Pixel Based Classificatrion Analyisis of Land Use Land Cover in Tarom Basin

The comprehensive management of a watershed requires basic information such as land use and land cover. The aim of this study is to conduct accuracy analyses of LULC classifications derived from Landsat-8 data, and to reveal that which kind of land use and land cover can be estimated more accurately. Tarom Basin and its near surrounding was selected as study area for this case study. Landsat-8 the data, acquired on 8 August 2017, were utilized as satellite imagery in the study. The RGB and NIR bands were used for classification. Required pre-processing and control of georeferenced of images were performed. After performing the required atmospheric corrections, using the FLAASH algorithm, classification maps were generated. LULC images were generated using 3 pixel-based supervised classification method method, Maximum Likelihood (MLC), Support Vector Machine (SVM) and Artificial Neural Network (ANN). As a result of the accuracy assessment, kappa statistics and overall accuracy for MLC method were 0.88 and 91.55 respectively. The obtained results showed that Landsat-8 OLI data, presents satisfying LULC images in water body, mountain and rock, bare land, Vegetation and forest classes. In addition, according to the obtained results, it can be stated that all three methods of classification in a region with heterogeneous (in terms of elevation elevation between 280 and 3000 m and land use and variety of vegetation) Such Tarom, can have good results. Among these methods, Classification with MLC method, had higher speed and lower complexity for execution, than two other methods in achieving the required maps.