The extraction of mangrove within intertidal zone based on multi-temporal HJ CCD images

Abstract

The coastal zone in Beibu Gulf is dominated by diurnal tide and there exists the largest mangrove community in China. The frequently used mangrove extraction methods seldom took the tide influence into account which would lead to extracted area loss on single instantaneous remote sensing image. The loss cannot be ignored when the mangrove submerged time is long. This study took one portion of Beibu gulf coastline as research site. Four temporal HJ CCD images with different tide levels were selected for inundation mangrove extraction and coastal terrain classification. Based on the analysis of targets image-spectra, several decision factors were proposed, and subsequently a multi-layer decision tree was constructed. After the classification, target distributions at research site including the submerged mangrove were acquired. The overall classification precision was high up to 91.79%, and the Kappa coefficient was 0.9064. The obtained submerged mangrove area was 2.155 km2, which comprised 4.5% of total mangrove area and would be lost if the extraction were only applied on single image.