Mapping of Land Use / Land Cover of Klerksdorp–Orkney–Stilfontein–Hartebeestfontein (KOSH) Region for the Years 2020 and 2019 Using Sentinel-2 Data
Year: 2020
Type: conference-paper
Abstract: The article presents mapping of current land use / land cover and of year 2019 of Klerksdorp–Orkney–Stilfontein–Hartebeestfontein (KOSH) region using Sentinel-2 data. This study made use of VNIR bands of Sentinel-2 acquired on 10 March 2020, 16 March 2019 and National Land Cover (NLC) dataset of year 2018 to investigate how land use change and vegetation alteration has occurred during 2018 to 2020. The Level 1 Sentinel-2 data were pre-processed to obtain ground reflectance using Sen2Cor algorithm. The classification system of NLC 2018 was used for identifying the training sites for 56 classes present. Further processing to produce intended classes and post processing of classification products were performed using Geomatica Object Analyst. The analysis involved image segmentation, attribute calculation for geometrical parameters, mean, standard deviation and 10 vegetation indices of pixels covered in segments of VNIR bands, training site selection, classification and post classification analysis. Though attempts were made to classify the images with training sites and different attributes, the best classification result was obtained with standard deviation of 4 VNIR bands and 10 vegetation indices. For classifying March 2019 image, the batch classification algorithm was applied using training model file generated for classification of March 2020 data. The percentage of total area covered by five major classes for the years 2020 and 2019 respectively are: grasslands - 49.32% & 42.14%; temporary crops (rainfed) - 32.90% & 32.92%; open woodland – 4.00% & 3.03%; mine waste (tailings) and resource dumps – 1.96% &2.46%; residential formal (grass) – 1.53% & 1.46%.
Author: A. Thomas
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
SDGs: Life on land, Climate action
Open Access status: green
OpenAlex ID: W4400891720