Mapping land cover in urban residential landscapes using very high spatial resolution aerial photographs
2012
Al-Kofahi, Salman | Steele, Caiti | VanLeeuwen, Dawn | St. Hilaire, Rolston
Accurate information on existing residential landscapes is essential for framing ordinances and monitoring residential water use in the Urban Greenspace Ecosystem. We classified residential landscapes of New Mexico's largest city, Albuquerque, to explore the spatial distribution of residential greenspace and its composition among zip codes and median incomes. Geographic Information System (GIS) vector files including parcels, city limits, zip codes and land-use maps, were integrated with ownership information. The database was stratified by Albuquerque's 16 zip codes. Four hundred eighty residential landscapes were selected randomly for study. Very high spatial resolution (0.15m) 2008 true color aerial photographs and the object-oriented supervised classification module in ENVI EX were used to identify residential features. Spatial and textural variables, created by image segmentation, were classified using the K-Nearest Neighbor (K-NN) algorithm embedded in ENVI EX. Classification accuracy was 89%. Larger greenspace, tree, shrub, and grass areas were in larger parcels. Landscapes in lower income groups and older zip codes include larger greenspace and tree cover because of mature tree sizes, while grass dominated landscapes of higher income groups and newer zip codes. This knowledge of residential vegetation distribution could serve as a basis for policy makers, planners, and water conservation officers wishing to enact ordinances and regulations that govern the urban residential landscape.
显示更多 [+] 显示较少 [-]