Where on Earth Are We

8/13/2019 Where on Earth Are We

1/24

2003

Where on Earth are We?The Global Positioning System (GPS)in archaeological field survey


2/24

IntroductionWhere on Earth are We?

Global Positioning System (GPS) is a term that is now probably familiar to most people,Surveying has always played a fundamental particularly sailors, pilots, hikers, motorists and a multitude of professionals androle in the recording and analysis of hobbyists who have a common requirement, to know their position in relation to a maparchaeological sites and landscapes. In or chart. For these groups of people, small, portable GPS equipment exists that canrecent years, there have been a number easily be carried in the hand or mounted on a vehicle dashboard. The variety andof significant technological developments availability of these hand-held GPS sets in the high street stores is testament to howin surveying equipment. As a result, the this once secret military technology has entered our everyday lives. In the professionalworld of archaeology has had to embrace world of surveying, GPS has also become an everyday tool. It is routinely used to makeand learn to translate a new language maps, to set out roads, to record the locations of street lamps and to measure volumesdeveloped by surveying equipment and levels in quarries, as well as a multitude of other tasks, including those undertakenmanufacturers, surveyors and mapmakers in the area of archaeological field survey.

to describe their ever-developing productsand techniques.The most recent of these It is impossible in a technical paper such as this to cover all aspects of archaeologicaldevelopments is known as GPS (Global surveying with GPS. The intention of the paper is to provide sufficient background onPositioning System). the subject to inform archaeologists as to whether this might be a suitable tool for them,

and to illustrate how GPS might be deployed. The Archaeological Investigation teams inThis paper is intended to provide guidance the former RCHME 1 and English Heritage have many years of experience of using GPS

to field archaeologists on the use and for earthwork and landscape recording, and it is hoped that, through this paper, some ofapplication of GPS in archaeological the experience and lessons learnt can be passed on to the wider archaeological

survey. Its principal aim is to simplify community. GPS has been deployed on numerous upland and rural landscape projectscomplex surveying and technological and is now the bedrock of any survey task undertaken by the teams. It has been used onprocesses to enable archaeologists to a wide variety of projects with differing survey requirements, ranging from recordingmake informed decisions as to whether vast tracts of surviving archaeological landscapes on moorlands at mapping scales, to

this is a technique that should be used recording individual stones in large-scale surveys of prehistoric roundhouses.for site-based surveys or for the mapping To help archaeologists to understand GPS, this paper has four main sections:of large landscapes. As well as providinga number of case studies, it explains in Global Navigation Satellite Systems an outline of the satellite systemssimple terms some of the background Understanding space and time important background information that it isand implications of using GPS that an necessary to understand before using GPS for surveying and mappingarchaeologist cannot ignore if the GPS equipment practical information about types, choices and applications of

technology is to be used for surveying and GPS equipment. The types of GPS equipment are broken down into three categories,

mapping purposes. navigation-grade (low accuracy positioning and small-scales mapping), mapping-grade(medium-scales mapping and GIS data recording), and survey-grade (high precisionsurvey control, large-scale site surveys and large-scales mapping).

Surveying with GPS the practical aspects of using this technology in the field. Fivecase studies, illustrating the use of GPS in archaeological projects, are included as aguide to the possible range of applications.

It should be remembered that GPS is purely a surveying tool that provides data that canbe turned into maps, plans and models. It does not provide any archaeological analysisskills, which are the foundations for the interpretation and understanding ofarchaeological sites and landscapes. These skills are still the province of the earthboundarchaeologist. However, a metrically accurate survey underpins the process of analysisand it is important that its results can be relied upon. For many archaeological projects,the survey process is the first stage towards recording and understanding, and there arenormally three main requirements of the survey data. The first is to know where anypoint is on a site, usually in three dimensions; second is to know where each point is inrelation to each other point so that a plan can be made of the site; and third is to knowwhere the site is in relation to a map. So what is GPS and how can it help thearchaeological community fulfil these requirements?

1 Terms and acronyms italicised at their first use are defined in the Glossary at the end of this publication.

2


3/24


4/24

for the user to have a very basicunderstanding of some of the principlesand terminology. This will enable informeddecisions to be made as to whether theright equipment and techniques are beingapplied to a particular project and willensure that when data have been collected,they can be deemed reliable for surveys,maps, databases and GeographicalInformation Systems (GIS).

Those who have worked with OrdnanceSurvey National Grid references, maps andarchaeological records systems will befamiliar with the importance of coordinategrid references for the accurate location ofinformation. Unfortunately, in many ruraland remote moorland areas these are notalways accurate enough for archaeologicalpurposes. The coordinate might beinaccurate, either because it was fixed by

insecure methods, such as compass bearingand pacing, or was simply guessed,particularly in areas away from mappeddetail. Traditionally, a coordinate is likely tohave been defined in relation to featuresshown on the map, such as walls, buildingsor other features, and thus the farther awayfrom the mapped detail, the more likely it isthat the grid reference will be wrong.Unfortunately, in many cases, theinaccurate grid reference still appears eitheron the current archaeological record or in

the GIS. It is also essential to recognise atthis early stage that a coordinate derivedfrom GPS might not necessarily be correctfor a point, the position of which you wishto find or confirm, and might be verydifferent from one derived from the detailshown on the map. To understand why thisis and how it can be remedied, it isnecessary to understand three terms,ellipsoids, geoids and coordinate systems . Allthree terms are relevant to understandingthe accuracy of a position, grid reference orheight derived from a GPS receiver. Theywill enable the user to understand thenature of the positional and heightinformation they are collecting with GPSequipment, and also to understand theprocess needed to transform that data tolocal maps. Also, and on a more practicalnote, they are terms that a user willencounter on GPS software in the field andin the office; therefore the better they areunderstood, the easier it will be to makeappropriate choices in the field and in thedata processing.

EllipsoidsWhether providing a position on theEarths surface by an old fashionedsextant or by a modern GPS receiver,

underpinning the mathematics andprojection of any map will be an ellipsoid(Fig 2) the mathematical surface thatbest approximates the curved shape of theEarths surface. Historically, however,accurately measuring the size of the Earthhas been difficult and as a result theellipsoids defined in the past were oftenchosen to best fit on a regional rather thanon a global scale.

Fig 2 Ellipsoid a mathematical surface that approximates the shape of the Ear th (based on Ordnance Sur vey 2002).

The Earth is a complex and dynamicbody and geodesists through the centurieshave applied themselves to the science ofdefining various ellipsoids on which tobase mapping projections. Mapprojections are necessary to allow theactual curved surface of the Earth (or forsimplicity, the ellipsoid) to be representedas a flat plane. To address local mapping

requirements, many countries havedevised their own map projections andreferencing systems. On a more globalscale, long before satellites had ever beeninvented, maps and charts covered thesurface of the globe using latitude andlongitude positions based on celestialobservations. However, it is only since theadvent of man-made satellites that therehas been the capability to measuredistances on truly international scales andthe possibility to construct a precise globalcoordinate system. In recent years,geodesists have had to respond to thisadvance and determine the relationshipbetween the old and new coordinatesystems in use. This relationship isachieved mathematically using atransformation.

This procedure means, in practical terms,that if points surveyed using GPS need tobe related to existing map detail, then theyrequire transformation from the globalreference system provided by the satellite

navigation system onto the local mapprojection and its reference ellipsoid. Inthe case of Great Britain, this is theOrdnance Survey National Grid, and theAiry 1830 ellipsoid. This process of

transforming coordinates is now mostlycarried out by computers using predefined mathematical models. OrdnanceSurvey freely provides a model to relateGPS derived coordinates to NationalGrid, the OSTN02 . This subject isdiscussed in more detail below, inUnderlying principle 2.

GeoidsThe geoid is another theoretical surface.The surveyor needs to visualise this inorder to understand heights derived fromGPS. The geoid is the surface formed bymean sea level over the Earth and itsimagined extension under the land areas(blue on Fig 3).

Fig 3 Ellipsoid, geoid and topographic surfaces (based onOrdnance Survey 2002).

In order to give meaning to the concept ofrelative heights around the world thegeoid provides a theoretical figure similarto the ellipsoid, thus providing aworldwide datum to which heights abovesea level can be related (Fig 4).

Fig 4 A simple representation of ODN height of a point Pon the land surface that is, its height above sea level.Thedotted line continued under the land is essentially a geoidmodel (based on Ordnance Survey 2002).

This acts as a fundamental reference forall height measurement. Historically, inmainland Great Britain, heights have beenrelated to Ordnance Datum Newlyn(ODN ), which is on the actualtopographic surface, and to which all

heights on Ordnance Survey maps andbench marks are related. Just as theellipsoid coordinates have to betransformed to local maps, so heightsderived from GPS have to be transformed

4


5/24

to ODN, or the other vertical datums 2 inuse in Great Britain (Table 1). Thetransformation from GPS coordinates tothe relevant national height datum forexample, ODN is most accuratelycarried out using Ordnance SurveysOSGM02 geoid model. Failure to takethis transformation into account canrender height values derived from GPSseriously inaccurate by up to 50m!

