Development of a PCA-based land use/land cover classification utilizing Sentinel-2 time series

Abstract

Land use/ land cover mapping and characterization is required for resource management and planning. In this aspect, remote sensing methods can be employed to classify the land use/ land cover classes over selected areas in an effective and economical manner compared to traditional surveys. In this research an improved supervised classification scheme for Sentinel-2 images classification of a selected area in El-Beheira governorate was developed. Field survey was carried out to collect ground truth data for from December 2020 to March 2021. The supervised classification was preformed after applying various principle component analyses (PCAs) on the used sequential Sentinel-2 images within the winter season. The results revealed that utilizing the proposed image classification technique, an overall accuracy of 86.8% could be achieved for the produced Land Use/Land cover map. The agricultural area covered about 89% of the studied area and was occupied by seven crops. Wheat and Egyptian clover were the major crops and covered about 67% of the studied area while green beans, potato and citrus covered about 21%.