Seyed Alireza Mousavi; Mahdi farahpour; Maryam Shokri; Karim Solaimani; Mahmood Godarzi
Volume 13, Issue 3 , February 2006, , Pages 186-200
Abstract
Landsat 7 ETM+ satellite data of 2002 versus vegetation cover map of 1976 were used to: 1- assess the capability of satellite data to prepare vegetation cover classes map and 2- study the vegetation changes trend in an area of about 26858.6 ha in Lar Dam Basin. Field Sample spots were defined after accomplishing ...
Read More
Landsat 7 ETM+ satellite data of 2002 versus vegetation cover map of 1976 were used to: 1- assess the capability of satellite data to prepare vegetation cover classes map and 2- study the vegetation changes trend in an area of about 26858.6 ha in Lar Dam Basin. Field Sample spots were defined after accomplishing necessary corrections of satellite images. Suitable band compositions were selected by considering the Optimum Index Factor (OIF), correlation matrix, Principal Components Analysis (PCA) and 2-dimensional diagram analysis. These compositions were classified using Maximum Likelihood, Minimum Distance and Box Classifier algorithms and then Majority Filter was used. Accuracy of resulted maps was evaluated by pixel to pixel method. Then Overall Accuracy Coefficient and Kappa Index were calculated. The map resulted from classification of band composition 123457 through Maximum Likelihood and Majority Filter was selected as the vegetation cover map of 2002. Vegetation cover map of 1976 prepared by Asgari-khah (1977) via field survey was used as "vegetation cover classes" map of that year. Then the changes happened in each class were assessed by operation of cross function on the mentioned maps. Due to complexity of initial classes, more homogenous classes were merged resorting to more detectable maps having only four classes. Overall Accuracy Coefficient of final map was promoted by using these classes. Then the changes happened during 1976-2002 were detected. The results showed that 28.55% of total area which was covered by rangeland vegetation types in 1976, is unchanged, 14.03% is dropped into lower and 57.42% into higher classes. The map of vegetation cover changes was produced, finally.
Mahmood Goudarzi; Mahdi Farahpour; Alireza Mosav
Volume 13, Issue 3 , February 2006, , Pages 265-277
Abstract
In Iran, like many other developing countries, high population growth rate causes unfairly uses of natural resources and consequently land cover change. Therefore, detection of land cover (rangelands, irrigated and rainfed agricultural lands, urban areas…) changes can influence local planning ...
Read More
In Iran, like many other developing countries, high population growth rate causes unfairly uses of natural resources and consequently land cover change. Therefore, detection of land cover (rangelands, irrigated and rainfed agricultural lands, urban areas…) changes can influence local planning and natural resource management. Present study efforts to find a rapid and exact method of recognition different land covers using Landsat satellite data. Methods used in this research were image enhancement, false color composite (FCC), principal components analysis (PCA) and Image classification, i.e. normalized different vegetation index (NDVI) and supervised classification. A GIS environment, ILWIS software, was used. Results showed that irrigated agriculture, rainfed agriculture, rock out crop, rangeland classes (fair, moderate, poor condition) could be separated with overall accuracy of 89%.