Mapping Land Cover in the Dead Sea Basin from Landsat TM Satellite Imagery

Mahmoud M. Rababa’a, Jawad T. Al-Bakri


A study was carried out to investigate the use of digital classification techniques of Landsat TM imagery to map vegetation pattern and land cover in the Dead Sea basin. Unsupervised classification methods, K-means and ISODATA, and density slicing of the Normalized Difference Vegetation Index (NDVI) were applied to the TM imagery to derive digital maps of land cover in the study area. Output maps were checked and verified in the field to assess their accuracy and to correct area estimate of each output class. Results showed differences in the overall accuracy of the three methods with the highest accuracy of 78%, obtained from NDVI method. Mapping accuracy was also variable at the class level, with lowest mapping accuracy of urbanized areas. Area estimate, therefore, was corrected using ground data included in the confusion matrices, originally used in accuracy assessment of maps. Results of area estimate were improved when correction of area estimate was made. The study suggests the possibility of using remote sensing data to map land cover and emphasize the role of ground data in correcting output maps.

الكلمات المفتاحية

Remote sensing, Unsupervised classification, NDVI, Mapping accuracy

النص الكامل:

PDF (English)

المراجع العائدة

  • لا توجد روابط عائدة حالياً.