ANALYSIS OF ENGINEERED OYSTER REEFS USING 3D STRUCTURE-FROM-MOTION PHOTOGRAMMETRY: PRELIMINARY APPROACH, METHODS, AND TECHNIQUES FOR QUANTIFYING REEF GROWTH

Photogrammetric methods have been used in geologic and environmental applications for more than three decades. Many such studies were completed with expensive technology and computers capable of high-level computation. Therefore scale of observation, logistics, know-how, and the data processing capacity of computers limited applications and accessibility to some disciplines. With the growing availability of low-cost personal devices and emerging software, these methods can be applied to a wider-variety of research questions by a range of users. While the majority of photogrammetry work has been on land, it is here being used to develop new efficient methods for investigating metrics of oyster reef health, growth, and recovery. In this feasibility study, photogrammetric methods are applied to the investigation of constructed oyster reefs (aka “oyster castles”) from the coastal bays of Virginia. Two projects, one by The Nature Conservancy and one by the Chincoteague Bay Field Station, resulted in installation of oyster castles in 2010 and 2014. The castles provide local substrates for oyster recruitment, reduce wave action, improve water quality, and ultimately help stimulate oyster recovery. Several studies and methods have been applied to monitor and quantify oyster size, density, and settlement patterns, but many of these require extensive field time for data collection, data entry, and analysis and could be of limited accuracy. In this preliminary study, photogrammetry is accomplished through the use of Agisoft PhotoScan Software to splice 2D photomosaics which are orthorectified to produce a 3D point cloud (3DPC) using software algorithms that auto-detect fixed feature points. By combining the camera position and triangulation of these fixed points, a dense 3D geometric surface can be created. The 360˚ photomosaic is then draped back onto the 3DPC to produce a full 3D rendering. The X, Y, and Z values are then exported into ArcScene for analysis. These data can be sorted, grouped, and plotted to quantify clusters of points with different shapes, sizes, orientations, etc. In effect individual oysters can be identified and patterns in the distribution of oysters and their respective shapes, and sizes can be evaluated. These methods are still being developed, and tested so additional results are forthcoming.