Performance of Supervised Classification for Mapping Land Cover and Land Use in Jeffara Plain of Libya




Conference paper

Conference title


Vol. 7 No. 55


Mukhtar Mahmud Elaalem


Different methods accessible for remote sensing image classification; they include supervised, unsupervised and fuzzy classifications. This paper investigates the performance of the supervised classification on remote sensing data for land cover/ land use in North-West Region of Jeffara Plain of Libya. The study used SPOT 5 satellite image taken on January 2009 as a main data. Maximum likelihood classification (MLC) which is based on the probability that a pixel belongs to a particular class were chosen to classifying land cover data/ land use in the study area. The land cover /land use classes for the study area were classified into 5 homogeneous land cover classes of a single land cover and 4 heterogeneous land cover classes. Ground verification was applied to verify and evaluate the accuracy of supervised classification. 48 field points were collected using Systematic Random Sampling. The results showed that 65 % of the study area was classified into heterogeneous land cover classes, while 25 % of the study area classified into homogeneous land cover classes. Derivation heterogonous or mixed land cover classes from the use supervised classification (i.e. crisp classification) led to producing uncertain or vague land cover classes in the study area. For future work, the authors will test the fuzzy image classification to derive land cover and land use data in the study area. Keywords: Supervised classification, spot image, land cover /land use, Jeffara Plain

Publisher's website