COASTAL ZONE MAPPING FROM RADARSAT AND OPTICAL IMAGES: SHORELINE EXTRACTION, SUBMERGED VEGETATION AND BIOPHYSICAL ELEMENTS MAPPING *

摘要

Geomat International and Procean Environnement have collaborated to evaluate the potential use of RADARSAT-I fine beam mode and RADARSAT-II simulated images for shoreline mapping. RADARSAT-I images have been acquired over the Manicouagan peninsula and the Montmagny areas in the St.-Lawrence estuary, Quebec, in August 2003 coinciding with high tide events. Several shoreline extraction tools were used. That included object-oriented segmentation and various region growing algorithms, automatic vectorization and smoothing. Results showed that the different approaches have lead to an accuracy range of 5.6 to 10 m. The results on the evaluation of the potential use of both radar (RADARSAT and ENVISAT Asar) and optical images for the identification of biophysical elements in coastal zones are presented in this paper. Submerged vegetation in the Manicouagan peninsula has been studied using ASTER and IKONOS images. Based on a field survey, the extent and density (dense and sparse) of the eelgrass (Zostera Marina) beds have been mapped successfully from IKONOS images with 95% and 77% mean accuracies, respectively.