Extracting demographic and socio-economic characteristics of urban/suburban areas using LiDAR remote sensing

Demographic and socio-economic information provides important inputs for urban/suburban environment mapping and monitoring, such as urban planning, conservation planning, management of natural resources, and facility allocation. This dissertation investigated extracting various demographic and socio-economic variables using Light Detection and Ranging (LiDAR) remote sensing data as a major data source.;Two studies (Chapters 2 and 3) estimated demographic (i.e., population) information in the suburban and urban areas, respectively, of Denver, Colorado. The two studies employed an area-based approach and a volumetric approach to estimate population at census block level. Both approaches resulted in successful performance for estimating population with high accuracy in suburban areas. However, the area-based approach failed to yield accurate results in urban areas where the building components were heterogeneous. Unlike the inconsistent performance of the area-based model, the volumetric approach resulted in accurate estimations in both suburban and urban areas.;The study in Chapter 4 investigated classifying buildings delineated from the LiDAR data into three types: single-family houses, multiple-family houses, and non-residential buildings. Four groups of spatial attributes described the shape, location, and surrounding environment of buildings were calculated and subsequently employed in classification. Two study sites located in suburban and downtown areas of Denver, Colorado were selected due to quite different building components and neighborhood environments. Machine learning approaches based on the remote sensing-derived attributes produced reliable classification accuracy for identifying building type information.;Residential house value, one of the most important socio-economic characteristics of urban/suburban areas, was estimated in Chapter 5. An optimized regional regression approach integrating a differential evolution optimization algorithm with the Ordinary Least Square regression method was proposed in this study to improve the house value prediction accuracy. Results showed that remote sensing-derived features were capable of accurately estimating house transaction price. In particular, the volume of residential buildings appeared to be an effective surrogate variable for total living area, the most important variable in typical house price estimation models.