The Discovery Programme collaborated with UCD School of Archaeology on a Heritage Council funded project to conduct advanced studies on the remains of wooden fishweirs exposed on mudflats of the Fergus Estuary near Boarland Rock. A part of the project aimed to assess the value in recording such complex and fragile remains at high resolution in 3D, generating data sets which could then be examined and studied in greater detail off-site. The wooden remains, situated on thick mudflats in the middle of the estuary, are only exposed during periods of extreme low tides and even then for a maximum of only 3 hours a day. Access to the sites had to be carefully planned, hiring small flat-bottomed boats and local boatmen with the detailed knowledge and experience of the tides and currents to land us and our equipment safely on site.
The aim was to concentrate the laser scanning on three of the best preserved fishweirs while simultaneously conducting a DGPS survey of the other remains, simply recording a point on each upright post. By the end of the four field days the intention was to have a plan overview of the currently exposed timber remains, with high resolution 3D data from our laser scanning for three examples.
Instruments and software
The laser scanner used on the project was a Mensi GS101 time of flight scanner. The instrument was powered by a Kippor Sinemaster 1000 petrol generator and controlled by Trimble PointScape 3.2 hosted on a ruggedized Panasonic Toughbook CF19. Processing was done using Trimble Realworks 6.5. DGPS for both georeferencing the laser scans and surveying the wider area fragments was done using a pair of roving Trimble 5800 receivers with a local base station provided by UCD, again a Trimble 5800. Due to the hostile environment (mudflats submerged by the incoming tide) a new base station had to be established each day, its position fixed in Irish Grid coordinates using the Trimble VRS NOW correction service via a GSM mobile phone link.
Why was this method of 3d data generation selected?
The subject matter, a complex arrangement of decaying wooden posts, was inherently three dimensional in character. In their decayed state few uprights are truly vertical and a complex collapse of wood and wattle remains form a subject matter for which a simple 2D plan would be an insufficient record. As the site is extremely fragile it would be been unsuitable to utilise any technique which required the surveyor to be walking extensively amongst the remains. Laser scanning offered the opportunity to scan from a sufficient distance with a non-contact approach. Laser scanning is also relatively fast, important considering the extreme lack of time on the site due to the constraints of the intertidal location.
What problems were encountered?
A number of issues were encountered which had not been anticipated. Our reccconnaisance visit to the site coincided with spring tides giving a misleading impression of how much of the structures would be uncovered, and for how long during the planned fieldwork date. This may have been compounded by a combination of wind strength and direction but the outcome was that some of the ideal set-up locations for the scanner identified in the recce were inaccessible, never exposed by the tide. The result was scans taken from acute angles to attempt some data capture at the estuary side of the structures.
Establishing a stable tripod set-up for the scanner and keeping cables free from mud and sea water was a challenge. Great care was needed to ensure that the scanner did not move or ‘sink’ between scanning the targets and scanning the object. This was identified by a mismatch when inspecting a part of the data set and had to be corrected by a re-registration using the cloud registration option.
A final problem which we hadn’t encountered in scanning projects previously was an issue of reflected data. In areas where there was extremely still water or a perfectly flat mud surface a mirror image, the reflected object, was recorded. This was solved by a manual edit of the pointcloud, segmenting the shadow data points from the main registered project.What were the final deliverables?
A georeferenced pointcloud dataset, segmented into the three structures recorded, was the primary deliverable of the project. This was supplied to the clients, (UCD School of Archaeology and the Heritage Council) along with the instructions to download free viewer software (Trimble RealWorks Viewer) giving them the ability to view and manipulate the 3D pointcloud in the native RealWorks environment.
A range of GIS compatible outputs including plan view orthoimages and derived shapefiles were also generated to be incorporated into the wider area DGPS survey. These GIS products will provide a useful framework for the further scientific analysis taken place on the fishweirs e.g. wood species and date analysis. Some experimental 3D reconstruction with extruded posts based on heights measured from the pointcloud has been undertaken.
The deliverables have been well received by the client and they have been using the Realworks viewer to further analyse the structural components of the fishweir sites. However, critically assessing our results it is clear that some compromises resulted from the extreme time constraints working in such an environment. A scan resolution of 10mm was selected as our target for overall coverage i.e. no part of the structure should have point spacing in excess of 10mm. Any finer resolution would have required considerably more time on site than was available (or a smaller area covered), but for such subtle and complex structures the result proved to be coarser than we would have liked. Amongst the remains of the fishweirs some wicker fish baskets could be seen in the mud but these were not discernable in the pointcloud. If the site had been easily accessible, with no constraints on time then we believe a resolution closer to 2mm would’ve been appropriate and attainable.