Table 1 The main height datums used inGreat Britain

It is based on a Transverse Mercatormap projection and has formed thebasis of all mapping in Great Britain. Itis the national coordinate system fortopographic mapping at all scales formost maps available both in print anddigital form, including Land-Line data,and is commonly used as the referencegrid for spatial records in archaeologicalrecords collections. The definition ofOSGB36 has recently undergone asubtle but fundamental change ( seebelow , The National GPS Network).

ODN Ordnance Datum Newlyn, atraditional vertical coordinate system

area datum

GB mainland NewlynOuter Hebrides StornowayIsle of Man DouglasShetlands LerwickOrkney Newlyn

St Kilda St KildaScilly Isles St Marys

As a new GPS user, it is easy to set up aGPS receiver and to produce somecoordinates and numbers relating tolatitude, longitude and height, or toeastings, northings and height. However,without some understanding of the aboveit will be hard to know which coordinatesystem they belong to and which

transformations will be necessary to relatethem to an Ordnance Survey map or benchmark. Some GPS receivers or software usevery basic transformations, which can be inerror by up to 10m, so it is important toknow what transformation is being used.The most accurate method in Great Britainis to pass the GPS coordinates throughOSTN02 or OSGM02.

Underlying principle 2: maps, gridreferences and coordinate systemsTo use GPS data for archaeologicalmapping in Great Britain, it is necessaryto understand the different coordinatesystems used by the Ordnance Survey.As the government agency responsible fornational standards in spatial positioning inGreat Britain, the Ordnance Survey hasdefined three national coordinate systemsand it is to these that all modern GPSdata are referred. These are:

based on mean sea level tidalobservations at Newlyn in Cornwallbetween 1915 and 1921. ODN isrealised by a network of 200fundamental bench marks and half amillion lower-order bench marks

throughout mainland Britain (Table 1gives the other main official verticaldatums used throughout Great Britain).All observations for this network wereundertaken by traditional levellingtechniques. All bench mark values, spotheights and contours on OrdnanceSurvey maps throughout mainlandBritain are related to ODN. This is theusual definition of heights above meansea level in Great Britain. The standardgrid reference used for most

archaeological recording systems inarchaeology, such as SMRs, NMR andother systems is a combination ofNational Grid (OSGB36 ) and ODN egeastings, northings and height.

ETRS89 European TerrestrialReference System 1989. This is thenational coordinate system for three-dimensional GPS positioning. It is amore precise version of the WGS84GPS coordinate system ( see below ) andis now the standard system usedthroughout Europe. If the accuracyrequirement of a survey is 1m or worse,then WGS84 coordinates can beaccepted and converted to NationalGrid (OSGB36 ) if necessary. If therequirement for accuracy is better than1m, then ETRS89 coordinates shouldbe derived using the National GPS

Network .

GPS coordinate systems:WGS84 The National Grid or Ordnance Survey and ETRS89

Great Britain 1936 ( OSGB36 ). While a grid reference is thought of in

Historically this is a traditional relation to a coordinate system on a map,horizontal (eastings and northings) such as the National Grid on Ordnancecoordinate system using the Airy 1830 Survey maps, a position derived fromellipsoid, originally realised by a GPS data is related to a global coordinatenational network of triangulation pillars. system known as the World Geodetic

2 See Glossary for specific definition of datum as used in this publication.

System 1984 (WGS84). WGS84 is aninternationally agreed, global referenceframework of coordinates within which apoint can be defined anywhere in theworld. Coordinates in this system areusually expressed as latitude, longitudeand ellipsoid height. While most peoplewill be familiar with latitude andlongitude, ellipsoid height might be moreconfusing. This is not the same as a heightabove sea level, as shown on a map, andexists only mathematically.

In Figure 4 it can be seen that, whileactual survey measurements at P is takenon the topographic surface (brown), landlevels are usually related to mean sealevel (blue). Heights derived directlyfrom GPS relate to the theoreticalmathematical surface of the ellipsoid(red). In order for GPS ellipsoid heights

to be related to mean sea level anothertheoretical datum, the difference betweenthe two surfaces, needs to be defined as ageoid model.

The introduction of WGS84 has meantthere is no longer a requirement forregional coordinate systems ( see Fig 2).WGS84 was designed for navigationapplications throughout the world, wherethe required accuracy is 1m or worse.A high-accuracy version of WGS84,

known as ITRS (International TerrestrialReference System), has been created in anumber of versions since 1989, and this issuitable for international high-accuracyapplications (it is used mostly bygeoscientists).

There is a problem, however, with tryingto use a global coordinate system forland surveying in a particular country orregion. The problem is that thecontinents are constantly in motion withrespect to each other, at rates of up to120mm per year. There are in reality nofixed points on Earth. In common withthe rest of Europe, Great Britain is inmotion with respect to the WGS84coordinate system, at a rate ofapproximately 25mm per year. Over adecade, the WGS84 coordinates of anysurvey station in Britain change byc 0.25m owing to this effect, which isunacceptable for precise survey purposes.

For this reason, the European Terrestrial

Reference System 1989 (ETRS89) is used asthe standard precise GPS coordinate systemthroughout Europe. ETRS89 is based onITRS, except that it is tied to the Europeancontinent and fixed in time (1989).

5


6/24

It is hence steadily moving away from theWGS84 coordinate system. In 2000, thedifference between the ITRS coordinatesof a point and the ETRS89 coordinateswas c 25mm, and increasing by c 25mmper year. The relationship between ITRSand ETRS89 is precisely defined at anypoint in time by a simple transformationof the coordinates, published by theInternational Earth Rotation Service,available from the information section ofthe National GPS Network web site.

In summary, while these global coordinatesystems might seem confusing, it isimportant to recognise that they haveimplications for using GPS for relatingobjects or sites to Ordnance Survey maps(see below , National Grid). ETRS89 hasbeen officially adopted as a standardcoordinate system for precise GPS

surveying by most national mappingagencies in Europe. By using ETRS89 it ispossible to ignore the effects ofcontinental motion because, to a highdegree of accuracy, the ETRS89coordinates of a survey station stay fixedas long as there is no local movement ofthe survey station. If control from theOrdnance Survey National GPS Networkis used, then the survey will automaticallybe in ETRS89.

The native format for any GPS receiver is the World Geodetic System (WGS84),expressed in latitude, longitude andellipsoidal height. Whenever results arerequired in a local grid system it isnecessary to use a transformation. In GreatBritain Ordnance Survey makes their latestand the most accurate transformations(OSTNO2 and OSGM02) availablefree through their web site. While using this

transformation cannot improve the basicquality of navigation solution positions fromsingle GPS receivers, using it will ensure

that no extra error is added in transforming from latitude and longitude toNational Grid coordinates.

Ordnance Survey mapping and theNational GridA true representation of the three-dimensional Earths surface cannot becorrectly reproduced on a flat piece ofpaper, nor can the properties of position

or relative positions of features all beretained. To best represent the surfacebeing mapped, a cartographer has todefine or choose a map projection. ForGreat Britain, the Ordnance Survey in the

19th century adopted the Cassiniprojection for its County Series mapping.The result of this exercise was to producea series of maps for each county, withindividual mathematical adjustments foreach. After the Second World War, it wasdecided that the County Series 1:2 500survey should be recast on Nationalinstead of County sheet lines and that asingle National Grid should besuperimposed on all maps. This resultedin a new series of National Grid maps,which were compiled from the earlierCassini projection onto a single, nationalTransverse Mercator projection. This newprojection was adopted to achieve a betterfit over the whole of Great Britain, towhich the earlier Cassini wasmathematically unsuitable. Thus, themajority of 1:2 500 maps in use today(many still forming the basis of SMR

systems) were created in this way andconsequently combine the errors andadjustments of that recasting from manyindividual county projections to a singlenational projection. Although this solutionenabled a rapid production of NationalGrid maps, it inherently contained manyinaccuracies (for more background seeHarley 1975; Seymour 1980; OrdnanceSurvey web site). It is important for thearchaeologist to be alerted to the potentialmetric differences between the 19th

century maps and the National Gridseries, as these earlier maps remain amajor source of map-based surveys ofsites and landscapes that havesubsequently been destroyed. This isparticularly true of industrial landscapes,for which it is a common requirement formodern surveys of surviving remnants tobe correlated with features on CountySeries and National Grid mapping. Thiscontinues to be a problem in many areasof archaeological recording where newsurveys using GPS have to be integratedinto a map regression exercise, and wheremetrical accuracy is important (theaccuracy of GPS in relation to OrdnanceSurvey mapping detail is discussed below,in Underlying principle 3).

With the advent of GPS, coordinates canbe produced that are more precise thanthe existing National Grid mapping.Accuracy has become such an issue forusers of maps that in recent years theOrdnance Survey has issued numerous

information and consultation papers onthe subject, and has had to makefundamental changes to the mappingdatabase to accommodate and address thisproblem. As Ordnance Survey maps are

fundamental to archaeological surveyingand recording systems it is sensible forpotential GPS users to familiarisethemselves with these changes (OrdnanceSurvey 2000af).The Ordnance Surveyhas initiated a programme of accuracyenhancement of its large-scale maps,which will address the historicalinaccuracies noted above, but this mighttake some years (Ordnance Survey 2002).

