- Labeling Poststorm Coastal Imagery for Machine Learning: Measurement of Interrater Agreement , EARTH AND SPACE SCIENCE (2021)
This project will address the problem of recurrent, shallow flooding in low-lying coastal communities. As local sea-level rise (SLR), land subsidence, and heavy rainfall events increase, so does the frequency of flooding in low-lying coastal areas. The tidal cycle now takes place on higher average sea levels, resulting in ÃƒÂ¢Ã¢â€šÂ¬Ã…â€œsunny-dayÃƒÂ¢Ã¢â€šÂ¬Ã‚Â flooding of roadways during high tides. Sea water also infiltrates stormwater drainage systems at low tidal levels, such that ordinary rainstorms lead to flooding. While these minor floods draw less attention than catastrophic storms, their high frequency imposes a chronic stress on coastal communities and economies by disrupting critical infrastructure services. The proposed work integrates outreach and research activities over the two-year project period to improve our prediction and communication of chronic flood hazards. First, we will couple an existing high-resolution hydrodynamic model used for prediction of estuarine flooding in the region (SWAN+ADCIRC) with a stormwater management model (SWMM5) to hindcast and identify the drivers of unexpected flood events in Carolina Beach, a community plagued by chronic flooding. In parallel, we will co-develop potential flood-mitigation actions with Carolina BeachÃƒÂ¢Ã¢â€šÂ¬Ã¢â€žÂ¢s Flood Working Group to inform future work using the coupled model framework. Second, we will deploy a real-time flood sensor network (in development by PIs Anarde, Hino, and Gold) in Carolina Beach to fill data gaps on the incidence and causes of chronic flooding. These data will inform an early-warning system, designed with local officials and community members, for real-time communication of flood hazard.
Hurricanes, with their strong wind, large waves, and storm surge, can profoundly reshape coastal landforms and damage near-coast structures. Mutual resilience of coastal communities, ecosystems, and landscapes to future storm impacts requires a clear understanding of the hydrodynamic forces impacting coastal systems during storms and accurate representation of these processes in numerical models. Previous efforts to collect coastal data during storm impact have relied on the ÃƒÂ¢Ã¢â€šÂ¬Ã…â€œdeploy and retrieveÃƒÂ¢Ã¢â€šÂ¬Ã‚Â model wherein successful data acquisition hinges on post-storm retrieval of data loggers. Inevitably under this paradigm, instrument damage and loss has resulted in sparse data sets with limited spatial and temporal resolution. Recent technological advancements in wireless monitoring and distributed sensor networks have the potential to catalyze a shift away from the ÃƒÂ¢Ã¢â€šÂ¬Ã…â€œdeploy and retrieveÃƒÂ¢Ã¢â€šÂ¬Ã‚Â framework toward ÃƒÂ¢Ã¢â€šÂ¬Ã…â€œreal-time monitoringÃƒÂ¢Ã¢â€šÂ¬Ã‚Â of storm impacts. Leveraging this potential, we propose to design and test a new low-cost wireless pressure sensor network for real-time measurement of waves and water levels during hurricane impact. The distributed wireless network will consist of multiple pressure sensor ÃƒÂ¢Ã¢â€šÂ¬Ã…â€œnodesÃƒÂ¢Ã¢â€šÂ¬Ã‚Â that transmit data via short-range radio to a central ÃƒÂ¢Ã¢â€šÂ¬Ã…â€œgatewayÃƒÂ¢Ã¢â€šÂ¬Ã‚Â, and thereafter to the cloud via a cellular modem. While the sensor in the proposed project records pressure, distributed wireless networks are inherently modular and future work will add utility to the instrument array by incorporating additional sensors (e.g., accelerometers). The proposed project will utilize existing infrastructure and expertise for laboratory (flume) and field-based testing of the prototype sensor network at the University of North Carolina ÃƒÂ¢Ã¢â€šÂ¬Ã¢â‚¬Å“ Wilmington (Dr. Mieras) and North Carolina State University (Dr. Anarde), as well cultivate new research collaborations among institutions and with a North Carolina-based startup, Agrinik Technologies, LLC. Once optimized, the new distributed wireless network will be used to address a myriad of data and knowledge gaps related to storm processes, including infrastructure fragility, feedbacks between structures and flow routing, and wave transformation during island overwash.