Identifying mangrove species and their surrounding land use and land cover classes using object-oriented approach with a lacunarity spatial measure

Soe Myint, Chandra P. Giri, Le Wang, Zhiliang Zhu, Shana C. Gillete

Research output: Contribution to journalArticle

68 Scopus citations

Abstract

Accurate and reliable information on the spatial distribution of mangrove species is needed for a wide variety of applications, including sustainable management of mangrove forests, conservation and reserve planning, ecological and biogeographical studies, and invasive species management. Remotely sensed data have been used for such purposes with mixed results. Our study employed an object-oriented approach with the use of a lacunarity technique to identify different mangrove species and their surrounding land use and land cover classes in a tsunami-affected area of Thailand using Landsat satellite data. Our results showed that the object-oriented approach with lacunarity-transformed bands is more accurate (over-all accuracy 94.2%; kappa coefficient = 0.91) than traditional per-pixel classifiers (overall accuracy 62.8%; and kappa coefficient = 0.57).

Original languageEnglish (US)
Pages (from-to)188-208
Number of pages21
JournalGIScience and Remote Sensing
Volume45
Issue number2
DOIs
StatePublished - Apr 1 2008

    Fingerprint

ASJC Scopus subject areas

  • Earth and Planetary Sciences(all)

Cite this