From the standpoint of archaeologicalrecords systems, however, it is importantto recognise that the Ordnance SurveyNational Grid (OSGB36 ) is staying thesame. Critically, the Ordnance Surveydoes, however, enable GPS users to tieGPS positions to its mapping through theuse of the National GPS Network andpublished transformations.

The National GPS NetworkTo ensure compatibility between thediffering coordinate systems, OrdnanceSurvey has defined a new GPS nationalpositioning infrastructure based onETRS89. This infrastructure has replacedthe traditional network of triangulationstations so familiar on mountain- andhilltops, which formed the realisation ofthe National Grid (OSGB36 ). This newnetwork of fixed points is known as theNational GPS Network. Using survey-

grade GPS receivers, data can therefore becollected on site by any user and post-processed in relation to this network, thusproviding precise positioning relative tothe ETRS89 coordinate system. GPS datacollected by any user can now be relatedto Ordnance Survey maps and recordingsystems based on them. The NationalGPS Network consists of GPS stations oftwo types:

Active stations a network of 32 GPSreceivers (December 2002) that have beenpermanently installed at locationsthroughout Great Britain. They have beendistributed so that any point in GreatBritain should be within 100km of an activestation and most major urban areas arecovered by more than one. Data from thesestations are continually monitored andprocessed by the Ordnance Survey inSouthampton and are available for eachstation through their GPS web site. Theyare termed active because they operatecontinuously and their inclusion in a GPS

survey does not require the user tophysically occupy the control point. Theprecise position of each is known in relationto the ETRS89 coordinate system. Theestablishment of these stations and the

6


7/24

availability of data via the internet from theOrdnance Survey GPS web site has made amassive impact on the ease of use of GPSfor high-accuracy surveying.

Passive stations a network of about900 ground-marks that have beenestablished throughout Great Britain.These are in accessible locations and theposition of each has been established inrelation to the ETRS89 coordinatesystem. Any point in Great Britain willnormally be within 2035km of a passivestation . The main difference (anddisadvantage) between this passivenetwork and the active station network isthat a passive station has to be occupiedby the users own GPS receiver.Coordinates, stations descriptions andphotographs can be freely downloadedfrom the Ordnance Survey GPS web site.

Using data from the active stations or fromthe coordinates of the passive stationsmeans that newly surveyed points can beaccurately related to the National GPSNetwork. By using this facility the surveyorcan obtain ETRS89 coordinates for theirunknown points. These points can then betransformed to eastings and northings onthe National Grid (OSGB36 ) by applyingthe Ordnance Survey DefinitiveTransformation (OSTN02), and to

heights above mean sea level that areconsistent with the traditional bench marksby using the National Geoid Model(OSGM02). These are computer-basedtransformations and are available withinmany GPS software packages, as well asfree through the Ordnance Survey GPSweb site (Fig 5).The National GPSNetwork and the online coordinate convertersensure that modern survey data collectedwith GPS are easier to integrate withdatabases and GIS mapping systems thanwas previously possible. A comprehensivebooklet describing in more detail thecoordinate systems and the use of GPSdatasets with Ordnance Survey mapping(Ordnance Survey 2002) is available fromthe Ordnance Survey web site.

Alternatively, GPS surveys can be locallyfitted to mapping by using a localcalibration/transformation (usuallyincorporated within GPS software). Thisenables surveyors to sample survey pointswith known map coordinates before

starting their surveys, to provide a good fitto local map detail. It should beunderstood that this type of survey will beno more accurate than the local map, sothat if mapping accuracy is poor in that

Fig 5 Conversion of GPS data to National Grid coordinatesand Ordnance Survey Datum heights (based on OrdnanceSurvey 2000a).

area, the resultant GPS survey willcontain the same errors and the surveywill be skewed by the mapping errors.

A word of warningMany software packages and GPS receivers

have different GPS (WGS84) to OrdnanceSurvey (National Grid) transformationsother than the more robust OSTN02and OSGM02. Using these simpletransformations might add between 1mand 10m of error into the survey as a resultof the poor level of modelling. This mightnot be an issue if the survey is in a remotearea, but if it needs to be fitted toOrdnance Survey mapping the user shouldbe aware that these errors will be added. Itis extremely important therefore for a userto find out whether the software usesOSTN02 and OSGM02.

In summary, collecting GPS data inETRS89 from the National GPS Networkwill make it possible for archaeological ortopographic features to be mapped againstexisting National Grid mapping and willraise future mapping based on ETRS89 toa defined European standard. GPS datawill also enable the user to establish trueheights in relation to Ordnance Surveyheight datums regardless of where the

nearest bench mark is.

Underlying principle 3: accuracy and GPSIn the context of GPS, the term accuracycan be confusing. For simplicity, accuracy

has three distinct meanings when used inconjunction with GPS and survey: relative,map and absolute. These terms arecharacterised as follows:

Relative accuracy means the relativedirection and measurement betweenpoints.

Map accuracy means the accuracy ofthose points on a map; in Great Britainthis would normally be an OrdnanceSurvey map.

Absolute accuracy means a measurethat indicates how closely thecoordinates of a point agree with thereal coordinates of the point on thesurface of the Earth.

Relative accuracyUnderpinning any surveying technology isthe concept of relative accuracy the

measurement and spatial relationshipbetween points in a survey. This applies toall surveying equipment and methodology,not only to GPS. The final scale of a plan tobe produced for an archaeological site orlandscape will usually determine the type ofequipment needed to give the requiredaccuracy. For example, this can be equatedwith the need to set up and produce planinformation within a 1m grid on a site, withheights relative to a site datum. Thearchaeologist needs to be able to rely on the

points on the grid being precisely 1m apartwhen measuring between them, and at rightangles to each other, as well as havingconfidence that heights are all relative toeach other and to a temporary bench mark.With such confidence in the relativeaccuracy, the site surveyor can produce aplan at an appropriate scale and assignheights to finds and stratigraphy. Where therecorded points are in relation to the large-scale Ordnance Survey map is not relevantat this stage.

Relative accuracy can be easily achievedwith traditional technology such astheodolites, tape measures, EDM(Electromagnetic Distance Measurement)and levels. The concept of relativeaccuracy and making choices in relation toappropriate surveying equipment andtechniques for archaeologists is coveredin detail elsewhere (Bowden 1999; 2002).In terms of relative accuracy, survey-gradeGPS offers no advantages over traditionalequipment such as theodolites and EDM.

In terms of speed of data collection,however especially where accuratethree-dimensional modelling is arequirement this type of GPS is a muchmore cost-efficient way to gather large

7


8/24

volumes of data to a high degree ofrelative accuracy. It can be a far moreeffective way of setting out grids over largeand undulating areas. Hand-held GPScannot produce sufficiently accuraterelative measurements to be used for planmaking or height calculations ( see below ,GPS Equipment).

Map accuracyTo use GPS successfully, it is important torecognise at an early stage that there aredifferences between relative accuracy andmap accuracy and therefore to ensure thatthe correct equipment and methodology isadopted for an archaeological project.Where GPS makes the most impact insurveying is the locating of site surveys,objects, monuments and landscapesrelative to maps. In the past, long traverseswith EDM and theodolite between

triangulation pillars followed by copiouscalculations were necessary to fitarchaeological features on the groundonto Ordnance Survey large-scale maps.This was a time-consuming and specialisttask. Because of this, easier and lessaccurate methods were devised, with theposition of archaeological surveys beingbest fitted onto paper maps, or ontodigital maps using CAD systems. Thesepractices were not very satisfactorysolutions to maintaining accuracy in

computer-based geo-referencing systems,where the correct spatial relationshipbetween features on the ground and themap is important.

This positional deficiency for futurearchaeological records management wasrecognised some years ago by thearchaeological field investigators of theRCHME. Throughout the 1980s theydirectly related surveys to large-scale,Ordnance Survey maps usingtriangulation pillars to ensure that theywere accurately recorded onto theNational Grid for longer-term integrationinto GIS datasets and archaeological datarecords. This was particularly relevant tothe more open landscapes of themoorlands, where the acquisition ofaccurate grid references was otherwisedifficult (RCHME and PPJPB 1993;Ainsworth and Barnatt 1998). At aboutthe same time, similar techniques werebeing used by other archaeological unitsfor recording excavations into large-scale

urban mapping, using Ordnance Surveycontrol points. However, the sort of skillsnecessary to do this with traditionalsurvey equipment lay mostly within thespecialist land survey disciplines.

accurate to better than c 10m ( see below ,GPS Equipment). At the other end of thescale, it might be because the map is notas accurate as the GPS-derived position.This is less of a problem in the urbanareas, where 1:1 250 mapping is moreaccurate, than in rural areas. Thepositional accuracy that might be expectedfor large-scale Ordnance Survey mapdetail is illustrated in Table 2.

