The survey area lies within University of Bristol grounds and was undertaken under departmental permissions. All flight activities were logged in accordance with departmental rules. The UAV was flown below 120 m a.g.l., within visual line of sight (VLOS), and operated under the A3 sub-category. A DJI Mavic Air 2 was used to survey an area of approximately 0.5 hectares, following a non-systematic flight path intended to produce a rough orthographic image of visible parch marks. Control points were established using a Topcon HiPer HR GNSS GPS in a base-and-rover RTK configuration. The camera gimbal was maintained at 90�, with an image overlap exceeding 80% and a two-second capture interval. A total of 124 images were collected72 at 25 m a.g.l. (98.18 m a.o.d.) and 52 at 40 m a.g.l. (114.50 m a.o.d.). Site alignment was achieved using known GPS points, with no coded targets employed. Data were processed in Agisoft Metashape Professional (v.2.1.3 build 18946). Images were aligned, and a tie-point and dense point cloud were generated, achieving an RMS accuracy of 0.384 (1.20 pixels) and a GSD of 8.95 mm/pixelwithin acceptable parameters for aerial photogrammetric survey. A textured mesh was created and georeferenced using GNSS control points, then converted to OS National Grid and Ordnance Datum (OD) coordinates via OSGM15 and OSTN15 transformations. An orthorectified image was exported as a GeoTIFF at 1.00 mm/pixel GSD. All GPS and image outputs were imported into ArcGIS Pro 3.0.3 for verification and refinement. Reclassification reduced the GSD to 0.05 mm for archiving. Digital Terrain Model (DTM; Figure 5) and Digital Surface Model (DSM) datasets were produced from the processed point cloud. The DTM was generated after classifying and filtering points to remove vegetation, structures, and other non-ground features (Appx. Sec. 6, Table 8). The resulting surface was resampled in ArcGIS Pro using bilinear interpolation from 1.79 cm/pixel to 50 cm/pixel. The DSM, generated from the unfiltered point cloud, was resampled from 1.79 cm/pixel to 5 cm/pixel. The survey has produced results that are of interest for archaeological interpretations of the site, especially with regards to the location of trenches and the alignment of the Royal Fort Wall. Twenty-nine features were mapped, with twenty-one interpreted as archaeological or man-made in origin. Several key parch marks aligned precisely with the hypothesised line of the seventeenth-century Royal Fort wall, validating earlier conjectural plans and linking the aerial evidence with previously excavated sections. Additional linear and curvilinear marks indicate eighteenth- and nineteenth-century reworking of the site during Humphry Reptons landscaping, while several modern service lines and past excavation trenches were also identified. Natural features were distinguished through their irregular shapes, slope alignment, and correlation with known geological variations. Overall, the survey provides an important spatial dataset confirming the survival and position of Civil War fortifications at Royal Fort. It refines the understanding of how the site evolved through subsequent estate landscaping and later university development. Beyond its immediate archaeological value, this work highlights the effectiveness of UAV-based recording as a rapid, non-invasive method for investigating heritage sites, and strengthens the regional research framework for Bristols post-medieval and Civil War archaeology.
Alex Birkett (Fri,) studied this question.