International Journal of Environmental Sciences

Volume 4 Issue 5 2014- March 2014    Pages: 945-955  <<Previous    Next>>

Transformation of satellite image data in class modeling of land use/cover of agriculture and forestry in tropical area

Author Information:

Akhbar1, Muhammad Basir2, BungaElim Somba2 and Golar1
1 - Faculty of Forestry, Tadulako University, Palu 94119, Central Sulawesi, Indonesia
2 - Faculty of Agriculture, Tadulako University, Palu 94119, Central Sulawesi, Indonesia


Development of classification model of land use/cover of agriculture and forestry in tropic areas using 30m resolution satellite image  (Landsat 7 ETM+ image) and 10m resolution image  (SPOT 5 XS image)is an urgent requirement. In relation to that point, research on image data transformation was conducted to seek the best class models by using Isodata method. This method was used to extract image pixel spectral value as a reference of making land use/cover class done in wide coverage area (>100,000 ha) and small coverage area (<50,000 ha) in Palu City and Sigi Regency,Central Sulawesi, Indonesia. From 3 class models (25, 50, 75 classes) tested, 50 classesmodel is the best and can be reliable in extracting image pixel spectral value and land use/cover class segmentation. The mean of pixel split power 1,817.64 to 1,972.08 (moderately to well separated) with split power of land use type 37.03% and land cover type 84.64% from the total area 117,931.32 ha by using Landsat image. Furthermore, 50 classesmodel in small coverage area (<50,000 ha) can improve split power and land use and land cover class segmentation respectively 83.31% and 99.08% by using Landsat image, as well as 67.10% and 77.76% by using SPOT 5 XS image. Accuracy test result of land use/cover class used confusion matrixmethod obtainedoverall accuracy72% to 88%, andkappa statistical value 0.7174 to 0.8774.

Keywords: Image, extraction, class segmentation, land use/cover.


© 2014 Copyright by the authors, licensee Integrated Publishing Association.This is an open access article distributed under the Creative Commons Attribution License (3.0) which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

The electronic version of the article can be downloaded below

Download full text pdf

SocialTwist Tell-a-Friend