Table 2 Positional accuracy expected forlarge-scale Ordnance Survey maps (basedon Ordnance Survey 2000a)

scale of achieved absolutemapping (map) accuracy 99%

63% confidence confidence

1:1250 urban < 0.5m < 1.0m1:2500 rural

resurveyed < 1.1m < 2.4m1:2500 rural < 2.8m < 6.0m1:10 000 moorland < 4.1m < 8.8m

It is possible to undertake independentsurveys using survey-grade equipment ( seebelow , GPS Equipment) and to transformthese to local map detail to fit in witholder large-scale mapping using a localcalibration/transformation, as explainedearlier. This solution, although only beingas accurate as the local map detail used, is

a pragmatic methodology for fitting smallarchaeological surveys into existing map-based records.

Absolute accuracyThis is a measure of the closeness of apoints coordinate to its true value.Although high relative accuracies havebeen achievable for some time prior tosatellite surveying techniques, precisemeasurements were restricted to line ofsight only. This meant that each continentwas positionally isolated and thatinterrelationships between continents wereimprecisely known.

It is this level of accuracy, on a globalscale, that is known as absolute accuracy.Because of the worldwide nature of GPS,and its underlying coordinate system, highabsolute accuracy on a global scale ispossible. As well as underpinning thecoordinate structure of Ordnance Surveymapping through mathematicaltransformations, the relevance of this fact

for archaeologists in Great Britain is to befound particularly in the recording ofintertidal zones and maritime archaeologyaround the coast. In the past thesemeasurements have been inconsistently

The advent of GPS, however, has meantthat the accurate location of surveys andmonuments onto maps and geo-referenceddatabases is no longer the realm of thespecialist land surveyor, and can beundertaken by the archaeological sitesurveyor. Now, survey-grade GPSequipment and software are sufficientlysophisticated to combine relative accuracyand map accuracy in the same package.To position sites, f inds, trenches andlandscape surveys accurately into thenational mapping of the Ordnance Survey

and subsequently to GIS, SMR andNMR databases the archaeologist nowonly has to identify the GPS equipmentthat will achieve the relevant mapaccuracy required. He or she will alsoneed to understand some of the conceptsthat are outlined in this paper related toOrdnance Survey mapping.

The relationship between relative accuracyand map accuracy can be illustrated bytaking two theoretical field corners (a andb on Fig 6). The relative accuracy of twofield corners might be correct (that is, themeasurement between them is correctcompared to the measurement on themap), but where they appear on the mapmight be wrong in relation to the NationalGrid printed on the sheet. If the positionof those same points were measured with

survey-grade GPS, and fitted to theNational Grid using modern mathematicaltransformations, they would potentiallyappear at a different place on the map tothose published (a1 and b1 on Fig 6), butwould retain their same relative accuracy.

Fig 6 Relative accuracy as opposed to map accuracy.

The GPS user has to recognise that planpositions of features recorded by GPS

might not agree with the same features asshown on a large-scale Ordnance Surveymap. At one level, this inaccuracy mightsimply be because the type of GPSreceiver is navigation-grade and cannot be

8


9/24

recorded in a variety of coordinatesystems, including National Gridreferences, geographical latitude andlongitude, and GPS WGS84 latitude andlongitude.

Now, global coordinates as defined byWGS84, and through mathematicaltransformations based on it areproviding a common referencing systemfor all data collected using GPS anywherein the world. Thus, through softwaretransformations, positions derived fromGPS can be plotted onto, or recoveredfrom, different map grids and navigationcharts. Also, most GPS equipment has anumber of worldwide map and chartprojections built into the software toenable WGS84 positions to be convertedto local maps, ensuring that GPS can beused anywhere in the world.

GPS accuracy and types of equipment:

Navigation-grade GPS. Map accuracyand absolute accuracy approximately10m. Good for finding location in relation

to maps and relocating sites, but notsuitable for site survey. Notoriously badfor height information.Mapping-grade GPS. Map accuracy andabsolute accuracy down to 1m can be

achieved, in real-time or if post-processed.Suitable for mapping up to 1:2 500 scalebut not suitable for site survey.Survey-grade GPS. Centimetre relativeaccuracy. Map accuracy and absoluteaccuracy to the nearest 100mm can beachieved either in real time or by postprocessing.

Increasing the accuracy of positionsderived from GPSThe most common technique forincreasing the accuracy of positionsobtained with GPS equipment is that ofDifferential GPS or DGPS . DifferentialGPS positioning is the correction of theraw GPS position given at a rovingreceiver. GPS data collected at a staticreference (base) station located at aknown coordinate is used in thecomputation of the position of aroverreceiver located at an unknownposition, either in real time or afterwardsin post-processing (Fig 7).

Two of the main error sources of GPS arethe predicted positions of the satellitesthemselves (which affects the distance) andthe delay in the transmitted signal due to

the atmosphere (which effects the timing).By locating a GPS receiver at knowncoordinates, tracking the same GPS signalssimultaneously and calculating thedifference between the known andcomputed positions, these errors can belimited. This technique works on theconcept that the errors in the signalsreceived at the reference station will alsobe present at the rover. However, as thedistance between the two increase thisassumption becomes less valid.

The technique of DGPS correction can beutilised either by collecting data in thefield at a base station and roversimultaneously and subsequently postprocessing it with computer software, or byusing GPS equipment that has thecapability of receiving and usingtransmitted base-station data in real time.

With post-processing, a base stationremains static at a fixed point and collectsGPS data, while a rover unit moves aroundcollecting data for a short period at eachpoint of interest in the survey area ( seebelow, Static surveying). At the end of thesurvey all the data are processed in acomputer and each of the survey points isaccurately fixed relative to the base station.With real-time surveying ( see below ,Underlying principle 4), the base stationcalculates the inherent errors in GPS (by

using its known position and the GPSsignal) and then continuously transmitsthis error correction factor to the rovingunit (Fig 7). Using the National GPSNetwork, the position of the base stationcan be tied to the National Grid formapping purposes, if required.

Real-time correction services are availablethat enable a user to operate a single

receiver. Accuracies of 1m to 2m areachievable (depending on the service andGPS equipment used) through correctionsignals broadcast from either ground-based transmitters or from geo-stationarysatellites. These correction services areavailable free in Great Britain from theGeneral Lighthouse Authority (GLAbeacon correction service:www.trinityhouse.co.uk), from commercialproviders (GPS suppliers can providedetails) or from the soon-to-be-fully-operational (2004), ESA-funded EGNOSsatellite system.

Underlying principle 4:coordinates now or laterAn important aspect of understanding thebasics of GPS sufficiently well to makeinformed decisions about appropriateequipment and methodology is identifying

whether positional data are required in realtime in the field, or whether it can simplybe collected in the field in a data-logger andprocessed at a later date on a computer inthe office (known as post-processing).

At the lowest level, a hand-held,navigation-grade GPS receiver is a real-time system as it allows the user to view orstore a coordinated position in the field.However, as explained above, thehorizontal accuracy of this procedure is

unlikely to be any better than c 10m. Somemore sophisticated hand-held GPS sets doallow the storage of GPS data that canthen be post-processed on a computer,which can increase the level of accuracy.

As explained above, positions can be fixedin real time in the field and these systemsare more commonly known as real-timekinematic (RTK) for survey-grade systems

Fig 7 Differential GPS. With real-time kinematic survey, data are broadcast from the base station to the rover via a telemetry link.

9


10/24

and differential GPS (DGPS) formapping-grade systems. The majoradvantage to real-time surveying is that theuser can be confident in the GPS fixedpositions before leaving the site. If there isa data-logging device equipped with avisual display, such as a pen computer orPersonal Data Assistant (PDA), theresulting coordinates can be instantlyplotted and visualised as they are recorded.This could have digital maps and planspre-loaded into it so that revision andupdating can be instant. The position offeatures that are not visible on the ground,but that are known from historic maps,aerial photographs or geophysical surveyscan also be set out on the ground. Data-logging devices might also accept inputfrom other survey equipment to completesurveys where GPS is not an appropriatecapture method; these include tapes,

theodolites, reflectorless distance lasers,and other equipment. Office time is savedas the whole survey can be completed andvisualised in the field. The disadvantage ofreal-time surveys is that the equipmentrequired is generally more expensive thansystems that use post-processingtechniques, but this aspect has to bebalanced against operational requirements.

Post-processing is performed at some timeafter the raw GPS data have been recorded.

This process is performed on a computerusing suitable processing software. Oneadvantage of post-processing is that theequipment is cheaper than the equivalentreal-time systems for a similar level ofaccuracy. Post-processing often requires thelearning of yet another softwareprogramme; however, most manufacturersproduce their own proprietary processingsoftware and run training courses ifrequired. A major disadvantage is thatsurvey data cannot be visualised until it hasbeen processed on a computer. Postprocessing is often used to provide controlfor a site with real-time methods being usedto locate the detail or to set-out points.

