The Use of Oblique Imagery and Ground Elevation Surveys to Generate a Time Series of Wet/Dry Shoreline Elevations

Various coastlines, including along the Gulf of Mexico, are seeing the effects of sea level rise and land subsidence. This combination of environmental conditions leads to increasing inundation frequencies on beaches and coastal structures. Better tools and models to predict water levels, including the impact of wave run-up, are increasingly necessary for coastal managers. To build better operational models, measurements more specific to the operational situations are necessary. Our research team is in the process of developing a deep-learning neural network to predict water levels and run-up on beaches. This model has been trained with human-delineated aerial images captured by an unmanned aerial system (UAS) of Texas and Florida coastlines; however, these drone images only provide a small sample of metocean conditions. To produce a more robust model, we installed beach and surf observing cameras on Horace Caldwell Pier, Port Aransas, Texas which capture oblique imagery of the beach every 30 minutes. The goal is to create a time series of the elevation of the wet/dry shoreline.

In previous steps of this project, we established how to consistently delineate the wet/dry shoreline based on aerial images. We have adapted the methodology to the current oblique beach imagery dataset including the challenge of how to best georeference the images while using targets placed and surveyed on the beach every two weeks. The development of the method included the selection of a GIS software and the selection of a transformation method all with the goal of obtaining the most accurate and precise estimates. After various attempts, including trials in QGIS, a system was established using ArcGIS Pro. As part of the process, the elevations of each delineated wet/dry shoreline are calculated using a digital elevation model (DEM) created based on data acquired during the biweekly ground surveys. The surveys are conducted over a 30.5 by 61m study area with points coordinates, including elevations, taken every 10 feet or 3.05 m in a grid pattern. Within this area, we use a smaller 21 by 22m study site for the wet/dry shoreline predictions. In addition, 14 wooden targets are placed on the beach to make georeferencing the oblique images possible. Of these 14 targets, a subset of 6 targets are actively used for georeferencing the area most relevant to the varying location of the wet/dry shoreline. The current presentation will include comparisons of wet/dry shoreline locations and elevations depending on conditions measured at a nearby tide gauge and our progress on oblique image georeferencing and elevation calculations. The overall objective of the research is to use the measured time series of wet/dry shoreline elevations to train an AI model predicting total water levels, including runup, on a beach.