#CAAPerth Day Three

I am jumping between sessions today.

<live blog>


Off to chair session S30 – Computational approaches towards artefacts studies (on behalf of Eleni Kotoula). Session starting at 11:00.


Examining spatial relationships along the street front.


Explored overlapping isovists to explore movement around the city, and visual overlaps in order to create visibility connections. <@tombrughmans work on analysis of visibility networks that was presented downstairs would be of interest here.


Recent laser scanning of site will further add to the possibilities of revisiting the site and also analysing its buildings. Now introducing Scalar Space Syntax – examining networks of connectivity in individual houses, an urban neighbourhood, and the whole street network.


Very little record survives of the 1938 – 1942 excavation and reconstruction of Ostia. This makes analysis hard. Interesting history of development of mapping. Originally focused on isolated, important buildings. Wider environment only considered later – in aprt tied to technological advances and opportunities e.g. vertical photographs from 1940s. 1990s first map of wider Ostial environment – the DAI.


Next up  H.Stöger – Roman Ostia: interpretation and understanding: what to do with nice maps? (Note: CAA 2013 programme is here).


Learn more at http://www.unicaen.fr/cireve/rome/index.php


Moved to S14 – Beyond Virtual Reality: documentation, interpretation and understanding of ancient cities. Paper by P. Fleury on A Roman Street at the Time of Constantine : Interactive Visit with Access to Ancient Source Materials. Emphasising need for authentication and transparency.


Transport networks do not seem to explain visibility networks. Sites with origins ion the Iron Age do not seem to act as hubs but visual control being significant. It seems that sites were taken out of the equation that sued to be hubs in the Iron Age. Ends by demonstrating need to address assumptions in detail in network analyses, and that robust techniques such as ERGMs help with this.


Summary: there is a clear difference between the Iberian and Roman networks. The networks of visibility fragment over time. Long distance decrease but local visibility increases. Sites with urban status tend to be less prominent. ERGM results – 50 million random networks generated per observed network. 1000 samples. These are very different from the observed ones – to be expected because site location is complex, and clearly non-random. For all periods reciprocity is significant and positive – propensity for inter-visibility. Significant all-in-star effect in Iberian, Republican and Early Imperial periods i.e. the visibility of a given site from others. Outgoing links (all-out-stars) also significant.


Probability viewsheds allow us to generate probability networks. Global measures of these networks – numbe rof nodes, arcs, clustering coefficient, average degree, connected components etc.. Tom will then look at all the components of the network and then generate random networks and compare them to each other – Exponential Random Graph Models (ERGMs). The random networks can include a hypothesis i.e. where a certain configuration is more likely e.g. more visible. Tom uses the analogy of a box of lego – you have the pieces but you have lost the instructions for a particular model. You keep putting the blocks together randomly. It is unlikely that you will end up with the planned model. But what if you start the process knowing that you want to end up with a particular building – the chance is much larger that you will end up with the hypothesised model based on dependent assumptions. This gives a sense of the processes that may have led to the observed distribution.


Discussing visibility analysis and the chronology, and also the attributes of each site e.g. urban status, road and river networks, etc. What can we do with these data? We can produce probability viewsheds, in this case via 100 binary viewsheds with random error in DEM.


Tom is interested in the processes that make these visibility patterns emerge. First endogenous processes e.g. extreme case where visibility is the only factor in site location (this is clearly not realistic but required to critique assumptions) and second exogenous processes – the influence of other site characteristics.


Tom introducing the AHRC Urban Connectivity Project from which his data derive. 195 sites from Iberian to Late Imperial period in the study area in Baetica, southern Spain. Examing the formal metrics of visibility and also the ones gathered on each site, (and hence taking more account of subjective factors and issues of weather etc.). Limited emphasis on intervisibility in the Roman research in the region – much more on networks of connectivity (roads and rivers) and also questions of urban status, hierarchy etc. In the Iberian period intervisibility (e.g. between lookout points) has been considered. Given the continuity of occupation visibility in the Roman period seems to be a useful focus of study.


@tombrughmans introducing his talk and emphasising how we go from exploratory to confirmatory techniques – dealing with the lack of neutrality and the dynamism of the interpretations and the data underlying them.


Moved through to S9 – Complex systems simulation in archaeology. Watching a paper by Tom Brughmans: When exploratory network analysis becomes confirmatory: exploring urban connectivity in Iron Age and Roman Southern Spain


Final case study is Tewkesbury. Using structural analysis to simulate buildings (Abbey Mill and 1 Mill Bank) based on AutoCAD building surveys. Using finite element analysis using Autodesk Simulation (formerly ALGOR). This takes account of material properties such as compressive strength. Example results – 1 Mill Bank can see displacement of localised areas. Ongoing work includes further FEM, thin sectioning of bricks and studying fabric of earth buldings.


Second case study is Odda’s Chapel at Deerhurst. There has been installation of remote climate modelling instruments here e.g. temperature, relative humidity, wind direction and wind speed. These data are captured and available in real time. Again photogrammetry and ERT have been used to examine absorption, in the context of the continuous weather readings. Also tried using the ERT on a horizontal plane – this was able to show the different material properties and their variable absorption rates.


Also undertaking photogrammetry as a means to monitor errosion of buildings over the study period. James introducing one of the case studies – Winchester Cathedral crypt. This floods every year. The photogrammetry has been used to check volumetric record of the building, as a means to contextualise the ERT results. Need to do more work comparing the ERT data to the lab tests conducted under control conditions by the project.


Parnassus Project has included a stone testing programme – testing absorption rate for stone such as Cotswold Stanway Stone and Wadhurst Stone. Also using Electrical Resistance Tomography to record water ingress and absorption in surviving stone buildings. Using up to 64 probes. The Southampton method has been non-destructive i.e. no drilling of holes – instead a medical paste is used to attach the probes.


James introduces simulation and analysis of wind driven rain impact on building materials. The Parnassus Project website has further details. Now discussing compressive strength of simulated medieval building materials and structural models.


In S12 – Three-dimensional computational analysis and simulation in archaeological research. @james_e_miles is talking about The Parnassus Project: Archaeology and Engineering collaboration for 3D data collection and analysis