“…Wu and Murray (2003), and Wu (2004) compared several spectral unmixing models for estimation of impervious surface distribution in Columbus, OH, and found that the three end-member V-I-S model (including brightness normalization) was the most effective. Other studies, however, have demonstrated the effectiveness of a variety of different end-member combinations, such as high-albedo, low-albedo, and vegetation (Small, 2002), the inclusion of shade as an end-member (Alberti et al, 2004;Lu & Weng, 2004), and allowance for variable numbers and types of end-members (Dennison & Roberts, 2003;Rashed et al, 2003). Lu and Weng (2004) demonstrated that green vegetation, shade, and soil or impervious surface were the most effective end-members for spectral unmixing in Indianapolis, IN.…”