GSA Annual Meeting in Phoenix, Arizona, USA - 2019

Paper No. 188-20
Presentation Time: 9:00 AM-6:30 PM

VEGETATION AND TOPOGRAPHY MAPPING OF COASTAL DUNE COMPLEXES USING SMALL UNMANNED AERIAL SYSTEMS AND GROUND-BASED IMAGERY


KREBSBACH, Jackson1, YURK, Brian P.1, PEARSON, Paul1, STID, Jacob T.2 and HANSEN, Edward C.3, (1)Department of Mathematics, Hope College, 27 Graves Place, Holland, MI 49423, (2)Department of Earth and Environmental Sciences, Michigan State University, East Lansing, MI 48824, (3)Department of Geological and Environmental Sciences, Hope College, 35 E. 12th Street, Holland, MI 49423

Active coastal dune complexes are dynamic environments characterized by interactions between plant populations, topography, and physical processes, including windflow and sediment transport. While patches of dune activity encourage ecological diversity, mobile dunes may also negatively impact human structures in coastal regions. Time series of high-resolution maps of vegetation and topography can be used for the management of coastal dune complexes and for the study of their ecological and geomorphological dynamics. We used multispectral imagery acquired by a small unmanned aerial system (sUAS) to estimate vegetation density within a dune complex in Saugatuck, Michigan, using ground-based photographs for calibration. First, a machine learning algorithm (random forest) was used to classify the pixels in the ground-based imagery into six different categories, including bare sand and marram grass (Ammophila breviligulata). The ground sampling distance (GSD) for the ground-based imagery is less than 1 mm, allowing for the resolution of individual blades of grass but not individual sand grains. The resulting classifier has an estimated overall accuracy of 76.9%, while correctly classifying pixels as sand 94.1% of the time and as live vegetation 99.7% of the time. Next, the same regions were identified in the sUAS-acquired imagery, and spectral and textural information from these images were correlated with vegetation density estimates from the classified ground-based images. Although the GSD for the sUAS-acquired imagery is much larger (approximately 2 cm), the coverage extent is much greater. The resulting model was used to process an orthomosaic created from sUAS imagery covering the entire dune complex produce a high-resolution vegetation density map. The same sUAS imagery was used to construct a digital elevation model (DEM) for the complex. Both the orthomosaic and the DEM were produced using Agisoft Metashape Professional (Version 1.5.3) photogrammetry software using high-accuracy (approximately 2 cm) ground control points.