GPS Equipment

Having looked at some of the underlyingprinciples that underpin GPS and its use, itis necessary to move to the next stage,which is to describe the equipment itself. Itis easy to become overwhelmed by the sheernumber of articles, technical reviews,

mathematics, science and advertisingliterature that accompanies the field ofGPS. To simplify this ever-expandingequipment market, the types of GPS thatshould be considered by field archaeologists

Fig 8 Navigation-grade (hand-held) GPS (photograph by Alun Bull, English Heritage).

can be listed in three categories: navigation-grade (hand-held); mapping-grade (GISdata collection) and survey-grade.

These types are principally defined by theirrespective levels of accuracy (as definedabove in Underlying principle 3) andsuitability to archaeological recordingpractices in the field. The differences among

the various systems are largely due to thedifferent signal processing techniques usedby the receivers to determine theirpositions. It is important to remember thatthe receivers are not the only pieces ofequipment necessary; there will almostcertainly be a need for tripods, poles,batteries, chargers, computers and otherbits, especially for mapping and survey-grade GPS. Without these peripheral piecesof equipment which can be costly topurchase the receiver will be of no use.

Navigation-grade (hand-held) GPSHand-held GPS receivers are generallymarketed at walkers and sailors, and forin-car and aircraft navigation (Fig 8).They are of very limited use as surveyinginstruments. However, they are still usefulas a means of positioning archaeologicalsites or finds in areas where there is littlemap detail ( see below , Case Study 1).Because they use the coarser code part ofthe signal they can often work better underwoodland canopy than survey-grade GPS

can. It must be understood, however, thatthis type of equipment will not generallyprovide a horizontal position to better thanc 10m. In remote areas this is certainlybetter than nothing, but attempting to plan

a site, or to record features relative to oneanother at this level of accuracy is not tobe recommended. Some receivers make itpossible to record positions againstmapping at smaller scales, which cansubsequently be transferred to othercomputer systems; and hand-held receivershave been used successfully to recordroutes and data from archaeological aerial

reconnaissance projects. The inbuiltsoftware in navigation-grade GPS oftenalso provides positioning within a numberof international and local map datums, forexample National Grid (OSGB36 ), butusually at the 2m to 10m level oftransformation accuracy. It is possible withsome units to download positions inWGS84 and then to use the OSTN02and OSGM02 transformations from theOrdnance Survey GPS web site. Somesystems can add differential capability tonavigation-grade GPS receivers at a smallcost.

Mapping-grade (GIS data collection) GPSThese are lightweight systems that can beused for surveying and mapping purposeswhere accuracies between 0.5m and 5mare acceptable.

This type of unit consists of either a back-pack-mounted receiver, linked to a data-logger, to a portable computer, or to ahand-held unit. Mapping-grade hand-held

GPS units are distinguished fromnavigation-grade ones by their level ofsophistication, specification and price tag.Mapping-grade GPS receivers oftenincorporate provision for viewing against a

10


11/24

pre-loaded map background and allowfeature and attribute coding to be attachedto surveyed points, producing plans withinformation attached to componentfeatures ( see below , Surveying with GPS).

This type of receiver can achieve accuracyin the 0.5m to 5m range with the use ofdifferential correction or by postprocessing. Differential correction can beobtained from radio beacons or fromgeostationary satellites, such as EGNOS(see above , Underlying principle 3). Postprocessing can be carried out using datafrom active stations in the National GPSNetwork, with just a few tens of secondsof data recorded at each point. This typeof system is ideal for mapping at scales upto 1:2 500 and recording associated detailsfor database and GIS systems. Becausethis type of system is often used in

wooded environments, they usually havebuilt-in software that helps pick up theweak signals when under trees.

Survey-grade GPSSurvey-grade equipment is the mostaccurate, and usually the most expensiveGPS option ( see Table 3). It is the onlytype of GPS that can produce surveyquality data at the centimetre level and iscomparable to higher order EDM survey.There are two principle methods of

collecting data with survey-grade GPS;static and real-time kinematic (RTK). Allsurvey-grade systems use a differentialapproach to survey processing.

Static surveyingStatic and fast static are the simplestforms of GPS surveying using survey-grade equipment. These post-processingtechniques require data to besimultaneously collected at two (or more)locations and used later in computationsto determine the baseline (or baselines)between the receivers. Baselines are thethree-dimensional distances betweensurvey locations. The time required tocollect sufficient data varies from minutesto several hours. Data must be acquiredfrom a minimum of four satellites, withthe amount of data increasing withbaseline length and accuracy required.

Static surveying is the technique used to tielocal site surveys to the National Gridusing the National GPS Network. With a

DGPS survey the base station is usuallystatic for a long time while the rover-receiver collects data to construct the plan,contour model, etc. The common dataenable calculation of the baselines between

base station and rover points (and thustheir positions in relation to each other) inreal time or during post-processing. Such asurvey could be plotted out, but it wouldnot be related to Ordnance Surveymapping. If the same base-station datawere processed against data from theNational GPS Network active stations forthe same time period, then baselinesbetween them could be calculated (andtheir positions in relation to each other). Asa result, the positions of the base stationand rover points would be fitted accuratelyonto the National grid. Baselines from asurvey site to the nearest active stations areroutinely up to, or more than 100km inlength and require data to be collected forseveral hours to achieve the best accuracy.With dual-frequency receivers, baselines oftens, hundreds or even thousands ofkilometres can be accurately measured.

Fast static is a derivative of staticsurveying, and, as its name suggests goodresults can be obtained more quickly,owing to advances in the hardware andsoftware used. These techniques can beused to measure distances accurate to onecentimetre over a few tens of kilometres.

Real-Time Kinematic (RTK) GPSThis is currently the ultimate in GPSsurveying equipment and techniques

(Figs 7 and 9). The receiver at a fixedpoint, or base station, transmits GPScorrection data to one or more roving units

Fig 9 Typical real-time, survey-grade GPS equipment in use atNorham Castle, Northumberland. In this photograph, therover GPS receiver and data-logger are mounted on the pole(see Fig 7); the batteries, circuitry and aerial (for telemetrylink to base station) are all carried in a backpack. In somemodels, all these are mounted on the pole ( see Fig 19).

in real time via a telemetry link, aVHF/UHF radio or mobile telephone. Thistechnique provides an accurate and flexibleway to carry out land survey on open sites.The rover can be set to recordcontinuously, or by short occupations ofthe survey points; the length of occupationcan be varied depending on the accuracyrequired, normally between a few secondsand a few minutes ( see below , Methods thatincrease the usefulness of GPS). Thesesystems achieve a relative accuracy of oneto two centimetres between the base stationand rover over a range of up to threekilometres with a radio and up to twenty-five kilometres with a phone connection.Because of the high accuracy, flexibility ofrecording at large and small scales, and thetime saved by not having to post-processdata, RTK GPS has shown itself to be themost practical and cost-efficient

methodology for GPS survey in the EnglishHeritage archaeological investigation teams.

It is perfectly possible to combine the collection of data at the base station for staticpost-processing at the same time as it istransmitting correction data for RTK use.

Navigation-grade (hand-held) GPS offers: low cost positioning to c 10m on the

ground

the ability to relocate approximately the positions of monuments from gridreferences

Mapping-grade GPS equipment offers: less complicated and less expensive

equipment than survey-grade less complicated data processing levels of accuracy that are acceptable

at mapping scales data collection tailored to GIS

Survey-grade, real-time, differential GPSequipment offers:

high levels of relative accuracy, and thuseminent suitability to site survey

high level of map accuracy to theNational Grid via the Ordnance SurveyNational GPS Network

rapid collection of accurate threedimentional points on the landscape forground modelling

the potential for creating a digital site plan that can be taken through to CAD & GIS

the potential to locate observed

features accurately from sources suchas historic maps, air photographs orgeophysical surveys, but which arenot otherwise visible in the landscape,and to set these out on the ground

11


12/24

Making choices taken into account when making choicesFor an archaeologist, making informed about using GPS for an archaeologicaldecisions about surveying technology can project. The best policy to adopt is tobe a difficult task. Some might be more think backwards, that is, define the endfortunate than others in having access to a product first and then determine theuniversity surveying department or have a appropriate survey strategy to achieve this.professionally trained surveyor on theteam. However, for many, surveying is still Other factors to consider are:viewed as a skill that can be picked upalong the way by one of the archaeological Do the archaeological unit staff have theteam; after all, some would say, with correct level of training to use themodern electronic kit, all you have to do equipment?is press a button and the black box will Does the unit have sufficient computerdo everything for you. Unfortunately, the power to process the voluminousblack box myth is frequently perpetuated amounts of data that GPS produces?by equipment salesmen their job is to Has the unit got high-quality printerssell equipment and software, not skills and plotters to produce plans or will dataand there has been much hype in recent have to be sent to a bureau for printing?years in selling GPS as the ultimate in Might it be more cost efficient to hirestand here and press the button and an GPS rather than to buy it?accurate survey will emerge technology. Should the work be contracted out to

Nothing can be further from the truth. a GPS surveying company (of whichthere are many) to do the positioning

