Tracking Coastal Inundation and Wave Dynamics at Public Beaches Using High-Resolution Surveillance Camera Images and Machine Learning
Presentation Type
Poster Presentation
In Person or Zoom Presentation
In-Person
Campus
Daytona Beach
Status
Student
Student Year and Major
Second Semester Masters (1) and Junior (2)
Organization, if requesting a table
Bethune-Cookman University
Presentation Description/Abstract
Barrier islands and beachfront systems are increasingly vulnerable to chronic inundation, which is driven by natural processes (tides, wave action, storm surge, etc.) and anthropogenic pressures, such as shoreline armoring and beach modification. These changes have contributed to the loss of natural dunes and narrowing beach widths, exacerbating the impacts of wave run-up beyond the influence of astronomical tides alone. Wave run-up, the movement of water up the beach beyond the still water line, is influenced by factors such as beach slope, wave height, and other coastal morphology, and plays a central role in determining the extent of coastal flooding. Remote sensing and machine learning offer tools to address the urgent need for early detection and prediction of coastal inundation events. This study applies image segmentation, an innovative computer vision technique, to time-lapse a publicly accessible beach camera livestream of South Dunlawton Beach in Volusia County. The images acquired from the camera livestreams are processed using convolutional techniques within Artificial Neural Networks (ANNs), applying semantic segmentation of the beach environment to distinguish regions of water, wet sand, and dry sand. This protocol offers high spatial and temporal resolution as well as the predictive ability needed to evaluate impacts of inundation on wildlife habitats, recreational use, and shoreline infrastructure within the county. The findings from this study will be shared with local coastal managers and would provide insights into a potentially shifting beach face, which will have broad implications for habitat conservation and resilience planning under future sea-level rise scenarios.
Keywords
Machine learning, remote sensing, coastal inundation, flood risk
Tracking Coastal Inundation and Wave Dynamics at Public Beaches Using High-Resolution Surveillance Camera Images and Machine Learning
Barrier islands and beachfront systems are increasingly vulnerable to chronic inundation, which is driven by natural processes (tides, wave action, storm surge, etc.) and anthropogenic pressures, such as shoreline armoring and beach modification. These changes have contributed to the loss of natural dunes and narrowing beach widths, exacerbating the impacts of wave run-up beyond the influence of astronomical tides alone. Wave run-up, the movement of water up the beach beyond the still water line, is influenced by factors such as beach slope, wave height, and other coastal morphology, and plays a central role in determining the extent of coastal flooding. Remote sensing and machine learning offer tools to address the urgent need for early detection and prediction of coastal inundation events. This study applies image segmentation, an innovative computer vision technique, to time-lapse a publicly accessible beach camera livestream of South Dunlawton Beach in Volusia County. The images acquired from the camera livestreams are processed using convolutional techniques within Artificial Neural Networks (ANNs), applying semantic segmentation of the beach environment to distinguish regions of water, wet sand, and dry sand. This protocol offers high spatial and temporal resolution as well as the predictive ability needed to evaluate impacts of inundation on wildlife habitats, recreational use, and shoreline infrastructure within the county. The findings from this study will be shared with local coastal managers and would provide insights into a potentially shifting beach face, which will have broad implications for habitat conservation and resilience planning under future sea-level rise scenarios.