The Saemangeum project resulted in the loss of approximately 208 km² of tidal flats, equivalent to about 8% of South Korea's total tidal flats. This loss has reduced carbon storage capacity by approximately 40,000 tons. Halophytes in the tidal flats ...
The Saemangeum project resulted in the loss of approximately 208 km² of tidal flats, equivalent to about 8% of South Korea's total tidal flats. This loss has reduced carbon storage capacity by approximately 40,000 tons. Halophytes in the tidal flats play a core role as blue carbon ecosystems. Quantifying their distribution status and carbon absorption capacity is essential but remains underexplored. This study investigates halophyte communities in the Saemangeum area using drone-based multi-sensor imagery combined with an artificial intelligence (AI) segmentation model to estimate blue carbon potential. Aerial photography using the DJI M300 RTK and Zenmuse P1/L1 was conducted to collect high-resolution orthophotos and digital surface model (DSM) data. For the species classification and area estimation of target halophytes, YOLOv11-seg and HRNet models were applied. Analytical results indicate that YOLOv11-seg demonstrates high detection speed and accuracy performance. At the same time, the HRNet model exhibits strengths in segmentation boundary extraction, confirming its potential for future use as a halophyte detection model. AI detected key dominant species of Suaeda japonica, Suaeda glauca, and Phragmites communis and presented their distribution patterns. In the Eco-Environmental Complex, the carbon sequestration amounts were found to be approximately 0.196 tC for Suaeda japonica and 0.012 tC for Suaeda glauca. In the Haechang tidal flat, the carbon sequestration amounts of Phragmites communis, Suaeda glauca, and Suaeda japonica were found to be 164.20 tC, 37.03 tC, and 24.40 tC respectively. These results suggest a useful tool for quantifying carbon sequestration in tidal flats, establishing a halophyte carbon inventory, and formulating a net-zero policy. This work suggests an AI-based halophyte monitoring platform that can contribute to managing sustainable coastal ecosystems.