Investigating and Monitoring Land Use Changes Using Geographic Information System, Remote Sensing Technique and Supervised Classification Methods (Case Study: Swadkoh City)

Journal of Radar and Optical Remote Sensing and GIS

مقاله 4، دوره 5، شماره 1، خرداد 2022، صفحه 39-51

نوع مقاله: مقاله پژوهشی

نویسندگان

Razyeh Shaban Mirfazlolah^{* 1}؛ Amin Mohamadi deh Cheshmeh²

¹Employee of document registration office of Mashhad, Mashhad, Iran

²Expert of Mapping Engineering

چکیده

Investigation and analysis of land use changes was done using remote sensing and GIS techniques with supervised classification methods. The selected images from the years 2000 and 2022 were taken by the Landsat satellite. Necessary pre-processing of the images was done and then the best band combination was selected. The best band combinations of 2000 and 2022 were selected as 245 and 467, respectively, using the OIF index. The area changes from 2000 to 2022, in the support vector machine method, the uses of dense pasture, poor pasture, agriculture, residential, forest have had area changes of 9580.53, 34267.49, 237.2, 1603.41, 26527.57 hectares. Therefore, the use of dense pasture and forest has decreased by 5.87% and 16.25%, and other uses have increased. The area changes from 2000 to 2022, in the neural network method, the uses of dense pasture, poor pasture, agriculture, residential, forest have had area changes of 6021.05, 33869.57, 360.79, 1492.16, 29701.47 hectares. Therefore, the use of dense pasture and forest has decreased by 3.69% and 18.20%, and the use of poor pasture has increased by 20.75%, agriculture by 0.22%, and residential by 0.91%. In the assessment of classification accuracy, kappa coefficient and overall accuracy in the support vector machine method in 2000 were 0.84 and 0.87 and in 2022, 0.86 and 0.88 were obtained. Kappa coefficient and overall accuracy were obtained in 2000, 0.94 and 0.95 and in 2022, 0.96 and 0.97 in the neural network method. Therefore, the neural network method has higher accuracy.

کلیدواژه‌ها

Land use؛ Kalardasht city؛ remote sensing؛ neural network