A Digital Surface Model was created in Agisoft Metashape v. 1.8.1 using the following general steps (see Over and others, 2021):
1. Project was created with imagery (located in larger work citation 2022014FA_Marconi_f1.zip) and position (located in larger work citation 2022014FA_Marconi_photolocations.csv) data imported with the NAD83(2011)/UTM19N and NAVD88 datums. An 0.15 m GNSS/IPS offset in the z direction was used to represent the approximate offset of the GNSS mount and camera on the Helikite. Photos were aligned at a low accuracy so that GCPs could be automatically detected in the point cloud. Twenty GCPs (2022014FA_Marconi_nav_GCPs.csv) were located.
2. Accuracies for the GCPs and camera positions were set to 0.02 m and 10 cm, respectively, but the positions were only used as check points and to speed up alignment. The photos were re-aligned with high accuracy (the pixels were not subsampled for increased processing speed) using a keypoint limit of 40,000 and unlimited tie points. Alignment uses the positions and matching pixels between images to create point clouds and puts the imagery into a real-world spatial context. Photos that failed to align/find tie points contained breaking waves and water.
3. The Metashape software used least squares to refine and optimize the camera positions, GCPs, and lens model with gradual selection and optimization parameters of: Ru = 12, Pa = 4, and Re = 0.3 to minimize reprojection error, or the distance between the measured points and the software created points.
4. A dense point cloud was generated using high-quality (images were not subsampled) and a low-frequency filtering algorithm. The dense point cloud was then edited to remove noise by filtering with point confidence before generating an interpolated digital surface model (the software calls all models elevation models, but a surface model includes vegetation/canopy returns, a terrain model would be only bare earth returns).
5. The digital surface model was used to build an RGB averaged orthomosaic, which was then exported from the software at a resolution of 10 cm in NAD83(2011)/UTM Zone 19N (ESPG:6348).
6. The orthorectified product was turned into a cloud-optimized GeoTIFF (COG) using gdal_translate with the following command: for %i in (.\*.tif) do gdal_translate %i .\cog\%~ni_cog.tif -of COG -stats -co BLOCKSIZE=256 -co COMPRESS=DEFLATE -co PREDICTOR=YES -co NUM_THREADS=ALL_CPUS -co BIGTIFF=YES (v. 3.1.4 accessed October 20, 2020
https://gdal.org/), where i is the name of each GeoTIFF section.
