Details

Autor(en) / Beteiligte

• Küçük Matci, Dilek

Titel

DEVELOPMENT OF A NEW INDEX FOR MAPPING URBAN AREAS in TÜRKİYE USING SENTINEL-2 IMAGES

Ist Teil von

• Advances in space research, 2023-12, Vol.72 (11), p.4677-4691

Ort / Verlag

Elsevier B.V

Erscheinungsjahr

2023

Link zum Volltext

Quelle

Elsevier ScienceDirect Journals Complete

Beschreibungen/Notizen

• Population growth and urbanization have been increasing simultaneously and rapidly in recent years. Urban growth causes changes in land use/cover. Monitoring and mapping these changes are important for issues such as disaster management, biodiversity and city and environmental planning and management. Images obtained by remote sensing methods provide data for mapping urban land use areas. Over the years, many approaches have been developed to accurately analyse these images. However, urban area mapping is still a problem due to its complex structure, the many different materials used in urban areas, artificial and natural green areas, roads and industrial zones. In this study, a new approach is proposed to accurately map urban areas. Band values of Sentinel-2 images are analysed in regions (Eskişehir, İnegöl area from Bursa and Sincan area from Ankara) belonging to various land classes in urban areas. Based on the results, a new index (Urban Area Index) was proposed. The performance of the index was evaluated by separability analyses (Bhattacharyya-distance, Jeffries-Matusita distance and Mahalanobis-distance). Besides the created index was classified with the Random Forests algorithm and an accuracy assessment was performed with the overall accuracy, producers’ and users’ accuracy and F1 score. Results obtained by classification using this index are compared to 17 indexes presented in the literature. The proposed index achieved an overall accuracy rate of over 82% for all study areas. This rate is higher than the overall accuracies obtained with other indices and raw Sentinel-2 bands. This result indicates that the proposed method successfully maps urban areas more effectively than existing methods.