In the previous sections, it has been part of the work and then do the resthighlighted that the potential with EDM?archaeological user has first to understand What quality control mechanisms willsome underlying principles before be applied?venturing anywhere near the hardware. Is redundancy of equipment andOnce the principles have been grasped, replacement built into the costs?judgements have to be made as to whether (Batteries, cables and other peripheralsthe final product is one that will lend itself can wear out and are costly to replace.to or benefit from GPS data capture, and Repairs can be very expensive.)also, the suitability of GPS to the terrain Does the unit have software to make

and task has to be assessed. full use of GPS data, such as featurecoding, CAD and GIS?A technical paper such as this cannot aim Can the GPS data be easily exportedto cover all possibilities related to GPS into CAD and GIS software?and decision making in archaeological Can the data-loggers be upgraded tosurvey, but Table 3 provides a summary of accommodate changes to the software?some of the basic considerations to be (The GPS equipment development is

Table 3 Criteria for making a choice of what equipment to use

end product map accuracy required

locations to be identifiedby a grid referenceand plotted on a1:10000 base map

objects and monuments to be plotted against anOrdnance Survey 1:2500map, or production ofplan of a similar scale

accurate, measured

survey plan at 1:1 000 andlarger scales, and three-dimensional data

c 10m

c 1m

c 0.10m

solution

navigation-grade GPS (hand-held)

approximate cost in 2003: 100300

mapping-grade GPS; accuracy can beimproved for free-standing surveys bydifferential survey technique and bypost-processingapproximate cost in 2003: 1,0008,000

survey-grade GPS; data available in real-

time or post-processed; c 0.01m relativeaccuracy can be achieved for free-standingsurveys using differential survey techniquesapproximate cost in 2003: 15,00022,000

driven by the surveying and engineeringprofession, and upgrades are frequentlymade to satisfy this market, sometimesat the expense of lesser users.)

Can the data be easily integrated withtraditional data from EDM and otherforms of survey?

Ask the salesmen to provide a list of usersof their equipment so that you can contactthem. (Because the GPS equipmentmarket is so competitive, choosing onemanufacturer instead of another isdifficult. The most honest appraisals ofequipment, support, benefits andproblems come from those who are usingit in an environment similar to yours.)

Many such questions need to be askedbefore deciding that GPS is the wayforward and saying yes to the salesman.As with computers, the technology of

GPS is developing at a fast pace. Newequipment and facilities appear on themarket almost daily. This might be asound reason for only purchasing veryexpensive survey-grade GPS if it is to beused regularly over a long period oftime. If the requirement for high-qualitysurvey is infrequent and limited, thenhiring the equipment or commissioning asurvey company to do the work might bemore cost effective. However, the relativelow cost of mapping-grade and

navigation-grade GPS is probably moreeasily justified for lower-accuracy survey.

For example, take a survey of ahypothetical, isolated archaeologicalmonument on moorland. The end productmight require a detailed survey at 1:500scale (relative accuracy), and an eight-figure National Grid reference for a recordsystem. This could be done withtraditional EDM, and tape measures. Ifonly the centre-point of location is neededto be located on the Ordnance Surveybasic scale mapping (1:10 000 formoorlands) then a hand-held GPS willfulfil that need for map accuracy, as atthat scale a point can only be plotted to c10m anyway. However, if the project ismore complex and will result in a numberof other products, such as:

a detailed 1:100 scale survey of amonument,

a 1:2 500 scale survey of itscontemporary and historical landscape

(which has be to fitted into thecontext of a wider landscape of othermonuments, and which will all besurveyed to similar specificationsover a number of years),

12


13/24

a landscape survey to be fitted ontothe Ordnance Survey digital map baseand with possible long-term furtherresearch through GIS,

establishment of permanent surveycontrol to aid excavation, waterloggingmonitoring, geophysical survey,fieldwalking and similar studies,

creation of a three-dimensional modelof the monument,

then consistency of relative accuracy andmap accuracy need to go hand in hand. Toachieve both types of accuracy using EDMtraversing and tape measures would be timeand labour consuming, tedious and proneto error, while using a hand-held GPScould not provide either relative accuracy ormap accuracy over a large area, with eachpoint being potentially 10m in error.Mapping-grade GPS could not achieve the

accuracy required for the detailed surveyand establishment of control for furtherwork, and therefore survey-grade GPSwould be the answer, as this can fulfil allneeds, as well as being significantly fasterthan traditional EDM survey.

The role of the ArchaeologicalInvestigation teams is to improve theunderstanding of our historicenvironment by identifying, interpreting,

and recording archaeological andhistoric landscapes, monuments andbuildings.This is achieved whereverpossible by working throughpartnerships with other organisationswith an interest in the historicenvironment.The results of the surveysundertaken are used by EnglishHeritage and other organisations toprovide the solid understandingnecessary to conserve, manage anddisplay the archaeological resource.

The work of Archaeological Investigationprincipally comprises field survey thecareful examination and analysis ofphysical remains surviving above ground,and is undertaken in conjunction with

the study of historic maps, documents,aerial photographs and many othersources of evidence, including anyprevious archaeological research andexcavation.The normal products areanalytical reports and surveys, which areavailable to the general public through

the National Monuments Record. English Heritage supply Ordnance

Survey with archaeological informationfor inclusion on its maps for England.GPS is routinely used by the survey

teams as par t of this process.

Surveying with GPS

It is important for any archaeologist usingthis paper to recognise that GPS is such aflexible technique that it can be applied tomost surveying tasks where traditionaltechniques have been applied in the past.Archaeological landscape survey,excavations recording, monitoring,contouring, site planning and topographicsurvey are but a few of the areas to whichGPS can be applied. As well as inarchaeological recording, survey-gradeGPS is now the principal technique usedby surveyors, mapmakers and many otherbodies and organisations who requireaccurate positioning of features, objects orassets in a plan or map-based form. Ofcourse, the ability to make maps and plansis nothing new, and most people inarchaeology have been happily recording

sites and landscapes using traditionalsurveying techniques based on theodolites,EDM, plane-tables, tape and offset, gridsand other equipment. Although it mightbe the latest surveying technology, GPS issimply another option, with its ownadvantages and disadvantages.

Some basic surveying proceduresWhile GPS might initially seem a ratherintimidating package, its use in the field isrelatively simple. As a means of providing

the fast and accurate location of points onthe Earths surface it has no rival whenused correctly and subject to the rightconditions. It should always beremembered that GPS equipment is onlyone of the tools in the survey cupboard,and many sites will require more than onesurvey methodology to produce theaccuracy and level of detail required, forexample reflectorless EDM, taping orother method. Choosing the best andquickest tool for the job is all part of theart of surveying. The basics of surveyingapply to GPS, just as they do to any othermethod of survey (for basic archaeologicalsurveying procedures see Bowden 1999,chapters 4 and 5).

The terrain, site topography andgeography should be assessed to see ifthere are any factors that might affectclear reception of signals from satellites,such as trees, buildings, cliff faces or otherobstructions. Where differential GPStechniques are to be used, a safe and

secure location for the base station shouldbe found. If surveying in real time, thenthere should be no obstructions to radiocommunications between the base androverreceivers. This requirement can be a

problem in very hilly terrain and urbanareas, and in these locations it might benecessary to fix more than one basestation, a repeater station or use a mobiletelephone. If there is a block of woodland,GPS might be used to fix control stationsaround the edge from which to run EDMtraverses under the trees; this samemethod can be used to good effect incomplexes of high walls and buildingssuch as castles and monastic sites.

Integration of feature coding and datasetsfrom GPS and EDM is a normal softwarefacility with most equipmentmanufacturers. For complex earthworksurveys, where subtle stratigraphy doesnot lend itself to rapid electronic featurecoding ( see below ), the positions of pegscan be fixed by GPS and then theresultant plots used for tape and offset

survey (Bowden 1999, chapter 4; 2002).For sites where geophysical survey andexcavation require grid-based recording,GPS can be used as a rapid and accurateway of setting out the grid on the ground,regardless of visible sight-lines. Thecombination of speed and accuracy alsomakes GPS an ideal tool for surveying inintertidal environments (Chapman et al2001 and below , Case Study 5).

Locating the position of an archaeological

feature on an Ordnance Survey map hasalways been one of the most fundamentalrequirements of the archaeological record.In modern computer-based GISenvironments, this has become even morerelevant to modern archaeological recordsmanagement. Whether it be a NationalGrid reference for a single find, or ananalytical plan recording complex multi-period landscape features, the preciselocation of finds and features has alwaysbeen the most difficult objective to achieveusing traditional survey methodology. It isin this area that GPS has been a majorbreakthrough in archaeological recording,and depending on the level of equipmentused (navigation, mapping-grade or survey-grade), objects and features can be given aNational Grid reference in real time and/orbe directly surveyed onto the OrdnanceSurvey map. Positioning is instant, simpleand reliable, and no other surveyingmethod can achieve the same speed,reliability, accuracy or cost efficiency.

