Barro Colorado Whole-Island Aerial Photogrammetry Products: Orthomosaics, Digital Surface Models, Point Clouds, and Raw Images for 2018-2023

This dataset is part of a larger initiative monitoring forests in Panama using drones (unoccupied aerial vehicles), an initiative led by Helene Muller-Landau at the Smithsonian Tropical Research Institute. As part of this initiative, we have been collecting repeat imagery of all 1543 ha of Barro Colorado Island (BCI), Panama, since June 2015 (see Cushman et al. 2022a for data products for 2015, 2018, and 2020). The dataset published here encompasses a total of 8 flight missions between June 2018 and June 2023, including in June or July of every year, and additional missions in February and March 2023. The June-July missions were timed to capture the flowering of the canopy emergent tree species Dipteryx oleifera; the February and March missions were timed to capture flowering of the canopy tree Jacaranda copaia. Dipteryx flowers are purplish pink, and Jacaranda flowers are bluish purple. Flights were conducted using an eBee senseFly drone and a S.O.D.A camera having a resolution of 20 megapixels, at a fixed elevation of 601 meters above sea level, and thus 430-575 m above ground, and ~390-575 m above the top of the canopy (ground elevation ranges from 26 to 171 m, and canopy height ranges from 0 to 55 m). Flights were conducted with lateral overlap of 77% and along-path overlap of 77%, which translated to 170 m between flight lines and 113 m between photos along a flight path. The drone imagery was processed independently for each date using Agisoft Metashape Pro 2.0 Python API (Agisoft LLC), employing a standardized workflow. Key parameters for this processing included highest setting for photo alignment, medium setting for point cloud construction, and aggressive point filtering; for additional details, see https://github.com/VasquezVicente/ForestLandscapes/blob/main/LandscapeScripts/UAV_photogrametry.py Note that these data products have NOT been aligned across missions and contain substantial errors of alignment. Cushman et al. (2022b) employed tiling and iterative closest point algorithms to align the point clouds for 2015, 2018, and 2020 to 2009 airborne lidar data, create associated digital surface models, and then differentiate these to quantify canopy disturbance patterns. Her original R code is available at Cushman et al. (2022a); it is based on a number of packages that have since been retired. An updated version of this code is available on GitHub at https://github.com/PanamaForestGEO/DroneCodeBCIwide/tree/main/scripts. Notes: The BCI_whole_2018_06_20_EBEE_dipteryx mission encompassed three different dates: 2018_06_14, 2018_06_20, and 2018_06_07. Distinct flight lines were conducted on each date, and the products published here are a composite of images from these flights. The BCI_whole_2019_06_19_EBEE_dipteryx mission features raw images taken on 2019_06_19, 2019_06_22, and 2019_06_24. The BCI_whole_2023_03_18_EBEE_jacaranda mission's raw image dates spanned 2023_03_16, 2023_03_17, and 2023_03_18.