8 févr. 2013

Conférence ICS 2013, Plymouth - Télédétection THR des blocs coralliens


Mapping the geomorphic impacts of high-energy events on coral reefs using very high resolution remote sensing: a case study in French Polynesia.
Antoine Collin, S. Etienne
12th International Coastal Symposium, Plymouth University, 8-12 April 2013



Remote sensing (from satellite imagery to airborne sensor) has been widely used for characterizing coastal environments: geomorphological mapping, vegetation mapping, ecological studies, coastal zone management, etc., but no study has yet focused on the use of satellite imagery for coastal boulder classification. This can be explained by the pixel resolution limit which was not in accordance with boulder size. However, very high resolution (VHR) imagery at submeter scale is now available so that the level of spectral mixing encountered by previous, lower resolution imagery becomes manageable. VHR remote sensing imagery allows the generation of thematic maps over large extents based on image classification. In this study, we examine the interest of VHR multispectral satellite imagery (WorldView-2, 8 bands at 50 cm/ 2 m spatial accuracy) for identifying and classifying coral boulder deposited by tropical cyclones on the reef flat of Tetiaroa atoll, French Polynesia. 
The methodological approach is innovative because it harnesses the biophysical properties of novel spectral while using a cutting-edge classifier, the Support Vector Machine. The resulting classification discriminates submerged coral colonies from emerged ones, as well as emerged coral boulders from emerged reef flat (palaeo sea-level). It therefore highlights the great interest of the new spectral bands offered by WorldView-2 imagery for coastal studies, especially the 400-450 nm “coastal” band and the 860-1040 nm “near infrared 2” band. Coastal boulder is a singular sediment class mobilized during high-energy events (tsunamis, tropical cyclones, storms) only. Hence, VHR remote sensing appears as a flexible and useful tool for post-catastrophic marine inundation mapping.

Book of abstracts here

Aucun commentaire: