Archaeology from the Air: UAV survey in Co. Donegal

Outreach and Communications Officer Hannah writes about The Discovery Programme’s recent survey work at McSwyne’s Castle, Co. Donegal.

In September I joined the Survey Team at McSwyne’s Castle, Co. Donegal, where we needed to undertake Coastal Zone Assessments (CZA) and Uncrewed Aerial Vehicle (drone) Surveys.

This was part of ongoing survey work in the area – work which is grant aided by the Royal Irish Academy Research Grants, funded by the National Monuments Service.

The location of McSwyne's Castle and St John's Point in Co. Donegal. Copyright Googlemaps2024.

The Discovery Programme is monitoring and recording this area because it’s being impacted by coastal erosion. This is a natural process, exacerbated by climate change, which we can’t stop. Therefore, it’s important to accurately record monuments whilst they are still here. McSwyne’s Castle had suffered a recent landslip, with sediment under the castle wall crumbling away. Moreover, there haven’t been extensive coastal surveys in the area so there is still a lot to learn about this stretch of coastline.

McSwyne's Castle

It can be helpful to use different survey techniques in the same area to build up a picture of the archaeology present and the effects of coastal erosion. This “toolkit” approach was something we used a lot during the CHERISH project, which you can read more about here. So whilst myself, Rob and Rebecca began the UAV work around the castle, Cormac and Ted set off down the coast to complete the CZAs.

Cormac undertaking Coastal Zone Assessments

Our attempts at completing the UAV survey had been foiled throughout the summer by high winds and heavy rain. Whilst Cormac and Ted braved the shoreline for the Coastal Zone Assessments in these summer storms, it isn’t safe to put a drone up in high winds.

We therefore chose a spell of late September sunshine to pack the car up with kit and head out to this beautiful stretch of coast to complete the survey.

Our UAV survey focussed on two key areas: firstly, the castle itself and secondly, the surrounding landscape. This is because Ted had observed earthworks (humps and bumps!) in the landscape here, which might indicate buried archaeological features which could be better seen from the air.

The landscape landward of McSwyne's Castle

A lot of archaeological investigation can be done without putting a spade in the ground. This is often called “non-invasive” investigation and has the benefit of not affecting the archaeology you’re investigating. Excavation, although the best-known of the archaeological investigation techniques, is ultimately a destructive one. When excavating, we only get one chance to preserve our findings on the site through detailed recording.

Photogrammetry is one such “non-invasive” technique in which many overlapping photos are taken and can then be digitally “stitched together” to create a 3D model. This model can be used to examine a site in detail, allowing a close-up view of places that wouldn’t be accessible, and an overall landscape view to help piece together different visible earthworks.

Photogrammetry can be used on a small scale using a camera, or a UAV for larger areas. For McSwyne’s Castle, we decided to use a combination of handheld photography, camera pole and UAV photographs.

The end result: a 3D model created from a UAV survey. This is Ballynarrid, Co. Waterford, which was recorded as part of the CHERISH project. You can see the full model here: https://skfb.ly/oUtwA

First off, Rebecca prepared the DSLR camera, ensuring the settings were correct for the lighting conditions and subject. These details were written on a whiteboard and photographed alongside a colour palette which helps the software identify the exact colours captured during the processing phase later on.

Rebecca and Rob preparing the camera and photographing the colour palette.

Preparations complete, Rob started the photography, taking overlapping photos all the way along the castle wall, including the area most recently affected by the erosion. The next stage was to repeat the process with the camera pole, allowing higher parts of the wall to be photographed and a downward angle to capture the rock tumble from above.

Although we were grateful for the dry weather, the sun actually made the job harder. The “flat light” created by cloud cover is more useful for photogrammetry than the shadows made by bright sunshine.

Rob taking photos using the camera pole.

It was then time to bring out the UAV, to capture images from above which could later be used alongside the DSLR images to create the 3D model. This was done “freehand” with Rob again taking regularly spaced photographs with plenty of overlap between each image.

Drone warning board.

Drone safety and privacy are always priority. When The Discovery Programme undertakes UAV survey, we aim to do it in an area where there are limited visitors, but it’s important to warn anyone who is in the area that someone is operating a drone. The drone pilot is always accompanied by an “observer”. Because the pilot needs to look at the control panel, they aren’t always able to keep their eye on the drone, so it’s important that someone always has it in their line of sight. There are many areas where we can’t fly a UAV as we might violate people’s privacy or it might not be safe – so it’s important to check with landowners, visitors and always respect “no drone” signs.

To survey the wider landscape, we took the approach of programming the drone to take photographs at regular intervals in a grid pattern. This means that, once the pilot has insured that the camera settings are all correct and the drone’s flight pattern is set, the drone will automatically take photographs at exact intervals. The pilot just has to watch for any irregularities and ensure the UAV comes back to base for necessary battery changes.

A screenshot of the flight path programmed by Rob and Rebecca.
Rob and Rebecca planning the survey.

Despite the drone doing the legwork in the photography, there’s still some graft for those on the ground. In order to be able to “geo-rectify” (accurately locate) the model in the software you need to have targets in the photographs that the software can recognise. With the sun high in the sky, Rebecca and I rolled up our sleeves and set out to place the targets across the landscape. This also involved carrying the GNSS kit – a tripod with survey-accurate measuring equipment on it – to log their positions.

The GNSS kit recording a marker point.

With the targets out and the drone programmed, we needed to find somewhere we’d be able to see the drone during its entire flight. Due to the size of the survey area, this involved a hot walk up a small hill – for which we were rewarded with a view of both the survey area and across the water to Sligo.

The view from the top.

UAV surveys complete, it was time to head back to the car and check that all the data had been properly captured. We checked in with Cormac and Ted who’d had a successful day finding and recording new sites of archaeological interest along the coast. After a hot and tiring day, there was really only one way to finish it off – with a quick dip in the sea!

McSwyne's Castle.

The data captured from this fieldwork will now be processed back in the office – watch this space for the survey results and new 3D models!