“…Methods such as artificial neural networks, expert systems, vegetation-impervious surface-soil (VIS) classifications, and support vector machines have been widely applied in urban land-use classifications (Foody, 2000;Pacifici, Chini, & Emery, 2009;Pal & Foody, 2010). Nonetheless, the resolution of the imagery and the heterogeneity characteristic of urban landscapes make it difficult to automatically map detailed urban lands solely using optical remote sensing methods (Cockx, Voorde, & Canters, 2014). The use of ancillary datasets such as census data, road networks, impervious surface coverages, landscape metrics, land parcel attributes, and radar data were recently documented to improve urban classifications (Abed & Kaysi, 2003;Berger et al, 2013;Chaudhry & Mackaness, 2008;Hermosilla, Palomar-V azquez, Balaguer-Beser, Balsa-Barreiro, & Ruiz, 2014;Schneider, Friedl, & Potere, 2014;Soergel, 2010;Wu, Qiu, Usery, & Wang, 2009).…”