Land Use Mapping by Classification of Landsat Images Using Synthetic Method

Land use mapping is very important in optimal land management. There are different methods of land use mapping that each of them has its own advantages and disadvantages. This research aimed to land use mapping by combining supervised, unsupervised, object-oriented, spectral methods and using vegetation and soil indices and ancillary maps (DEM and slope) in the Fandoghlo region.

This research was carried out by a descriptive-analytical method in the Fandoghlo region. At the first, the Landsat8 image was pre-processed and then the normalized difference vegetation index and brightness index were extracted and stacked with image bands. Agriculture, fallow, pasture, and water body classes were extracted using the maximum likelihood method. Then matched filtering analysis and density slice methods were used to extract barren lands and forest classes respectively. Residential and garden classes’ were extracted using object-oriented method. Finally, using expert knowledge and recognition of the study area and based on ancillary maps (slope and DEM), the final land use map was extracted.

The overall accuracy and Kappa coefficient of the synthetic method were 97.28% and 0.96, respectively, that is more suitable and practical for land use planning and management than the conventional methods of satellite image classification and land use mapping.

Based on the results of this research, it can be said that in order to land use mapping, according to the characteristics of each region in terms of homogeneity and the number of classes, as well as the type of satellite image, it is better to use different combinations of methods, indices, and ancillary maps are used for land use mapping.