U-Net-based time-series mapping of tidal flat areas and carbon sink variations under fluctuating tidal levels

초록

Tidal flats are crucial for carbon storage but are threatened by human activities. Satellite remote sensing data and deep learning methods help extrapolate field-based carbon measurements to broader spatial and temporal scales. However, inconsistent tidal heights during satellite image acquisition can bias carbon sink estimates. This study employed a U-Net model to map tidal flat areas and estimate their organic carbon sequestration rates from 2016 to 2024 in Seocheon, South Korea. Tidal flats were delineated at the lowest tidal height available each year from Sentinel-2 data and calculated carbon sequestration using a locally derived coefficient. Results showed tidal levels fluctuating between 51.0 and 202.5 cm caused tidal flat areas to range from 53.2 km & sup2; to 71.5 km & sup2;, corresponding to changes in carbon sequestration from 734 Mg C yr-& sup1; to 989 Mg C yr-& sup1;. These findings emphasize the need for a fixed tidal height for consistent carbon sink monitoring.

키워드