one or more coordinates used to define an object in Cartesian space. In the vector structure, real-world entities are represented using one of three geometrical primitives: points, lines or polygons. Each primitive is defined using one or more x, y-coordinate pairs called vertices, and are thus described as discrete objects because of their precisely defined locations ‘and boundaries (Fig. 2.10). Vertices that are located at the ends of discrete lines, or ‘at their intersections, are called nodes. For example, points are zero-dimensional objects (for they have no length or breadth) defined by a single coordinate pair, and lines and polylines (often also referred to as arcs or edges) are one-dimensional vectors (having the property of length, but not breadth) defined by two or more coordinate pairs. Polygons, Or areas, are two-dimensional objects defined by three or more coordinate pairs. Three-dimensional objects are referred to as volumes, but despite the fact that CAD systems routinely use three-dimensional vector objects, the three-dimensional ‘vector structure has not yet been widely implemented in GIS. ___ The discrete nature of every vector object means that, in addition to possessing its own unique spatial location and morphology, it is a trivial process to provide cach vector object with an identification (id) number, On the basis of this unique identifier, ach and every object can thus be linked toa set of additional non-spatial attributes that describe additional properties of that object. These properties most often consist of real-world quantitative and/or qualitative variables that give the vector object meaning within the GIS (Fig. 2.11). Vector topology An extremely important concept that underlies the vector structure is the geomet= rical relationships between vector objects, referred to as topology. The analys of topological relationships is explored more fully in Chapter 11, so here it sufficient to note a few basic concepts. Firstly, topological relationships define incples RSP a Attribute data vector objets jygonid | soll type oan clay t—— 2 | clay with flint [—_ 3 | chalk bjt inked tatibute dat. In this example, euch Fa vest inks it ety tan abut table that defines nique reseed by that oye. Polygon has g the Sol type the connection and teltionships between vector objects rather than their spati lato, For xanple, when wo als cross each othe, two different topologica tcaonshis can potentially exist between those entities. If the lines simply cro, without sharing node the ines are not topologically connected. This is e«uivaen to road crowing another via an underpass and itis not possible to get from one roa t another atthe point of intersection, Ifthe roads do share a node, they are topological inked n this case, it would be equivalent to the two roads meeting aan intersection. Topological relationships are therefore defined by the presence of shared nodes between vector objects, In practice, many GIS require nodes at boa crosing and meeting points, in which case additional methods must be used to provide adequate topological information (see Chapter 11). ‘Topological relationships also define how polygons relate to each other, For ‘tanple, two adjacent polygons, perhaps representing separate parcels of land or Winey tones, ae topologically related if they share one or more nodes or arcs ae (Fig. 2 Without common nodes this relations! ip does not exist, . ae ie Tha ta get overlap and/or have a gap between them. I Wo pte anon tetionally overlap or have a gap to reflect a real ite seat abe er ually adjacent polygons have an assumed, Feats ship. The calculation of spatial relationships and trivial process, and is dependent both on the sls During the data collection phase and laionsh ent Yee" objects, care should t0 be t "ps are properly maintained and defined. that vec tha take ct lay las Provide a set of topological rules to ensure suey areas migh have 4g PE PHate Ways, For example, polyZons 7 Must Not verap’ rue, so that any instances 2.4 Data models and data structures 27 i : ‘ Fig. 2.12 Three topologically related polygons. Polygons 1, 2 and 3 share arcs (edges) ; defined by nodes cm, my and mg. ; Where this occurs are identified and the appropriate action taken (e.g. the over- lapping area is subtracted from one polygon, or a new polygon defined by the ‘overlapping area is created). | Topological accuracy also makes for more efficient storage of vector data as yector objects can then share data. Some GIS systems take advantage of this when storing the geometric definitions by only recording an arc (and its vertices) once, and then defining its relationship to polygons. In Fig. 2.12, for example, arcs cm, ‘mj and mg need only be stored once instead of twice for each of the polygon ‘boundaries they define. On large, complex, polygonal maps such as those routinely encountered with soil or geological series, this can result in a significant saving Of storage space and computational time, an issue examined in further detail in Chapter 4, 2.4.2 The raster data structure Unlike vector graphics, which use coordinate geometry to define the spatial param- eters of objects, raster graphics use a grid matrix of equally sized cells or pixels to represent spatial data (Fig. 2.13). Raster maps are therefore defined only by the number of rows and columns in the grid and the size of each pixel in terms of actual area covered. Each cell also has a value associated with it that represents the attribute status of the object at that location, In a digital elevation model (DEM), for example, each cell has a quantitative value that signifies the mean elevation across the area defined by that pixel, whereas the pixels in a vegetation map may be coded to reflect modal vegetation type. The reliance on pixels and the use of a single attribute per pixel may appear to be very limiting in comparison to the vector structure, but within the simplicity of the raster structure lies its strength, Raster Line Polygon 213i cd py pve a epreseted on aster grid anets a esly combined and mathematically manipulated as computers cay proces dip raser data considerably more quickly than vector data because the efceny with which they can store and handle grid data. The simplicity ofthe seucture doesnot reduce its functionality for, as we shall see in the nex, chapl,the aster data structure canbe used to model some extremely complex spatial phenomena. ‘Actica variable inthe raster structure isthe size of the cells, since they define the resolution ofa map by providing a minimum