Whereas previously, surveying was governedby line of sight along the ground, whether itbe across the landscape or among featuresin excavation trenches, GPS has removedthat constraint and this has made it a much

13


14/24

more flexible tool than traditional EDMsurvey. The only governing factor with GPSis that the receiver must be able to see thesky. This is probably the easiest concept tograsp, although in practice it is not quitethat simple. The amount of sky visible isone prime factor; how many satellites are inthe bit of sky that can be seen is another;and finally, the positioning of the satellitesin relation to each other. Fortunately, theoperators on the ground do not have toconcern themselves with this too much, buthave only to be aware that these factors willaffect the quality of the data and resultantpositioning. This is one of the bigadvantages of real-time survey-gradeequipment, as the operator knowsimmediately whether the data areacceptable or not. In contrast, with postprocessing, awareness that some data hasnot been captured would only emerge later,

and could be costly to put right.

Summary of Ordnance Surveyinfrastructure to aid GPS Surveys

Ordnance Survey basically provides access through www.gps.gov.uk to an array ofservices, utilities and information on how

to carry out surveying with GPS in GreatBritain. A summary of the services offeredincludes:

access to a comprehensive informationfrequently asked questions resource,including an overview of GPS, how touse GPS with Ordnance Survey mappingand a guide to coordinate systems;

access to raw GPS data from anetwork of thirty-two permanent highquality GPS receivers across thecountry; this facilitates the location of alocal base station in the area to besurveyed, and enables the location of

the survey within the Ordnance Surveycoordinate system;

access to the coordinates and stationdescriptions of about 900 GPScoordinated stations across the country;

this also enables the accurate coordinationof a local base station within the surveyarea, but requires two receivers;

access to a coordinate converter facility changing from Ordnance Survey toGPS coordinates and vice versa.

Methods that increase the usefulnessof GPSThe application of differential GPStechniques with survey-grade equipmentprovides for improved relative accuracy,

enhanced map accuracy (using a suitabletransformation) and better absoluteaccuracy than a stand-alone position. Theeffectiveness of survey-grade GPS in thefield can be increased by permutations ofhardware and software applications.Descriptions of some examples of theseapplications follow.

Multiple roversMore than one roverreceiver can be usedwith a base station. Thus, assuming thereare sufficient trained surveyors to usethem, large volumes of data can becollected speedily. It is often efficient tohave two rovers working on a site,receiving correction data from a singlebase station ( see below , Case Study 3).This can be used as either a post-processed technique or in real time (wherethe signal broadcast by the GPS base

stations can be used to provide correctionsto any GPS rover units within range).

Ordnance Survey active stations in theNational GPS NetworkThe Ordnance Survey active stations inthis network are examples of the potentialfor many surveys to benefit from datacollected at a single, static base station,or in this case a network of static basestations covering the entire country. Thedata are supplied in Receiver

INdependent EXchange Format(RINEX) and so can be used in the postprocessing of data collected with anymanufacturers equipment. Informationon surveying with the National GPSNetwork can be found at the OrdnanceSurvey GPS web site.

Ground modellingUsing survey-grade equipment, largequantities of three-dimensional points canbe collected by walking or driving over thesite and these points can then be readilyturned into ground models usingcomputer software. Ground models canthen be used to generate contours, three-dimensional grids as the basis of computervisualisations and analysis in a GISenvironment, as well as enablingcalculation of volumes and analysis ofslope and drainage. Data collection canthen be repeated at intervals to providecomparative data for monitoring sites forconservation purposes to measure coastalor other erosion (Fig 10; see also Figs 17

and 18; Case Studies 3, 4 and 5).

Feature codingIt has to be remembered that although ahigh degree of relative accuracy can be

achieved using survey-grade equipmentand differential GPS techniques, it stillonly produces a series of three-dimensional points. To turn these pointsinto lines, shapes and colours that arenormally associated with a plan or map,they must be combined with featurecoding. Feature-coding software isnormally installed on the data-logger orpen computer and processed throughCAD. With such software, the GPS datacan be converted into a meaningful planof the site. With time and experience it ispossible to create feature codes that willprovide a clear representation of featuresin the landscape, archaeological ormodern, with little need for post-surveyediting. It is rare that feature coding canbe used to record every aspect of acomplicated site, and it would be normalto take the plot back into the field for final

checking and possible additional survey(see Fig 14).

Feature coding is a technique that can beapplied to all electronic survey datacollection, with varying degrees ofsophistication. It involves setting up alibrary of the features to be surveyed;these can be points such as survey pegs,trees, cairns or other fixed points, or linearfeatures such as fences, roads, tops andbottoms of slopes (Figs 11 and 12). As

each point is recorded, a feature code isattached to it by the surveyor, whichdetermines how it will be treated when itis processed. A point may have a symbolinserted on it, or points may be joinedwith a particular type of line. Featurecodes might also have associatedattributes, for example the height anddiameter of a cairn, its state ofpreservation, grid reference and otherattributes. All these data contribute to thepreparation of a site plan or map, and canbe used in a GIS ( see Fig 13).

Real-time GPSThe ability that GPS surveying equipmentaffords, of knowing precisely where in theworld you are at any moment in time, isone of its greatest assets. The GPSreceiver can be connected to a fieldcomputer that has existing maps loadedonto it, so that the relationship betweenwhat is being surveyed and what has beenpreviously surveyed can immediately beseen. On complex sites this procedure

offers the advantage that the surveyor cansee how effectively data coverage is beingachieved by viewing it on the screen as itis being collected. Equally, the position offeatures that are not visible on the ground,

14


15/24

Fig 10 Contour model of the topography of Yeavering Bell, Northumberland derived from survey-grade GPS.The plot shows the spatial distribution of roundhouse sites and their relationships to the topography within the Iron Age hillfort.

Fig 11 Using survey-grade GPS and feature coding to record tops and bottoms of earthworks within an Iron Age/Romano British settlement complex at West Hill, Northumberland(photograph by Alun Bull, English Heritage).

but that are known from historic maps,aerial photographs or geophysical surveys,can be digitised in CAD from the originaldata, transferred to the GPS logger andthen set out on the ground. Grids for fieldwalking and geophysical survey can besimilarly designated on the GPS receiverand set out in the field. This is possible

because real-time equipment provides afacility to give directions on-screen bydisplaying bearing and distance to aparticular coordinated point. Apart fromolder GPS equipment, which was built for

collecting data for post-processing only,this facility is available with all grades ofequipment, making it possible to achieveresults at different levels of accuracy, asdiscussed earlier.

Kinematic GPS (surveying on the move)GPS equipment has the ability to track its

position over time, thus providing theability to produce a continuous plot of itsmovement. This is sometimes referred toas a kinematic survey. This, coupled withfeature coding, enables the surveyor to

record linear features such as fences ortracks by setting the receiver to record itsposition every few seconds, or every fewmetres, and then walking along thefeature. This technique is particularlyuseful for gathering large amounts of datafor contouring and three-dimensionalmodelling, as the surveyor simply keeps

moving and does not have to stop atevery point and press a button. To acquirethis type of data for large areas oftopography, GPS receivers can bemounted on vehicles.

15


16/24

Planning GPS surveys in advanceMost GPS processing software includesfunctions that can predict satelliteavailability at a site at any time of any day.These functions use the broadcastephemeris or almanac (orbital information)data that contains predictions of currentsatellite positions and is received as part of the GPS signal.This can be of greatbenefit when working on sites that aredifficult for GPS data collection becauseof obstructions to sky visibility, forexample close to cliffs or buildings or inwoodland clearings.

Factors that limit the use of GPSDespite the fact that GPS offers manyadvantages to the archaeological site orlandscape surveyor, there are a numberof factors that effect how and where itmight be used:

GPS signals are weak radio signals thatare easily blocked by foliage, buildingsor other obstructions. GPS receivershave to be able to see a minimum of four satellites to work, five to work inreal-time kinematic mode.

Signals might be affected near high-voltage power lines and transmitters andthese are best avoided if possible.Problems have also been encountered

close to airfields and militaryestablishments.

The environment can introduce anerror source in GPS, for examplemultipath. Multipath occurs when thesignals received by the antenna havenot arrived by a direct path but havebeen reflected off another surface,such as a building, foliage or fencing.Multipath cannot be corrected bydifferential GPS as it is specific to a siteand satellite constellation. The antenna,receiver and post-processing softwaredetect and resolve multipath errors,but it might still introduce errorsto a survey.

Other major error sources in GPS aredue to the troposphere, the ionosphere,the satellite and receiver clocks, and thesatellite orbits. Careful survey practicetogether with choice of equipment and

software will limit these effects. GPS satellites operate in circular

20,200km orbits, in six orbital planes.Because of the way in which theseorbits have been arranged, satelliteavailability is always biased towards theEquator.Thus in the UK and the restof the Northern Hemisphere,obstructions to the south of the user,such as steep slopes or buildings, canbe a problem.

16

Fig 12 Feature-coding library devised by English Heritage and used on GPS and EDM data-loggers.

Since GPS reached full operationalcapability in April 1995 the availabilityof sufficient satellites for surveying israrely a problem. Very occasionally,however, satellites are switched off electronically for maintenance, or theybecome unhealthy because there is aproblem with the broadcast signal. Suchan occurrence might create troughs of lower coverage that cannot be predicted.

