They say a picture is worth a thousand words. What if that picture was taken from hundreds of feet above one of the most diverse and beautiful marine ecosystems in the world? Such images from above — captured with unoccupied aerial vehicles (UAVs) or drones — offer a glimpse into the beauty and wonder of nature and marine ecosystems. These snapshots in time can provide valuable data on change at the scale of land- and seascapes, inspire others to protect and conserve our environment, and encourage stakeholders to engage in and advocate for science. A group of public scholars, including faculty, post-docs, students, community members, and K-12 children and teachers, has been working together since 2017 in Belize to capture this beauty from above with drones to support science and conservation. Open Reef, a project of Citizen Science GIS at University of Central Florida, uses affordable, consumer-level drones to capture high resolution imagery of the Belize Barrier Reef and its vast ecosystems. And they share their imagery and processes for free as open data.
The Citizen Science GIS drone team mapped reef ecosystems around Carrie Bow Cay as part of the 2019 MarineGEO field campaign. Drone flights to map the local seascape were coupled with a series of georeferenced transects, along which divers collected underwater photoquadrats to validate habitat desigantions based on drone imagery. UAV flights over Carrie Bow Cay were conducted by the Smithsonian MarineGEO team and the Citizen Science GIS drone team from September 29th to 30th, 2019. Figure below shows an example of high-resolution drone imagery over CBC. Normally drone permits for international researchers are challenging to secure in Belize, but the Citizen Science GIS team has been working on multiple projects in Belize so they have permission for flying drones for research activities through partnerships with Civil Aviation and Belize Coastal Zone Management Authority and Institute. This trip’s drone mapping was conducted to generate a series of high-resolution maps of Carrie Bow Cay and to explore the surrounding area’s coral reef and seagrass habitats. Multiple flights were conducted on clear days with good visibility. The team used DJI Phantom 4 Pro drones flying at different heights. Flying at different heights provided different resolution and detail. For example, the spatial resolution of the mapping product at 400ft is about 5 cm covering a relatively larger area, on the other hand, the mapping product has about 1 cm spatial resolution at 100 ft, but the mapping coverage is relatively smaller. In a unique university and industry partnership between the UCF team and Esri, the world’s leading GIS company, the images collected by the DJI Phantom were processed using Esri’s Drone2Map software as 2D and 3D products, including orthomosaic, Digital Surface Model (DSM), Digital Elevation Model (DEM), and point clouds.
Figure below shows an example of the drone mapping product at 400ft on September 30th. The spatial resolution of the product is 6.26 cm for both orthomosaic imagery and elevation products. The water clarity is outstanding for the area around the CBC. In the drone imagery, the spatial patterns of various reef habitats are vivid and clear. For example, the parallel “spur and groove” pattern on the forereef are clearly visible as dark strips stretching out from the island ot the east. The pattern reflects reef development in response to the prevailing trade wind direction. In addition, the halo pattern around the coral patch reefs in the lagoon is clear, representing the limit within which herbivorous fishes venture out from the reef to graze the seagrass. These halos could be also used as barometers to understand reef health.
Acquiring data with UAV approaches is often less expensive and more convenient than hiring out occupied aircrafts, especially in more remote, and inaccessible places like the Belize Barrier Reef. In addition, although satellites capture images of remote areas and difficult terrain, they often have infrequent and inflexible temporal revisit cycles. UAVs, on the other hand, can collect on-demand data for field campaigns like this one. A typical UAV/drone mapping project will involve multiple stakeholders (e.g. scientists, engineers, pilots), technologies (e.g. drone platforms, controllers, software packages, sensors), parameters (e.g. flight altitude, scientific sensor calibration date and processes, scientific parameters, FAA airspace regulations), and complex processes (e.g. data stitching, data management, data pre- and post- processing), many of which can influence the utilization and interpretation of the data.
Using a high-performance Garmin R1 GNSS system, the team also “ground-truthed” their imagery by collecting about 20 points for CBC. These GPS points created tie points in open water in particular for stitching all of our individual images together to create a complete mosaic of the study area. For points on land, the team used obvious objects, such as corners of the docks that could be easily identified in the drone imagery. For image referencing over the water, they used light-colored buoys with anchors fixed underwater to serve as the ground control points (GCPs, Figure 2). The coordinates of the GCPs were input to the mapping software to geo-register (or stitch together) the images. Georeferencing is the name given to the process of transforming a scanned map or aerial photograph so it appears “in place” in GIS. By associating features on the scanned image with real-world x and y coordinates, the software can progressively warp the image so it fits to other spatial datasets. Because the drone mapping takes images in a perpendicular way, most of the imagery will have horizontal shifts, therefore the team uses a technique called similarity transformation to rectify the imagery. The similarity transformation is a first-order transformation which tries to preserve the shape of the original image. The RMS error tends to be higher than other polynomial transformations since the preservation of shape is more important than the best fit. The team recorded GCPs and observed how the RMS error changes with the increase in the GCP number and with the refinement of the GCP positions. The researchers then refined the positions of GCPs to minimize the RMS error. If there is a GCP point with a large individual error value, the team removes the point to account for GPS measurement error. Moreover, they also deployed two tarps on the island (Figure 3). One is light color and one is black color. Two tarps not only serve as GCPs on the ground to rectify the imagery, but are also used for atmospheric correction.
The imagery collected by the team supports the 2019 field campaign at CBC, building a high-resolution habitat map of the vicinity of the Smithsonian’s long-term observatory at Carrie Bow Cay, and also supports future citizen science and education efforts in Belize. All drone imagery collected from this fieldwork is provided as open data from the team at University of Central Florida team to support science, discovery, and citizen science. The Citizen Science GIS team looks forward to expanding its work around the globe — and at other MarineGEO sites — with additional partners and communities interested in participatory drone science. With the advent of drones, interdisciplinary teams can reach new heights, sharing the beauty of nature from above to inspire others to engage with science and support conservation efforts.