Recent advancements in satellite remote sensing have greatly enhanced the monitoring capabilities of coastal environments, primarily through improvements in sensor resolution and data processing techniques. This progress is particularly evident in algorithms designed for the (semi)automatic detection of shorelines on high-resolution images. However, existing algorithms are often tested under optimal conditions on sandy beaches, lacking validation for shorelines with diverse materials such as vegetation debris. This study aims to validate Satellite-Derived Shorelines (SDS) obtained from the SAET and CoastSat algorithms, applied to Sentinel-2 (S2) multispectral images. The validation process involves comparison with RTK-GNSS surveys conducted during fieldwork at Arborea and Mare Morto beaches in the Gulf of Oristano, Sardinia, carried out during fieldwork campaigns as part of the OVERSEE project. The chosen beaches present different types of exposure due to their positions in the Gulf of Oristano, as well as various morphologies controlled by the presence or absence of banquettes of Posidonia oceanica. The study focuses on analysing the effect of these biomass deposits on the accuracy of shoreline detection algorithms. The conducted analysis revealed that the SAET algorithm outperformed CoastSat, demonstrating higher accuracy in shoreline detection on both beaches. It exhibited performance more accurate on Arborea beach, despite the greater challenges posed by the presence of the Banquette. The study underscores the robustness of the SAET algorithm in handling these challenges and highlights its effectiveness in generating accurate SDSs, even in complex coastal environments. Additionally, the study emphasizes the advantages of adopting different algorithms with distinct image processing modes and spectral indices, providing valuable insights into their accuracy levels and constraints.
XX Congreso de la Asociación Española de Teledetección y Cambio Global: Retos y Oportunidades para un Crecimiento Azul
Terracciano S.Primo
;Brunetta R.;Ciavola P.Penultimo
;
2024
Abstract
Recent advancements in satellite remote sensing have greatly enhanced the monitoring capabilities of coastal environments, primarily through improvements in sensor resolution and data processing techniques. This progress is particularly evident in algorithms designed for the (semi)automatic detection of shorelines on high-resolution images. However, existing algorithms are often tested under optimal conditions on sandy beaches, lacking validation for shorelines with diverse materials such as vegetation debris. This study aims to validate Satellite-Derived Shorelines (SDS) obtained from the SAET and CoastSat algorithms, applied to Sentinel-2 (S2) multispectral images. The validation process involves comparison with RTK-GNSS surveys conducted during fieldwork at Arborea and Mare Morto beaches in the Gulf of Oristano, Sardinia, carried out during fieldwork campaigns as part of the OVERSEE project. The chosen beaches present different types of exposure due to their positions in the Gulf of Oristano, as well as various morphologies controlled by the presence or absence of banquettes of Posidonia oceanica. The study focuses on analysing the effect of these biomass deposits on the accuracy of shoreline detection algorithms. The conducted analysis revealed that the SAET algorithm outperformed CoastSat, demonstrating higher accuracy in shoreline detection on both beaches. It exhibited performance more accurate on Arborea beach, despite the greater challenges posed by the presence of the Banquette. The study underscores the robustness of the SAET algorithm in handling these challenges and highlights its effectiveness in generating accurate SDSs, even in complex coastal environments. Additionally, the study emphasizes the advantages of adopting different algorithms with distinct image processing modes and spectral indices, providing valuable insights into their accuracy levels and constraints.I documenti in SFERA sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
