Global Urban Area Mapping using Global ASTER Satellite Images
Title | Global Urban Area Mapping using Global ASTER Satellite Images |
Publication Type | Conference Proceedings |
Year of Conference | 2010 |
Authors | Miyazaki, H, Shao, X, Iwao, K, Shibasaki, R |
Conference Name | 31th Asian Conference on Remote Sensing |
Pagination | TS07–2 |
Conference Location | Hanoi |
Keywords | ASTER, cover classification, high resolution, land, LLGC, logistic regression, urban area mapping |
Abstract | We present development of global urban area map using ASTER satellite images, which has much higher resolution than that of existing global urban area maps. To complete the map for world’s cities, we developed automated algorithm for mapping urban area in high resolution. The algorithm consists of two steps: classifying pixels of ASTER satellite images into urban or non-urban by Learning with Global and Local Consistency (LLGC) technique; and integration with existing urban area maps using logistic regression. We implemented the algorithm and demonstrated it on 775 scenes of ASTER satellite images. LLGC classified pixels of ASTER satellite images into urban or non-urban in 15-m resolution, though it had considerable amount of misclassification due to similarity in surface reflectance among non-vegetative land cover. To complement the misclassifications, we integrated LLGC results with existing urban area maps using logistic regression. The misclassifications were corrected well, especially in dry zone. We also developed ground truth database using global gazetteer of world’s cities so that we conduct comprehensive accuracy assessment on the urban area map. We visually interpreted land cover of urban or non-urban on 3734 points coordinates derived from global gazetteer, and combined them with 4211 data points of Degree Confluence Project into a database, which had 2185 data points of urban and 5559 of non-urban. Accuracy assessment using the database indicates that our map is more accurate than existing urban area maps. Finally, we applied the method on broad coverage of ASTER satellite images rather than single one scene. The result showed quite well classification as a whole, indicating possibility of developing global urban area map in high-resolution; however considerable problems due to availability of cloud-free satellite image is still remained. |
- Log in to post comments
- Google Scholar
- BibTeX
- RTF
- Tagged
- EndNote XML
- RIS