Assessment of different methods of land use changes detection using satellite images (Case study: Dry region of Mehran)

Document Type : Research Paper

Authors

Abstract

Mapping Land use from satellite images facilitate to reveal changes over a period. There are different methods to detect the changes. In this study, Landsat satellite images have been used for the 1988-2007 periods, and six techniques were applied to detect changes in desert areas of Mehran in Ilam province with an area of 146,265 hectares. Changes detection techniques used in this study are Including of image differencing, principal components analysis, the normalized difference land cover index, conventional components analysis, difference Tasseled Cap (light) and post-classification method. In this study the threshold in the methods that requier the threshold were determind statistically. On this basis, it is known that the threshold change in the region is ±1 standard deviation from the mean. After detecting the change threshold, areas with decrease and increase in changes and no change were identified. In order to assess the accuracy of the changes detection techniques ground truth map and field data were used and overall accuracy and Kappa coefficient wewre calculated. Based on the results it was found that the difference method of infrared band with overall accuracy and Kappa coefficient of, respectively, 95.66 and 0.94 was the most accurate, and difference method of CCA2 with overall accuracy and Kappa coefficient of, respectively, 29 and 0.16 was the least accurate method in detecting changes in land use in the area under the study.

Keywords

Journal of Arid Biome
Volume 3, Issue 1 - Serial Number 1
January 2013
Pages 56-68

Files

Share

How to cite

Statistics