The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Download
Publications Copernicus
Download
Citation
Articles | Volume XLII-4/W4
https://doi.org/10.5194/isprs-archives-XLII-4-W4-123-2017
https://doi.org/10.5194/isprs-archives-XLII-4-W4-123-2017
26 Sep 2017
 | 26 Sep 2017

VEGETATION MONITORING OF MASHHAD USING AN OBJECT-ORIENTED POST CLASSIFICATION COMPARISON METHOD

N. Khalili Moghadam, M. R. Delavar, and A. Forati

Keywords: Vegetation Reduction, Urban Development, Mashhad, Landsat Satellite Images, Object-Oriented Post Classification Comparison

Abstract. By and large, todays mega cities are confronting considerable urban development in which many new buildings are being constructed in fringe areas of these cities. This remarkable urban development will probably end in vegetation reduction even though each mega city requires adequate areas of vegetation, which is considered to be crucial and helpful for these cities from a wide variety of perspectives such as air pollution reduction, soil erosion prevention, and eco system as well as environmental protection. One of the optimum methods for monitoring this vital component of each city is multi-temporal satellite images acquisition and using change detection techniques. In this research, the vegetation and urban changes of Mashhad, Iran, were monitored using an object-oriented (marker-based watershed algorithm) post classification comparison (PCC) method. A Bi-temporal multi-spectral Landsat satellite image was used from the study area to detect the changes of urban and vegetation areas and to find a relation between these changes. The results of this research demonstrate that during 1987-2017, Mashhad urban area has increased about 22525 hectares and the vegetation area has decreased approximately 4903 hectares. These statistics substantiate the close relationship between urban development and vegetation reduction. Moreover, the overall accuracies of 85.5% and 91.2% were achieved for the first and the second image classification, respectively. In addition, the overall accuracy and kappa coefficient of change detection were assessed 84.1% and 70.3%, respectively.