Long-Term Shoreline Change Analysis using Optical Satellite Images of the east coast of the Korean Peninsula

Coastal zones, as dynamic interfaces between land and sea, are critical for economic activities, ecological conservation, and human habitation. However, natural sediment systems are increasingly disrupted by artificial interventions and rising sea levels accelerated by climate change, leading to erosion and uncertainty in the stability of the coastal zone. Effective coastal management requires not only accurate monitoring but also large spatial scale and long-term temporal coverage of shoreline observations. While traditional in-situ and aerial survey methods provide high precision, they are labor-intensive and limited in scope. Optical satellite imagery emerges as a viable alternative, offering continuous, broad spatial coverage. Furthermore, advances in image processing with artificial intelligence enable shoreline extraction and long-term change analysis.

This study marks the initial step in utilizing optical satellite imagery for analyzing long-term shoreline changes along the Korean Peninsula, where artificial approaches such as gray structural coastal disaster prevention methods are mainly applied. Specifically, this study focused on the East Sea region of the Korean Peninsula, a high-risk area for coastal erosion, and examined approximately 40 years of shoreline changes using publicly available satellite data, including Sentinel-2 and Landsat series satellite images. First, an automatic satellite image download system was designed using the Google Earth Engine API, incorporating appropriate parameter settings for the region of interest and ensuring uniform data quality based on the characteristics of satellite imagery. Second, effective preprocessing techniques were applied to improve shoreline recognition from each satellite image. Third, the lower resolution of Landsat images relative to Sentinel-2 was enhanced through super-resolution generative adversarial network, enabling more precise identification of shoreline features. Fourth, the open-source software, CoastSat(Vos, 2019) was utilized to extract shorelines, and this study analyzed shoreline changes based on comprehensive coastal engineering knowledge. Finally, the feasibility of the proposed method was validated by analyzing the cross-sectional time series of the shoreline at the littoral cell of Wonpyeong-Chogok beach, an area where various coastal structures have been installed over the past decades to mitigate coastal retreat. These findings illustrated the shoreline responses and geomorphological changes resulting from the sequential construction of coastal structures.

This study underscores the potential of using elaborate image enhancement techniques, including contrast stretching, spatial registration and super-resolution, to analyze long-term shoreline dynamics with high accuracy. By applying these methods to satellite imagery spanning four decades, we provided insights into shoreline responses to sequential coastal structures. These findings contribute to supporting proactive coastal management in the face of growing uncertainties, emphasizing the importance of integrating advanced image processing and unsupervised learning for effective shoreline extraction and geomorphological analysis.

 

Reference

Vos, K., Splinter, K. D., Harley, M. D., Simmons, J. A., & Turner, I. L. (2019). CoastSat: A Google Earth Engine-enabled Python toolkit to extract shorelines from publicly available satellite imagery. Environmental Modelling & Software, 122, 104528.