unit of representation. Whereas ‘rater map depicting density of archaeological sites across a large region ma consi ofa got pls tat each represent an area of a square kilometre or more, | strap showing the densiy of artefacts aross a site may use pixels that define an area of asquare mete or ess, Although computer processing speed and space iscotnuouslyinreasng, there ei catia inv mh noe eRe a revels some practical resections Stour wie ence mba can efficiently process. This, sed vent al atte mer, bu oss of performance may nip erecting a wet hg et lion or more (e.g. araier ‘ised! 00 cunt st with individual pixel sizes of 10m? will ‘25000 il) Deron ee therefore storage of information ibe oll ls pen as ¢tresoion need to be made very early Inga conegnences enc PE We show that such decisions 208 of sail patems "Pttaton of the results of analysis and nas Coming a dt ‘many inst of ave Where Piel ee ae a en ms vie the only sensible means aster data iat One migt ling the sort SOB for tn ma fag ings able to represent continuous Phy ang € analysis and cf i img a ltsfcation routines Many modern GIS systems siructure 2.4 Data models and data structures 29 “nybrid’ and offer capabilites for manipulating both raster and toalarge extent ; vector dats Wi jemow eommon to find both data structures being used vector datasets. As a resul qna single program environment, thus reducing the choice berweet the raster versus aa aneactures to tha of appropriateness for the particular needs and questore at and. From the philosophical perspective adopted in Chapter 1, the vector suretnre js generally most appropriate when the subject matter has Been conch using the nity model of space. Conversely, the raster structure is usually a Bester choi the subject matter has been conceived as a continuous field. ‘Advantages and disadvantages of the vector structure g ‘Advantages ofthe vector structure A major advantage of the veetor structure iS 1% Spatial precision, Real-world entities can be drawn and positioned with an accvi"y restricted only by practical limitations such as the precision of the recording equip rent, Artefacts, features, sites and other archaeological entities can be integrated Jha single environment, each mapped with as much spatial detail a is required for analysis, While centimetre-scale precision would not be required for the analysis of the spatial distribution of sites across a large study region, this level of precision ‘would, however, be essential for the study of the distribution of chert flakes on a knapping floor. Finding the balance between spatial precision and the minimum scale of analysis is crucial: most importantly to prevent spatial errors from influ- encing pattern recognition. In practice, it is important to recognise that increased resolution also means increased file sizes, with a corresponding burden on stor- ‘age and processing time. Another significant advantage of the vector structure is that vector objects are maintained as distinct entities and can be easily linked to attribute data records in an internal or external database. For this reason, vector- based GIS programs have traditionally led the way in terms of database integration, as complex attribute queries can be performed with relative ease. A vector map may therefore act as a window into a database, in which each object is described in great detail. Disadvantages of the vector structure Vector objects are computationally demand- ing, Every vertex and node of a vector map must be stored in computer memory and drawing vector objects requires a considerable amount of processor time. For this reason, vector data are often much slower to generate on a computer screen than raster data. The manipulation of vector data is correspondingly intensive; spatial queries involving, for example, the calculation of areas of overlap within a large set of polygons needs considerable computer-processing unit (CPU) time. Vector data also impose properties onto real-world objects that do not necessarily cor- respond with reality. The most important imposition is ‘boundedness’. Although many real-world objects do indeed have precise and discrete boundaries, certain types of dataare more ‘fuzzy’ and do not lend themselves to the hard, precise, edges of vector objects. As vector data cannot readily deal with fuzziness or imprecision, this can result in artificial precision in some scenarios. An issue related to fuzzy —_—_— Fars princes ® ses ai implied nein Pena, 4 FOE Te density of artefacts in tha enumeration 8? Fee ae tat Prion of artefacts i that Area, Which in pq js a conumans "an additional attribute COuIL Be used 19g, ay the ca8e (CBT an, but there is 20 simple way spatially to map the aay wating PDE Elevation is therefore inherently ditheys tives change wid YEE Pas such a8 points OF lines: contour linge’ nt using jon of topographic variation at set intervals, but i qq ween the lines. There are special vector yr” diticuo pedi aed imegular nenworks (TINS) that overcome these diffi nhl nee rane ens thal some pes OF Jaa ae LSS Wel suited ie vector structures state of an aMtDUIE ACTOS a Vectog ta discrete SUFVeY area may po, OX vantages of the raster structure ‘ean tern The speed at which raster data can be proce ‘ies advantages for some applications involving very lange datasets and they are ral other key areas in which the raster structure can offer advantages gy, arto formas Firstly, raster data ae very good for mapping continuously varyig, Fewena,soch 8 elevation, asthe continuous cell-based structure is akin to, ‘Continuously varying surface. The raster structure is also very good at represen, ing real-world entities that have fuzzy boundaries. For example, a distribution « anefocts collected in a ploughed field could be represented more realistically b, sing raster cells that show the changing density of material rather than a singl polygon that arbitrarily defines the site's area with a single density value. Wher this type of information is crucial, then the raster data structure offers a clew advantage. Secondly, raster datasets can be mathematically manipulated and com. bined more easily than polygon maps, making it an exceedingly powerful tool for spatial modelling. A simple model of agricultural potential, for example, may be ‘constructed by combining data from several different sources, such as raster ‘maps of elevation, slope, aspect, soil drainage and soil type in a process called map algebra, Thirdly, aerial photographs, satellite images and geophysical surveys pro-

    You might also like