On site at a Roman Fort – the challenges of new tech, volunteers and a fast flowing river...

I run a facility in the Department of Archaeological and Forensic Sciences called ‘From Land To Sea’. This is a geophysical prospection ‘laboratory’ that uses variety of tools to look at landscapes, archaeology and people, especially in environments where water is a crucial factor such as the coastline, wetlands and river corridors. The facility (and my job) is funded by the AHRC as part of the Research Infrastructure for Conservation and Heritage Science (RICHeS) programme. RICHeS provides 31 facilities across the UK where people involved in heritage and conservation can access cutting edge research tools. Costing £80M, RICHeS is one of the largest heritage science programmes in the world. 

RICHeS run an Access Fund to facilitate people outside of traditional research environments to use these facilities and gain training and experience. We were lucky to be part of the pilot round of funding in a partnership with the Trimontium Trust and AOC Archaeology. The Trust and AOC were able to engage a team from Bradford to go up to Trimontium in the Scottish Borders for a 10-day data collection and hands-on training mission in February 2026. Trimontium is a Roman Fort on the banks of the River Tweed just east of Melrose and the village of Newstead. First excavated at the turn of the 20th Century, the site was a major military installation on the route between Hadrian’s Wall and the Antonine Wall, and its apparently hurried abandonment resulted in an incredibly rich suite of finds that is some of the best evidence about Roman Cavalry units anywhere in Europe.

We had three goals for this mission – firstly, to use specialist airborne geophysical tools to find the Roman crossing point on this stretch of the River Tweed. Secondly, to complete a survey (started in 2022) of the fort interior and immediate environment using multichannel ground penetrating radar (GPR) to achieve a high resolution three dimensional image of the buried archaeology. And finally, to train local and regionally based volunteers in the use of geophysical prospection for archaeological research. 

Using Ground Penetrating Radar in different ways to survey the site

The first five days on site were largely spent ironing out the method for the data collection with the team from AOC Archaeology, a UK-wide commercial company providing archaeological services. They brought with them a specialised all-terrain vehicle (ATV) with an autosteer tool (that uses RTK (Real-Time Kinematic) GPS) and their experience of using it for very large-scale magnetic prospection. This tool takes over the steering of the vehicle and guides it along predetermined lines designed to achieve a specified density of coverage. It was developed for precision agriculture, but they have been using it to guide geophysical surveys in large land parcels. We brought with us a compact Mala Mira GPR array that is normally used in a hand pushed buggy system. This is a multichannel ground penetrating radar system that has multiple antennae that fire in a complex pattern giving an effective data resolution of 6cm x 6cm, up to 4m below the ground surface. GPR is ideal for use on Roman sites as it responds very well to buried stone structures, allowing them to be imaged in detail. 

This was an iterative process and involved the manufacturer of the equipment as well as we worked on the best physical and software configurations of the equipment to allow it to run towed behind the vehicle. For example, it was readily apparent that we needed wired connections from both the GPR antenna and the external GPS system to overcome issues with the reliability of Bluetooth and local Wi-Fi connections.  

We had to work carefully with the manufacturer to optimise the programmed coverage density to allow for all of the possible positioning errors present in the system. These were difficult to predict between the navigation system on the ATV, the RTK system on the GPR and the complications from towing on undulating ground. We also think a solar storm might have contributed to the errors (space weather is not something I thought would affect my career when I started out in archaeology 25+ years ago!). 

Eventually we settled on a planned traverse separation of 50cm. This required quite some skill by the operator of the ATV even with the guidance tools. We learned a lot from Kinnie Wade, the archaeological geophysicist from AOC about how to do this well and how to plan the driving pattern best for the conditions and efficiency. As part of the collaboration, we taught Kinnie and colleagues how to work with the compact Mira system and multichannel GPR more generally. 

Watch a video of the ATV in action collecting data.

Collaborating with community volunteers

The second week on site saw 25 volunteers from the community (both local to Newstead and as far away as Northumberland and County Durham) to learn from me and Lucy, one of our BSc students doing her industry placement as a geophysical research intern. Their goal was to learn how to collect GPR data using a smaller hand pushed instrument with a single antenna. Also, and perhaps more importantly, I taught them how to plan their survey in terms of resolution and orientation for their site and research question, regardless of methods, to maximise the usefulness of the resulting data.  

Together we collected an area of data over the location of the amphitheatre north of the fort that sits on the edge of very steep ground overlooking the river, and the likely crossing point, and we covered the same area using earth resistance. Earth Resistance measures the electrical conductivity of the ground. Buried stone will be less conductive, and things like ditches will be more conductive. The results from this were fascinating and need to be further deciphered but they hint at a ditched feature just outside the northern entrance of the amphitheatre. Is this part of the amphitheatre or some earlier or later (and therefore non-Roman) structure?

Meanwhile, the AOC team were completing a heroic effort with the ATV towed multichannel GPR in the fort, producing the most comprehensive picture of what is still underground on the site following the first excavations at the beginning of the 1900s. This covered our second and third goals nicely. But what about the first goal – finding the river crossing? 

Flying the UAV drone over the river

As often happens with brand-new methods, we didn’t have a clear enough understanding of how the altimeters on the UAV (Unmanned Aerial Vehicle) for the airborne surveys were going to respond to the fast-running River Tweed. This was important as we needed to fly within 0.7m of the surface! Ultimately, it was too risky to fly with the geophysical payloads, but we completed higher altitude photogrammetry of the river, banks and existing bridges. We’ll be going back later in the year to do the geophysical flights using the photogrammetry to plan them rather than relying on live altimetry. We will also test a miniature remote operated surface vessel with sonar arrays to look for the Roman Bridge. It’s hard to be sad about having a good reason to go back up to the site though – it is an amazing place and from the early view have had of the survey results so far, we’ve uncovered not only confirmation of the below-ground survival of many of the structures described by Curle (the first excavator), we also have some possible structures, particularly deeper in the NE quadrant of the fort that were not described by him- these could be pre-Roman, or represent an earlier undocumented phase of Roman Military activity on the site. AOC are now going to analyse and interpret the survey results with the help of the Trust so there will be plenty more to come from this story! 

Image 1: Dr Kayt Armstrong.

Image 2: view of the Trimontium GPR data from 2022 and 2026 from the main fort complex - watch a video that illustrates the data at different depths.