The intensification of climate change poses significant threats to coastal regions worldwide, manifesting in increased storm frequency, sea level rise, and consequent flooding risks. This study addresses the urgent need for innovative monitoring strategies by introducing an advanced coastal hazard monitoring system specifically designed for areas with underdeveloped monitoring infrastructure. Employing a blend of traditional methods and cutting-edge technologies, including the Segment Anything Model (SAM) for high-resolution image segmentation and Dynamic Mode Decomposition (DMD) for pattern recognition, we provide a comprehensive assessment of coastal water dynamics. The study highlights the application of SAM in identifying water-land boundary despite challenges such as image distortion and variable lighting conditions. Additionally, the innovative use of monoplotting with DEM provides a robust framework for accurate mapping in complex coastal terrains. This research advances our understanding of coastal dynamics under the impact of climatic changes and sets a new benchmark for environmental monitoring, offering substantial improvements over traditional methodologies by integrating technological advancements with practical fieldwork. The findings demonstrate significant implications for disaster preparedness and the sustainable management of coastal regions, emphasizing the necessity of adopting advanced technologies to enhance the resilience of vulnerable coastal communities against the escalating threats posed by climate change.
Climate change and the associated rise in sea levels are escalating coastal vulnerabilities, increasing risks of flooding and erosion in coastal communities. Climate change also poses a dire threat by amplifying the frequency and intensity of storm events. Consequently, coastal regions, especially those that are economically and socially vulnerable, must urgently implement protective measures to safeguard both the populace and critical infrastructure against these escalating threats.
For instance, September 2017, when Puerto Rico was struck by Hurricane Irma, a Category 5 storm, followed closely by Hurricane Maria, a Category 4 storm, exemplifies the devastating impact of such climate phenomena. These hurricanes unleashed catastrophic flooding and destructive winds, leading to a total blackout and the collapse of essential communication infrastructures. Given these precedents and the ongoing rise in sea levels, Puerto Rico's coastal communities face imminent danger of recurrent flooding incidents. It is imperative that comprehensive and immediate actions are undertaken to improve defenses in vulnerable coastal areas to mitigate the increasing flood risks engendered by sea level rise.
The recurrent flooding issues necessitate a proactive approach to monitoring and analyzing coastal resiliency. However, measuring water levels along coastal areas presents a range of logistical and technical challenges. One of the primary obstacles is the high installation and maintenance cost associated with creating a comprehensive sensor array. Another significant challenge is the limitation of remote sensing in terms of sampling frequency. While satellite and aerial data can provide extensive coverage and valuable insights into coastal dynamics, the temporal resolution is often insufficient for real-time or near-real-time monitoring. Satellites typically have fixed orbits and revisit times, which means that they can only capture data for a specific location at certain intervals-often measured in days or weeks. This low sampling frequency makes it difficult to monitor rapid changes or short-term events such as storm surges or tidal fluctuations. In addition, coastal cloud coverage prevents wide application of satellite imagery for environmental analysis.
The advancement of social media and citizen science has opened up innovative avenues for flood data collection, offering a valuable complement to traditional methods. However, despite these fruitful developments, several challenges and limitations persist. The accuracy of social media-based flood monitoring is still unreliable. Factors contributing to this uncertainty include the variability in the quality of user-generated content and the inherent limitations of the platforms themselves in addition to privacy issues. These challenges underscore the need for innovative approaches to coastal monitoring that balance accuracy, cost, and long-term viability.
Emerging technology provides new opportunities to enable innovative coastal monitoring technology. The Segment Anything Model (SAM), unveiled by Meta AI in April 2023, signifies a leap in image segmentation technology with its unparalleled ability to generalize across diverse image datasets without the need for extensive training on new objects. Groundbreaking for its minimal training requirement, SAM leverages zero-shot and one-shot learning principles enabling it to recognize and outline targeted objects through a broad comprehension and incorporation of a single example to refine accuracy. This model has great potential in reducing the resources required for data training and annotation, suggesting a transformative impact on coastal monitoring.
In addition to image processing, methods are needed to project the imaging data to 3-D world for quantitative analysis. An example is monoplotting which is a technique for interpreting oblique imagery from smartphones, UAVs, and monitoring systems by integrating single images with a 3-D Digital Elevation Model (DEM) to produce georeferenced data. This technique is pivotal across geoscience fields for geolocating objects and quantifying image content against a DEM or map, requiring only a single image and a DEM for 3-D scene reconstruction, thus avoiding the need for stereo imagery. Historically significant for analyzing earth surface changes and observing ground-based photographs, its recent applications include disaster rapid response and stakeholder communication. The method also enhances low-cost sensing in monitoring environmental dynamics and 3-D deformation reconstructions. Development of tools like OP-XFORM, JUKE, and WSL-MPT underscores its application in natural hazard analysis, glacial processes, and land change assessments. Despite challenges in ground control point placement, especially in coastal areas, and the reliance on optimization algorithms promises enhanced robustness and efficiency, addressing limitations of previous methods.
To process georeferenced data, emerging data-driven approaches such as Dynamic Mode Decomposition (DMD) are valuable in revealing hidden patterns and dynamics. DMD emerged from the fluid dynamics community as a technique for identifying spatial-temporal patterns in high-dimensional time-series data. Its application has expanded to diverse fields such as epidemiology, robotics, neuroscience, quantum control, power grids, and tidal dynamics, despite facing challenges like sensitivity to noise and limitations in modeling dynamics. Building on Singular Value Decomposition, DMD offers a powerful approach to extracting coherent linear behaviors over time, distinguishing itself from traditional Fourier analysis by capturing comprehensive spatial-temporal dynamics. Innovations like the total least-squares formulation and the forward-backward DMD method have enhance DMD's ability to deal with noisy data for practical applications. This study presents a novel application of DMD to field data analysis in coastal water level dynamics, marking a significant advance in understanding and predicting complex environmental systems by uncovering detailed spatial-temporal patterns crucial for coastal management and response strategies.
In this study, we are targeting developing a coastal hazard monitoring system in an area with under-developed monitoring capability to fill the data gap. Surveillance cameras were installed to monitor the water level and extent, as well as the coastal seawall. AI-based data processing algorithms were implemented to assess and map the coastal water level dynamics. The research methodology and results are helpful in informing asset owners and managers about a new method of coastal monitoring for regional vulnerability.
