GCL_FCS30: a global coastline dataset with 30-m resolution and a fine classification system from 2010 to 2020

Area of Science:

• Earth and Environmental Sciences
• Geographic Information Systems
• Remote Sensing

Background:

• Coastlines are crucial indicators of coastal processes and dynamics.
• Existing global coastline datasets primarily focus on morphology, lacking detailed typological classification.
• A comprehensive classification of global coastlines is needed for better coastal management and research.

Purpose of the Study:

• To develop the Global CoastLine Dataset (GCL_FCS30), featuring a detailed classification system for global coastlines.
• To provide a continuous and smooth line vector format dataset covering high latitudes.
• To address the gap in global coastline typology datasets.

Main Methods:

• Coastline extraction using a combined algorithm: Modified Normalized Difference Water Index (MNDWI) and adaptive threshold segmentation.
• Coastline classification via a hybrid transect classifier integrating a random forest algorithm with multi-source geophysical data.
• Validation using location data and third-party datasets to assess accuracy and spatial relevance.

Main Results:

• The GCL_FCS30 dataset successfully captures various coastline types, including artificial coastlines, with high accuracy.
• Achieved an overall accuracy exceeding 85% and a Kappa coefficient over 0.75 for coastline classification.
• Demonstrated strong alignment with validation data and high spatial relevance for each coastline category.

Conclusions:

• The GCL_FCS30 is the first global dataset providing detailed coastline typologies, particularly valuable for high-latitude regions.
• The dataset's accuracy and comprehensive classification offer significant advantages for coastal research and management.
• This novel dataset enhances our understanding of global coastal environments through continuous, smooth vector data.