Drone photogrammetry and KMeans point cloud filtering to create high resolution topographic and inundation models of coastal sediment archives

• Published in: Earth surface processes and landforms Vol. 43; no. 12; pp. 2603 - 2615
• Main Authors: Callow, J.N., May, S.M., Leopold, M.
• Format: Journal Article
• Language: English
• Published: Bognor Regis Wiley Subscription Services, Inc 30.09.2018
• Subjects: Accuracy; Aircraft components; Archives & records; Backwaters; Coastal sediments; coastal washover deposit; Coastal zone; Colour; Cyclones; digital elevation model (DEM); Drone aircraft; Dynamics; Earth; Filtration; Geomorphology; Ground penetrating radar; Hurricanes; KMeans; Landforms; Model accuracy; Overwash; Photogrammetry; point cloud filtering; Preservation; Radar; Radar data; Remotely piloted aircraft; remotely piloted aircraft systems (RPAS); Resolution; Runoff; Sediment; Sediments; Stratigraphy; structure from motion (SfM); Target recognition; Terrain models; Terrestrial environments; Three dimensional models; Topography; Tropical climate; Tropical cyclones; Tsunamis
• ISSN: 0197-9337; 1096-9837
• DOI: 10.1002/esp.4419

Coastal areas are vulnerable to the impacts of tropical cyclones (TC), tsunamis and other water super‐elevation events, but the frequency of these events is often poorly represented by conventional records. Coastal overwash deposits (including washover fans) can provide a longer‐term archive of event frequency. Because of their low‐gradient geomorphic form, washover fans require high accuracy (centimetre‐resolution) topographic models to understand patterns of connectivity and dynamics that control archive formation. Using images collected by a remotely piloted aircraft system (RPAS, or ‘drone’) and Structure‐from‐Motion (SfM) photogrammetry techniques, we apply a novel point‐cloud filtering technique based on KMeans classification of the R‐G‐B colour of each X‐Y‐Z point to remove vegetation and create a centimetre‐resolution and accuracy bare‐earth digital terrain model (DTM) of a washover fan in Exmouth Gulf (Western Australia). Using the RPAS‐SfM orphophoto and DEM data, supported by ground‐penetrating radar (GPR) and field stratigraphic analysis, we show how this approach can be applied to understand dynamics controlling low‐gradient geomorphic landforms, using an example of a washover fan sedimentary archive in northwestern Australia created by extreme overwash events. Our approach reveals the likely role of backflooding and terrestrial runoff in creating backwater environment for sub‐aqueous deposition and good sediment preservation and identifies key areas to target for detailed dating and stratigraphic analysis of a potentially decadal to sub‐millennial resolution sediment archive of TC activity. Copyright © 2018 John Wiley & Sons, Ltd. Drone photogrammetry point cloud data created by Structure from Motion (SfM) is filtered using Kmeans filtering of the R‐G‐B point cloud colours to make high accuracy and resolution bare earth Digital Elevation Models (DEMs). We show how this can assist in interpretation of a coastal overwash fan and connectivity mapping of the sedimentary landform.