Deriving urban neighborhood characteristics using spatial autocorrelation measures.

This paper addresses the detection and characterization of urban neighborhoods from remote sensing images. Object based image analysis (OBIA) procedure is implemented and spatial statistical methods are applied to create homogeneous zones. Both Getis‐Ord statistic and shape metrics are used. These methods are applied to the data of Pune city, India. Five types of neighborhoods were identified, all clearly dissimilar from each other. Such understanding of the neighborhood structure from data is considered well‐ suited, in non-motorized transportation (NMT) planning studies. Also, neighborhood scale is a fairly neglected level in transportation planning, and this research suggests a method to extract rational neighborhoods for such simulations. The study closes, demonstrating a promising and consistently applicable procedure to provide an urban neighborhood structure from remote sensing images. Validation is done using travel data from a trip survey. The study calls for further research to develop transport models from the neighborhood scale.