Drone orthomosaic, side scan sonar data, and geotagged underwater photos of the nearshore environment near Wendtorf/Bottsand, Germany (Baltic Sea)

This dataset contains imagery acquired at Bottsand Strand, near Wendtorf, Germany (54.43°N, 10.29°E) on 27 August 2024. Aerial Drone Imagery 20240827_Wendtorf_Ortho.tif - orthomosaic produced using original drone images 20240827_Wendtorf_Ortho_CLAHE.tif - orthomosaic produced after applying histogram equalization autel_raw.zip - 509 drone aerial images used to produce the orthomosaics Uncrewed Surface Vehicle Data humminbird_raw.zip - data recorded by Humminbird Helix 7 side scan sonar Wendtorf_Ardu_Log.csv - navigation data logged by the autopilot geotagged_continuous.zip - 4694 underwater images produced by exporting GoPro video frames at 1 second intervals, aligning with navigation data, and writing positions to image EXIF. 20240827_Wendtorf_Humminbird_Depth.txt - text file containing date, time, lon, lat, depth Aerial drone imagery was acquired by an Autel EVOII RTK Pro quadcopter. UAV flights were conducted between 09:00-10:00 CET to allow for sufficient sunlight to reach the benthic features of interest (primarily eelgrass, macroalgae, and bare sand) but before sun glitter contaminated large portions of the imagery. Flights were carried out in auto mode using missions that were pre-programmed in the Autel app. Missions were planned to acquire images with 70% overlap and 50% side lap at an altitude of 100 m and at a speed of 4 m/s. Image contrast settings (shutter speed and aperture) were monitored continuously and adjusted manually when necessary. RTK position corrections were obtained from an NTRIP server. The study area could not be covered in a single flight due to the limited battery life of the UAV. Four mission were flown and 509 images were acquired. UAV images were processed using Agisoft Metashape Professional (version 1.7.1) following the methodology described by Over et al. (2021). Camera groups were created for each of the four flights and the corresponding images were added. In the ‘reference’ tab, the overall position accuracy was changed to 0.5 m. Image masking was not required. The image alignment, camera optimization, and iterative gradual selection steps were automated using a python script. For the alignment step, the key point and tie point limits were set to 60000 and 0, respectively. Generic and reference preselection were disabled. In the gradual selection step, target values of 12, 10, and 0.5 were set for Reconstruction Uncertainty, Projection Accuracy, and Reprojection Error, respectively. For each parameter, the value was adjusted until the number of points selected for deletion reached 10% of the total number of remaining points. Camera optimization was performed after each set of points were deleted. This process could be repeated up to 10 times for each parameter. After gradual selection, the sparse point cloud was manually inspected and any remaining ‘sinkers’ or ‘fliers’ were manually deleted. The dense point was computed at ‘ultra high’ quality using the ‘mild’ depth filtering mode. The dense point cloud was cleaned by filtering according to confidence. Points with confidence values of 0-2 were deleted. The digital elevation model (DEM) was then computed from the dense point cloud, followed by the orthomosaic. The orthomosaic was exported as a geotiff (WGS 84 / UTM zone 32N (EPSG::32632) with 2.3 cm resolution. The UAV survey covered an area of 0.256 km2. A second orthomosaic was produced to examine the use of histogram equalization in improving contrast in images of submerged features. The original images were duplicated and enhanced by applying the contrast limited adaptive histogram equalization (CLAHE) function in openCV. To ensure compatibility with Metashape, the EXIF metadata were copied from the original images. After the UAV missions were completed, a small uncrewed surface vessel (USV) equipped with a side scan sonar and underwater camera surveyed the same area. The USV consisted of a foam boogie board propelled by two motors (Blue Robotics M200). The motors were mounted to paddle board fins and were attached to the boogie board using 3D printed brackets. The speed and heading of the USV were controlled by a GPS-equipped autopilot (Pixhawk 2.4.8) running ArduRover firmware. Batteries and electronics were housed in a waterproof enclosure (IP67 ABS junction box). A Humminbird Helix 7 sonar with side imaging was mounted on the USV. The sonar data was stored on a micro SD card in the head unit. The head unit was installed in the electronics enclosure. The transducer was located 10 cm below the surface and was mounted on a 25 mm aluminum pole. The underwater camera (GoPro Hero9) was mounted on the same pole.