Hierarchical land use and land cover classification of Sentinel2-A images and its use for corine system

Thesis Type: Postgraduate

Institution Of The Thesis: Orta Doğu Teknik Üniversitesi, Faculty of Engineering, Department of Civil Engineering, Turkey

Approval Date: 2017




The aim of this thesis is to investigate the potential of Sentinel-2 satellite for land use and land cover mapping. The commonly known supervised classification algorithms, support vector machines (SVMs) and maximum likelihood classification, are adopted for investigation along with a hierarchical classification model CORINE. The main classes for land cover and mapping are selected as water, vegetation, built-up and bare-land in the first level, which is followed by inland water, marine water, forest/meadow, vegetated agricultural land, barren land and non-vegetated agricultural land in the second level. The study area for the experiments are selected as the two biggest cities of Turkey, namely Ankara and Izmir, providing sufficient number of classes for comparison purposes. During the utilized methodology, water and vegetation are first extracted by using the normalized difference water and vegetation indexes. Then, sufficient number of pixels are collected from the remaining parts for the first and second level classifications to perform a training and comparison for supervised learning algorithms. The experimental results first indicate that the support vector machines are significantly superior to the maximum likelihood classification with an average vi of 8 percent accuracy rates. Second, the hierarchical classification is also superior to non-hierarchical classification with the gains between 4 to 10 percent. The overall accuracy rates of the proposed hierarchical methodology are obtained as 85 % and 84 % for the first level classes and 84 % and 72 % for the second level classes, respectively for Izmir and Ankara.