Real-time kinematic GPS often usesVHF/UHF telemetry to transmit thecorrection data between the base stationand the rover. In the UK the power andfrequency of radio transmissions arecontrolled and they limit the operationalrange to about 3km. Obstructions suchas hills or buildings can also adverselyaffect radio communications. Real-timesurveying systems, however, do allowthe collection of data for post-

processing if the radio link is lost; thismeans that the survey can carry on untilthe link is re-established. This can be acommon occurrence in very hilly areas,but is easily overcome by simply movingthe base station to a more suitablelocation or by using repeater stations.Other delivery methods are available,such as the use of mobile phones; thissolution, however, incurs higher runningcosts.


17/24

Case Studies

The following case studies are used toillustrate some of the ways in which GPShas been used to record archaeologicalsites and landscapes.

Case Study 1 Archaeological survey ofthe Quantock Hills ANOB (Area ofOutstanding National Beauty)Use of navigation-grade (hand-held) GPSto record the locations of newly discoveredsites in Slaughterhouse Combe and SomertonWood (a Level 1 survey as defined inRCHME 1999).

Introduction and objectivesThe Exeter Office of English Heritageundertook a survey of the extantarchaeological sites in the Quantock HillsAONB. The location and distribution of sites

of all types and time periods were within thescope of the project. Most of the sites couldbe recorded using survey-grade GPS orEDM, but the woodlands presentedsignificant problems in terms of survey work.

The wooded combes of the easternescarpment of the Quantock Hills are acharacteristic feature of the landscape.They contain mostly oak, with somehazel and beech. Most of the oak hasbeen coppiced at some time in the past,

but the woodland is not now managed.Two of the major combes,Slaughterhouse Combe and SomertonWood (centred ST 142 401), wereselected for an initial study. There wereno archaeological sites recorded in thearea in either the SMR or NMR. Initialreconnaissance showed significantnumbers of charcoal-burning platformsand building platforms on the valley sidesand valley floor respectively. Locations ofindividual sites or groups of sites wererequired to assess the nature of thearchaeological resource in the woodlandand to facilitate decisions on the type offurther survey work required.

Survey methodologyThe fieldwork was carried out in earlyMarch and mid-April. In March, the treeshad yet to come into leaf, and there werefew problems in receiving signals from fiveor more satellites. Topography was not aproblem during these sessions. In April,however, the leaves were beginning to

appear and this caused some problems withsatellite availability on the valley floors.

The centre of each site was recorded usinga hand-held GPS. Care was taken to

maximise the number of satellites available,and to minimise the error display on theunit. The points were downloaded onto acomputer, then transformed to theNational Grid using the OrdnanceSurveys coordinate converter, available ontheir GPS website. A note of the positionof each site in relation to neighbouringsites and the topographic or map detail wastaken in the field, then sketched onto the1:10 000 base map. The results showed thatthe accuracy of the location of the sites at1:10 000 scale was acceptable in mostcases. Errors were in the order of 1015m,possibly placing a site on the wrong side ofa track, for example. A test, which recordedthe same site on different days, gave adifference in position of less than 10m.

The GPS was used to record the locationof four sites on the open moorland; these

same four sites were then recorded usingsurvey-grade GPS. The coordinates forthese four sites are given in Table 4.Generally, the results were good, withthree of the four sites giving the sameeight-figure grid references. The fourthsite, however, was some 35m to the northeast of its plan position as determined bysurvey-grade GPS. Height differencesranged from 3m to 116m!

Conclusions

From this work, a number of conclusionscan be drawn:

The hand-held GPS proved to berobust and easy to use. It is portable,and stood up to long periods of use inwet conditions.

The results were viable in fairly openwoodland during the winter months.

The results were adequate for plottingsites at 1:10 000 scale, providing thatadditional positional information wasnoted as a check.

Hand-held GPS is an excellent tool forquickly locating large numbers of newlydiscovered sites, enabling a rapidassessment of the resource.

Hand-held GPS with no postprocessing facility should not be used asa substitute for locating or surveyingsites with mapping-grade or survey-grade GPS or EDM.

Case Study 2 Archaeological surveyof DartmoorUse of post-processed and real-time kinematic,survey-grade GPS and differential techniquesto record archaeological landscapes onDartmoor (a Level 2 survey as defined in

RCHME 1999).

Introduction and objectivesDartmoor is the largest expanse of openmoorland in the south of England andcontains large tracts of prehistoriclandscapes surviving above ground. Since1988, the RCHME and latterly theArchaeological Investigation Section ofEnglish Heritage have carried out severalarea-based recording projects on themoor. Work began in the form of a joint

project with the Duchy of Cornwall toinvestigate the entirety of the enclosedland on the Duchys Dartmoor estate. Itcontinues in the form of partnerships withthe Dartmoor National Park Authority,the National Trust and the Ministry ofDefence, among others. By 2002 some240 square km (35% of the areaadministered by the Dartmoor National

Table 4 Coordinates of open moorland sites

(All coordinates are in metres; eastings and northings are on National Grid (OSGB36

) andheights are above ODN.)

point

Cairn 1, Higher Hare Knap ENH



Cairn 1, Black Hill ENH

hand-held GPS(navigation-grade)

314995.485139366.601290314715.812139494.827282314757.573139200.601283314560.191138199.685237

differential GPS(survey-grade)

314996.104139360.813307.512314717.955139490.712293.649314734.764139172.357279.600314557.655138196.889353.154

positionaldifference (m)

0.6195.78817.5122.1434.11511.64922.80928.2443.4002.5362.796116.154

17


18/24

Park Authority) have been recorded. Atotal of about 2,500 monuments havebeen surveyed.

The enormity of the archaeologicalresource had been alluded to by previousfieldworkers, although its true extent onlybecame apparent after the first season offieldwork. The range of monuments thatwas recorded was expanded from thepreviously traditional concept ofarchaeological categories to include suchdiverse items as slotted gate-posts (goodindicators of pre-19th century enclosure)and military observation posts and targetsdating from the Second World War. Theextent and importance of the industrialarchaeology, mostly relating to theremains of the early tin industry, but alsoincluding gunpowder mills and early 20thcentury clay works, had also become

apparent over the years.

Survey methodologyIn the early days of the project, only apost-processed, survey-grade GPS facilitywas available. Since then, new equipmenthas been developed enabling real-timekinematic techniques to be employed.This has formed the backbone ofsubsequent projects and is used as thetool of first preference. It providesaccurate three-dimensional points relative

to each other and to the National Grid.Feature coding has enabled theinvestigators to assign a number oflinetypes, colours and symbols to theindividual points. When processed withinCAD software these create satisfactorydepictions for most monuments, withlittle need for post-survey editing. Whilethis system of feature coding wasdeveloped specifically for the archaeologyof Dartmoor, most elements are of use inother upland areas.

The discovery and interpretation ofmonuments was supplemented by theiraccurate location a process not alwaysstrictly observed by earlier workers which in itself led to the creation of a newsurvey methodology applicable to mostupland areas. From the outset monumentswere located using EDM within theNational Grid (OSGB36 ) framework,which, while a relatively easy process onthe enclosed fringes and central basin ofthe moor, became more time consuming

in the more remote parts.

Perhaps the most ambitious projectundertaken during this period was acomplete survey of the open moorland

Fig 13 Typical HSIS window; each monument is linked to a text-based record.

owned by the Duchy of Cornwall. Thisinvolved the investigation of some 140square km centred on the highest parts ofthe north and south moors during athree-year period. While the archaeologyof these areas was relatively sparse, muchwas concealed from general view innarrow valleys and on broad ridge crests.

EDM survey became increasinglyarduous and inappropriate. By thebeginning of the third season, when GPSbecame available, only the most accessible80 square km of the designated area hadbeen investigated. The use of GPSenabled the remaining area to becompleted within nine months. It provedthe ideal tool in this environment,providing the required accuracy andunhindered by the line-of-sight problemsthat beset EDM survey.

ConclusionsBy-products of this digital process areseamless maps of the surveyed areas at1:10 000 and, since 1996, 1:2 500 scales.The results of this project have overcomeearlier problems of piecing together largehand-drawn surveys. For the first time it ispossible to examine large archaeologicallandscapes in sufficient detail to beginexplaining, rather than simply describing,the archaeology.

Additional value is added to thesesurveys by their insertion into GIS,whereby each monument depiction islinked to the relevant text-based record(Fig. 13). The principal archive for this

work is the Heritage Spatial InformationSystem (HSIS) operated by EnglishHeritage. HSIS graphical data can easilybe translated into a variety of outputs tosuit the needs of both national and localorganisations, and also the various typesof GIS now in use. The HSIS dataset forthe parts of Dartmoor so far surveyed is

being integrated into the GIS of theDartmoor National Park Authority foruse in the day-to-day management andconservation of the moorlandarchaeology. This accurate depiction anddescription of the archaeology of themoor is also helping the Department ofEnvironment, Farming and Rural Affairs(DEFRA) to evaluate the archaeologicalresource when compiling

Documents

Where on Earth Are We