Automatic land-sea classification in a nearshore environment using satellite-based photon-counting LiDAR data

Area of Science:

• Earth Observation
• Geospatial Science
• Remote Sensing Technology

Background:

• Accurate land-sea classification is crucial for high-quality data products from the Ice, Cloud, and Land Elevation Satellite-2 (ICESat-2).
• Existing land-sea classification methods require manual input or assisted data, limiting automation for massive datasets.

Purpose of the Study:

• To develop an automated land-sea classification method for photon-counting LiDAR data.
• To introduce a novel index, the normalized photon rate-elevation ratio (NPRER), for distinguishing land and sea.

Main Methods:

• Designed the normalized photon rate-elevation ratio (NPRER) index based on land-sea differences in photon-counting LiDAR data.
• Developed an automatic classification workflow involving preliminary classification, reclassification, and post-processing enhancement.
• Validated the method using ICESat-2 data in Cook Inlet, Alaska.

Main Results:

• The NPRER index effectively measures sea appearance probability in nearshore environments.
• The automatic classification method achieved an overall accuracy of 97.98%.
• The method demonstrated robustness across different coastal types, data collection times, and feature sets.

Conclusions:

• The proposed automatic land-sea classification method offers a reliable technical solution for satellite-based photon-counting LiDAR data.
• This advancement significantly enhances the automation capabilities for processing large volumes of geospatial data.
• The NPRER index provides a valuable tool for nearshore environmental analysis.