Tamworth Castle: survey enhancement

Introduction

Tamworth Castle is a Grade I Listed Building now housing a museum. Originating as a late 11th century motte and bailey castle it has undergone several phases of repair and expansion. The castle is sited in the Pleasure Grounds close to the town centre of Tamworth and is a popular tourist attraction.
 
Wessex Archaeology were commissioned by Tamworth Borough Council to undertake an enhancement project of an existing metric survey of Tamworth Council. The existing survey was undertaken by NGM Surveys Ltd and provided architectural style outputs comprising elevation drawings and plans. Whilst being of the highest quality, these were of limited use for archaeological work as architectural drawings typically do not contain enough detail of the historic fabric; only the outlines of major features such as walls, doors and windows are included with no representation of the stonework itself. Our job was to produce stone by stone drawn elevations of the castle to the highest level specified in the English Heritage Metric Survey Guidelines.
 

A tricky proposition

The original specification to enhance the existing survey with the requisite level of detail suggested the use of photogrammetric techniques but it was quickly realised that to capture the necessary photography and survey control for such work would be unfeasible. As the original survey drawings were based on a point cloud created by terrestrial laser scanning (TLS) techniques, Wessex Archaeology suggested reusing this existing scan data to create stone by stone drawings of all faces of the castle, based on our experience with similar projects.
 
Existing scan data was limited in that it was captured using a phase scanner which captured no colour information. Drawing individual stones from the scan data alone was in many cases impossible, even using advanced visualisation techniques to enhance the surface detail as much as possible. The limited access available to scan the external walls meant that a very short standoff was necessary and the combination of scanning from positions close to the walls combined with protruding features such as windows meant that there were considerable areas of shadow (i.e. no data) on the upper surfaces of the walls. The oblique angle onto the upper surfaces of the walls also resulted in a much lower point density on these upper surfaces (laser scanners have a fixed angular precision so the laser beams are fired at fixed angular intervals resulting in a variable density of points on the subject depending on how far away the subject is and the angle of incidence of the laser beam onto the surface).

The sheer volume of TLS data posed issues for data processing, manipulation and visualisation. Comprising over 1.4 billion individual measurements, the dataset is certainly one of the larger datasets produced as part of a heritage project.

203

A novel solution

It was looking like a complete resurvey would be necessary to meet the project specifications but any survey would suffer from many of the same problems due to the shape and position of the castle walls. Photogrammetry would be too costly and rectified photography would not work due to the curved surfaces of the walls. After a bit of lateral thinking, a combination approach was opted for which would use the existing point cloud to provide a metric framework but supported by a photographic survey to provide the additional detail required to allow stone by stone drawn elevations to be produced.
 
Traditional photogrammetry involves very carefully controlled photography and survey control but recent developments in survey software now provide a means to integrate any photography, even historic photography, with any survey control, including laser scan data. Our solution was therefore to undertake a photographic survey to provide a visual record and combine this with the TLS data to allow us to produce stone by stone elevation drawings. In addition to photographs taken from ground and from vantage points on the castle, photographs were taken from elevated positions using specialist kite and pole based systems to provide a good record of the upper surfaces of the structure.

204

The drawing up process then used the TLS data and the photographic record combined to produce a 3D wireframe model with each stone drawn using the visual detail from the photographs, metric data being provided by the TLS data. Orthographic views onto this model then formed the basis for scale drawings of each elevation.
 
Working with the data was less than straightforward given its sheer volume. Thankfully, our dedicated GIS/Survey custom built workstations are designed for this kind of work and whilst standard laptops and PCs floundered, the specialist workstations were able to manipulate, process and visualise the data easily.
 

A combination approach

In addition to the drawn elevations, the causeway leading to the castle needed to be recorded. Unfortunately, this was not surveyed as part of the original TLS survey. As such, there was no metric framework to allow us to use the same methodology as was applied to the rest of the castle. Instead, rectified photography was used to provide a visual record that can also be used for measurement.
 
Also, a set of orthoimages produced directly from the TLS data were produced. These look very much like photographs, albeit black and white in this case as the TLS data contained no colour information.
 

Conclusions

With the right set of skills and tools, a great deal is possible, even with existing data that may not have been captured to ideal specifications. This is the first time the technique of combining TLS data with photography in this way to produce detailed stone by stone elevation drawings has been applied on this scale and the methodology developed for this project is a very good and cost effective alternative to photogrammetry.
 
This project exemplifies the complexity of many archaeological survey projects; rarely is survey in the heritage sector straightforward and a team comprising geomatics and archaeological specialists is often better placed to make the right survey decisions to get to the desired outcome.
 

Acknowledgements

 
Bill Blake Heritage Documentation – kite and pole based photography.
 
NGM Surveys Ltd – original survey work including TLS data capture.