Identifying coastline positions and types based on both the moisture content and feature knowledge of ground objects

The coastline is one of the most important basic geographical elements in the coastal zone. Traditional methods often fail to accurately identify coastline locations due to the instantaneity of remote sensing imaging and the dynamics of tidal waters. This study was conducted to develop a model for identifying coastline locations and types that considers both the moisture content and feature knowledge of ground objects based on a long-time series of satellite remote sensing images. The validation test showed that: (1) the model could accurately identify different coastline types with clear water-land boundaries and precise spatial positions; (2) the average distances between the coastlines identified by the proposed method and the true coastlines was 3.82 m with a root mean square error of 8.78 m, with 97.56% of the distance errors being less than one pixel width; (3) from 1985 to 2022, the total coastline length increased from 2,152.42 km to 2,264.79 km, with an average annual increase of 3.03 km per year. A clear trend of coastline artificialization was observed, with the proportion of natural coastlines decreasing from 87.31% to 63.07%. This study provides technical support that will enable accurate extraction of coastline remote sensing information and has significant implications for scientific management and sustainable development of coastal zone resources.