Methods

Fieldwork seasons 2005-2007 collected data for a high-precision topographic model of the area from the westward boundary of the estate ('Villa del Confine') as far as the eastern limit of the vicus Augustanus. A second topographic model has also been initiated in the area in and around the so-called fish farms, together with a series of fixed points along the intervening 1.2km stretch of the Via del Telefono.

The model aims to map both the archaeological and natural features of the present landscape and will form the basis for both the more detailed documentation and analysis of Roman occupation along the ancient sea front (c.200 BC - c.AD 500) and the study of its environmental and geomorphological context (being carried out by Prof. Helen Rendell, Loughborough University).

It is intended to use the model to explore and explain how the architecture of the Roman villas, the associated vicus and a possible harbour responded to the continuing evolution of the coastline and to the environmental changes associated with this development.

Since the site is heavily forested, the most feasible way of recording the surface co-ordinates is using a total station and a prism. The dense foliage impedes the use of a reflectorless laser setting for the total station, differential global positioning system and also remotely sensed images. Instead a traditional topographical survey is conducted where one person operates the total station while a second person positions the prism on the surfaced to be surveyed in a grid-like formation with 0.3-5 metres intervals between each measurement. The Total Station employed is a Leica TPS400.

The area surveyed around the vicus Augustanus presently covers 660 by 290 metres, within which around 19,750 points have been taken and used to create a three-dimensional digital elevation model (DEM) in the geographic information system ArcView. The DEM has already highlighted several areas which require further investigation by other means, such as test trenches, geophysical survey and geomorphological coring. The whole model in its current state is reproduced in Fig. 1.

The second topographic surveyed around the so-called fish farms, including a 450 metres section of the Via Severiana, covers 450 by 150 metres, within which 1475 points have so far been taken and used to create a three-dimensional digital elevation model (DEM). The topographical model of the fish farms in its current state is reproduced in Fig. 2

The total station was not used for surveying every structural feature since these were drawn by hand and, as they eventually also had to be incorporated into AutoCAD, this in effect meant doubling the amount of work. Instead only a limited number of total station point shots were taken at easily definable points in order to link the drawings in AutoCAD.

At home the hand drawn plans were imported into AutoCAD, together with the coordinates of the known points. By scaling the drawings, and then shifting and rotating them to the actual coordinates they were then geo-referenced correctly in 3D space. This system proved sufficiently accurate and more time saving than digitising the entire site by traditional total station survey.

Once the drawings have been digitised they are linked into our geographic information system for further analysis together with the DEM model, the geophysical, geomorphological and palaeoenvironmental surveys.

The geographic information system employed on Laurentine Shore Project is ESRI's ArcView. ArcView is used because it does not require massive computing power in the field, it is also easy to use and it does almost everything required for this project.

The following is a brief summary of the method of the geophysical survey undertaken by Timothy Evans (Birmingham Archaeology Unit) during the April 2006 season. The geophysical work took the form of a resistivity survey using the Geoscan RM15 module with a 0.5m PA1 twin array. All results were processed using the associated Geoplot 300mx software. With all the data, the standard processing for resistivity data as recommended within the Geoplot manual took place:

• A clip to 3 standard deviations off the mean average.
• A high pass filter using a gaussian filter.
• A de-spike process at 3 points off the median average.
• For presentation purposes the data was subjected to the Interpolation procedure along the Sin/x and y-axes.

The survey of Area B" (SS3) covered an area approximately 180m x 20m over the location of a known portico structure. The area was divided into 10m x 10m grids and traversed at a 0.5m sample interval on both the x and y axis, thus providing a maximum total of 400m per grid. This strategy was adopted in view of the suspected width of sub-surface features (less than 1m). The output current was constantly 1MA and the gain setting adjusted to x10 in order to compensate for a background resistance of less than 100 Ohms.

Initially results proved disappointing, with blanket areas of high resitivity masking any discernible architecture within the dataset. However, knowledge of the site formation processes derived from previous trial excavations proved invaluable. A large proportion of the survey was carried out on a raised area comprised of significant depths of wind blown sand. Analysis of the numerical data showed that this type of background drift geology was providing results +200% above a mean average (c. 70-74 Ohms) and thus introducing a massive bias into the visual data. Even high resistance features such as tufa and brick walls should only vary a maximum of 20-30% away from the mean average. Subsequent processing and re-evaluation of the data set showed areas of anomalies forming right angles of relatively lower resistance (80-100 Ohms) within these areas of massively high readings. One of the current theories in use is that both during the lifetime and after the abandonment of the site large deposits of windblown sand accumulated between the standing walls. The overall effect of this is to produce a 'negative' image of the relative resistance-the opposite to what one would normally expect. Combined with the survey DEM and small scale excavations the geophysical data confirmed the existence of the portico structure, with an apparent 90 degree turn towards the area of the ancient coastline. Furthermore, possible architecture can be seen abutting the possible portico to the Northwest, perhaps linking to the villa site further along the coast.

An area 20m x 20m was surveyed using the same settings as on area SS3, with the area again split into 10m grids. The results appeared to confirm the trend noted previously, with potential architecture only providing a relatively weak response above the background mean. Further explanations for the lack of a strong contrast could lie with the sandy and well drained nature of the subsoil layers. This area was also surveyed with a 1m-twin PA5 array to provide a deeper depth of reading (max 2m) to compare against the shallower PA1. Results showed the same trends but with slightly less clarity, perhaps suggesting a limited archaeological presence at this depth.

The 2006 survey season demonstrated that a densely sampled resistivity survey can be effective on the subsoil conditions particular to the Castelporziano estate, especially when combined with the detailed DEM and small scale excavation.

Peter Rose also collected data for experimentation with digital photogrammetry on the archaeological features excavated during the April 2006 season. He has previously employed the method at Ostia in the modelling and analysis of standing buildings where the approach has offered great advantages over traditional building survey and recording systems. It permits an objective visualisation of the structures in their present state; reconstructions can be produced from these visualisations and used in more detailed analyses. The approach has proven to be both an accurate and easy way of capturing a large amount of data in a short period of time. The result of the photogrammetric image of Trench 1 H4 is illustrated here (Fig 5).

The following is a brief summary of the methods for the Geomorphological and Palaeoenvironmental studies undertaken by Helen Rendell, Fiona Brown and Andrew Bicket during the April/September 2007 seasons. The aim of this investigation is to develop a better understanding of the geomorphology and environment of the coastal zone of Castelporziano before, during and after the period of Roman occupation of the site. A number of cores were extracted during the two seasons. Three sets of 2.0 m cores were recovered from the central part of the so-called fish farms using a Russian peat corer. One other core was extracted from a more marginal, and drier, location using a percussion corer. A series of cores through interdune sands were also taken along a transect running inland from the Vicus site and Tor Paterno. A detailed contemporary vegetation survey was undertaken along the coastal strip between the Vicus and Tor Paterno (Fig 6).