Application of deep learning model and DSAS in assessing coastline changes of Da Nang Bay

Abstract

The effects of climate change and the rapid increase of urbanization are exerting pressure on coastal areas. Meanwhile, coastal accretion and erosion are natural hazards that affect coastal communities, particularly in Da Nang Bay, where both processes are influenced by natural dynamics and anthropogenic activities. The study aims to assess coastline changes in Da Nang Bay under the influence of storms and urbanization. The authors used DSAS and the deep learning UNet model on high-resolution UAV images to analyze coastline change during the period from 2002 to 2024. The results show that the UNet model achieved an overall accuracy of 98.5% with a Kappa coefficient of 0.97, demonstrating the effectiveness of the combined UNet–DSAS approach in shoreline change analysis. The southern area of the Cu De River tends to experience severe erosion, with a maximum rate of 3.43 m/year. In contrast, the Cu De River mouth shows a tendency toward accretion, which prevented transportation there. Land reclamation has caused the Da Phuoc urban area to grow quickly, with a maximum pace of 22.78 m/year. The study showed that the UNet model improved the accuracy of coastline detection. The results provide an effective methodological framework for coastal planning and management, and offer scientific evidence to support sustainable coastal planning and protection in Da Nang Bay.