Evaluating the Effects of Vertical Urban Forms on Diurnal Land Surface Temperatures Using Google Street View Images

Land surface temperature (LST) directly responds to diurnal surface (incoming and out-going) radiation exchanges which are strongly influenced by shading conditions and vertical urban forms. This study compares the fine scale spherical and horizontal urban form fractions in explaining the diurnal LST variations at Phoenix, AZ. We found that the spherical fractions from Google Street View provide better LST estimations than the horizontal fractions from remote sensing. The finding highlights the importance of using more comprehensive spherical measurements for urban climate research versus the conventional bird-eye view remote sensing data. In addition, geographically weighted regression (GWR) was employed to uncover the spatially varied relationships between LST, urban form, and social variables. GWR provides a considerable model improvement over the commonly used ordinary least squares regression. More importantly, it identifies the dominant factors of LST change at different neighborhoods of Phoenix, AZ, which are critical for optimizing diverse cooling strategies to combat the increasing temperature across the semi-arid city. Our study demonstrates the essential impacts of the spherical/vertical urban form on surface-urban heat island, and provides valuable insights on applying Google Street View and spatial regressions for urban climate and sustainability research.