Extraction of tidal channel networks from airborne scanning laser altimetry

摘要: Tidal channel networks are important features of the inter-tidal zone, and play a key role in tidal propagation evolution salt marshes flats. The study their morphology is currently an active area research, number theories related to have been developed which require validation using dense extensive observations network forms cross-sections. conventional method measuring cumbersome subjective, involving manual digitisation aerial photographs conjunction with field measurement depths widths for selected parts network. This paper describes semi-automatic technique extract from high-resolution LiDAR data zone. A multi-level knowledge-based approach has implemented, whereby low-level algorithms first fragments based mainly on image properties then high-level processing stage improves domain knowledge. adopted at low level uses multi-scale edge detection detect edges, associates adjacent anti-parallel edges together form channels. higher includes repair mechanism. algorithm may be extended as well data. Its performance illustrated two sites, River Ems, Germany Venice Lagoon. For Ems data, error omission automatic extractor 26%, partly because numerous small channels lost they fall below threshold, though these less than 10 cm deep unlikely hydraulically significant. commission lower, 11%. Lagoon 14%, but 42%, due difficulty interpreting natural scenes. As benchmark, previous work shown that this type specifically designed extracting able achieve substantially improved results compared those obtained standard drainage extraction Digital Terrain Models.