Methodology

Gary Nobles

Methodology

3.0 Methodology There were various methodological complexities within this study; this is mostly attributed to the variety of computer and software technologies employed. There was a need for two methodological approaches, one for the questionnaire and the other for the quantification of the horizon.

3.1 The Questionnaire To achieve a definition for circularity an online interactive questionnaire was conducted using a multimedia approach. In order to construct this questionnaire various technical processes had to be put in place, these included: •

The creation of an Apache Server

•

Creation of a MySQL Database

•

Configuration of Php with the Server and MySQL

•

Manipulation of a series of photographs to form a panorama

•

Converting the panorama into a Flash based 360˚ viewer

•

The creation of a website using HTML and Php.

The Php served as a communication device adding the data from the html forms into structured query language (SQL) for the database. The forms allowed for the user to submit their name and email into an unrelated table for a partial dissemination of the results with complete anonymity in relation to their answers. The remaining forms utilised a single table, the structure of which can be found in appendix I. Firstly the user was asked their interpretation of the meaning of circular, was it: o Like, involving, resembling, shaped like, relating to, or in the form of a circle

21

Gary Nobles

Methodology

o Traveling, occurring, or recurring in a cycle; or ending in itself; roundabout; circuitous o Both o None of the above Following this the user was presented with a series of 360˚ panoramas, these panoramas were created by the author using a 10 mega pixel SLR camera and a tripod, the tripod was stationary and a series of photographs were taken around this point, an overlap of at least 50% was maintained. These photographs were stitched together using panorama software (hugin with a PT sticher from pano tools available under gui licence at www.sourceforge.net). These panoramas, once created were fed into a flash 360˚ viewer which converted the panoramic image into an interactive dynamic image. The resolution of the initial image had to be reduced by 90% in order to reduce download time from an hour to a few minutes; it could not be reduced further as this would have resulted in a vastly distorted image. The locations were chosen largely by examining a map of Dorset and selecting areas which had a variety of attributes. Some were chosen due to the horizon’s uniformity, proximity, variance in elevation and irregularity. Dorset was chosen as a location due to a variety of factors; the author was born there and has spent the majority of his life there, therefore he has a familiarity for the landscape. Dorset also has large areas of agricultural landscape and is largely rural; this provides open areas with minimal disturbance from vegetation and modern constructions which could detract from the far horizon. Once the user had experienced the landscape through this media they were asked which ones they perceived to have an impression of circularity in relation to their chosen definition of circularity. They were only allowed to answer yes or no to avoid the problem of ambiguity. The next three sections displayed the resulting attribute of the previous landscapes, inclination, distance and elevation, and the same question was posed. To avoid the association of these

22

Gary Nobles

Methodology

graphs with the previously viewed panoramas they were displayed in a different order. The final section displayed plans of the areas displaying only the horizon and viewing point, continuing with the same question. The user was also allowed to comment on the questionnaire, this identified a big problem; a large proportion of the users voiced concerns about the length of the questionnaire and its complexity, this was backed up when looking at the user statistics, after 500 unique hits on the website only 33 people had fully answered the questionnaire, many of the users had entered their details but failed to complete it fully. As a result the questionnaire was reduced from the 41 questions to only 9; this left only the definition of circularity and the panoramas. The answers gathered from this questionnaire aimed to establish the current impressions of a circular horizon, and to have a landscape which could be quantified in relation to inclination, distance and elevation of the far horizon. The method for obtaining these quantifications is described in due course within this section.

3.2 Digital Elevation Model Three Digital Elevation Maps (DEMs) where needed for this study, the first and largely unrelated DEM was of Dorset. This served to provide the questionnaire with various graphs of inclination, elevation and distance of the horizon, as well as providing a method for quantifying the circularity of the horizon, as described in the previous section. The vast bulk of the analysis was aimed at two DEMs, those of South West Ireland and North East Scotland. Both the Dorset and Scottish based DEMs were obtained through the EDINA (www.edina.ac.uk) web service which provides OS land-form profile digital terrain model at a resolution of 10km for Britain in conjunction with the British Ordinance Survey. The Irish based DEM was obtained through the Manchester University based Landmap web service (www.landmap.ac.uk) which provides a DEM of Ireland in 100km tiles at a resolution of 25m. All of these maps contained various artefacts through the Interpolation method; this over representation of contour lines could not be resampled or smoothed through a low filter due to the nature of the study. A re-interpolation would invalidate the DEM as the artefacts would be also resampled and a filtering of these artefacts would run the risk of chopping the tops off of

23

Gary Nobles

Methodology

peaks and ridges which are likely to be restricting and therefore creating the horizon. The Scottish DEM was 120km by 100km, this was an appropriate size as too allow the containment of the viewsheds of the monuments. The Irish DEM was unavoidably larger due to its delivery in 100km tiles; this DEM was 200km by 200km which was more than sufficient to hold the viewsheds. Edge affects for the analysis was mitigated by utlising a DEM which held an area which was of sufficient size.

3.3 The analysis The analysis was conducted within the open source GRASS GIS package, the main reason for this choice of software was due to the development of the module r.horizon which allows for the calculation of values for the inclination, distance and elevation for the far horizon from a specific point, the viewing point. The module r.horizon is not contained within the standard GRASS GIS package, it was developed by Mark Lake at University College London and is mentioned in Lake and Woodman 2003 from which this study is drawn. It draws upon an existing module r.los; it takes the resulting inclination data from the line of sight and calculates the far horizons values from this. Table 2. Parameters used for r.los and r.horizon r.los Description coordinate Position of observation point max_dist Maximum distance of analysis obs_elev Elevation of observer

Parameter x,y 60,000m 1.75m

r.horizon coordinate max_dist

Parameter x,y 60,000m

Description Position of observation point Maximum distance of analysis

This data is delivered within a text file with a value for each azimuth and the resulting values for inclination, distance and elevation, upon this data a Monte-Carlo test and Standard Deviations are conducted upon it, for a more detailed explanation of each analysis see the appropriate section.

3.4 Dissemination

24

Gary Nobles

Methodology

This document is the main method of dissemination, it will be available through UCL and the author, also the digital data is provided on the accompanying DVDs/CDs. These DVDs/CDs have an autorun feature which automatically launches a html browser which contains a contents of the DVD/CD with links to the appropriate folders. Rather than just having the data accompanying this document it is also planned to have this available at www.noblesweb.co.uk and through the Archaeological Data Service (ADS) based in York in due course.

3.5 The Contents of DVDs/CDs Disk 1: Documents (CD) •

Front Page.doc

•

Abstract.doc

•

Contents.doc

•

Aims and Objectives.doc

•

Background.doc

•

Methodology.doc

•

Analysis.doc

•

Discussion.doc

•

Conclusion.doc

•

Appendices.doc

•

Bibliography.doc

Disk 2: Data (DVD) •

Ireland Profile Data

•

Scotland Profile Data

•

Random points Profile Data

•

GRASS ASCII files

•

ArcGIS files

•

Profile Data (Combined into a spreadsheet)

•

Data Analysis (Spreadsheet) 25

Gary Nobles •

Methodology The Questionnaire Files o PHP files o HTML files o Image files

26