First USGS Coastal Maps from Unmanned Aerial Systems
The USGS Coastal and Marine Geology Program made another technological step forward and produced two high-quality maps of coastal regions in Cape Cod, Massachusetts, using photogrammetry from images taken by unmanned aerial systems (UAS).
On March 1, 2016, a long permitting and planning process culminated with two UAS flights to map Coast Guard Beach in Cape Cod National Seashore. The project was a proof-of-concept exercise to demonstrate that UAS operations could be safely, legally, and effectively used to make maps of coastal features. The project was supported by the USGS Innovation Center for Earth Sciences and two ongoing Coastal and Marine Geology Program projects.
Coastal change is episodic. Significant changes in the beach and nearshore regions; erosion of dunes, bluffs, and cliffs; overwash; inlet formation; and changes in habitat occur in a matter of hours during storms. Major geomorphic changes can occur with only 24 to 48 hours advance notice, often after long periods of relatively slow change.
Prediction of storm impact in coastal regions requires accurate and up-to-date maps of coastal morphology on land (bluff or dune height, beach slope and width) and in the water (nearshore bars and shoals, offshore bathymetry). Evaluation of geomorphic response models requires accurate maps of the same features immediately after the events, before anthropogenic or natural fair-weather processes modify the storm-related changes. Thus, the ability to map before and after infrequent but significant events is critically important.
Structure-from-motion is a new but proven technique for making high-resolution maps from multiple photographic images. Structure-from-motion uses automatic point matching and least-squares fitting to reconstruct a three-dimensional scene from a set of images from different camera locations. Unmanned aerial systems provide the ability to acquire these images and map coastal features quickly, safely, and inexpensively, on short notice, and with minimal impact. By contrast, lidar surveys of coastal regions are infrequent and costly: the most recent measurements on Cape Cod were obtained in 2011, and logistics prevented timely mapping after the series of winter storms that occurred in January and February 2015.
Permits and planning documents from the National Park Service and the U.S. Department of the Interior were required prior to the flights. While the USGS is authorized to fly UAS under the terms of a Memorandum of Understanding with the Federal Aviation Administration (FAA), which provides a Certificate of Authorization, the USGS UAS team was busy, so we hired Raptor Maps, Inc., to fly and acquire the imagery. Contractors can now fly under a Section 333 Exemption from the FAA, which means that the USGS can take advantage of UAS techniques in many coastal regions.
Raw data from the UAS flights consist of thousands of high-resolution digital images, tens of surveyed ground control points, and about 100 independent surveyed points for comparison. The primary products from photogrammetry are point clouds containing hundreds of millions of georeferenced x, y, z points with associated RGB (red-green-blue) color values. These point clouds are similar to those generated by lidar surveys, and can be stored in the same formats and processed with the same software tools. Final products include digital elevation models (DEM) and georeferenced mosaics of the images.
One of the objectives of this study was to determine just how accurate these maps are. It turns out that, given adequate ground control, the maps can be very accurate. Barry Irwin walked transects on the beach and measured elevation at about 140 independent points with a precision of +/- 2 centimeters (cm) using differential GPS while the UAS were flying. These points were compared with the closest points on the DEM produced from the UAS imagery. More than 90% of the points agreed to within +/- 15 cm, and most (>80%) agreed to within +/- 10 cm. However, in regions with no ground control, the DEM tended to drift away from ground truth elevation values and was off by as much as 0.6 meters until corrected with ground control points based on earlier lidar surveys. This requirement for extensive ground control is one of the tradeoffs associated with structure-from-motion methods, however, in many cases, it is offset by the low cost and rapid turnaround time associated with UAS data collection.
In addition to the studies in Cape Cod National Seashore, we worked with collaborators and contractors to make multiple maps of Sandwich Town Neck Beach, Massachusetts. Repeat mapping there has allowed us to document changes associated with both natural events and beach nourishment projects. The maps made with UAS images provide us with data needed to evaluate numerical models of coastal geomorphological change.
AcknowledgmentsWe are grateful to the National Park Service, and particularly George Price (Superintendent of Cape Cod National Seashore) and his staff members Sophia Fox and Mark Adams, for permission to fly in the park and help with permits and logistics. Thanks also to the Town of Eastham, the Eastham Conservation Foundation, and the Massachusetts Audubon Society for permission to overfly their private parcels in the study area. We also thank Jeff Sloan and Bruce Quirk of the USGS UAS team. The project was supported by the USGS Innovation Center for Earth Sciences (http://geography.wr.usgs.gov/ICES/) and two ongoing USGS Coastal and Marine Geology Program projects: the Barrier Island Evolution Project (http://coastal.er.usgs.gov/bier/) and the Coastal Model and Field Measurements project (http://woodshole.er.usgs.gov/project-pages/coastal_model/).
in this issue: