Understanding dynamic interactions between human activities and land-use structure in a city is a key lens to explore the city as a complex system. This dissertation contributes to understanding the complexity of urban dynamics by gaining knowledge of the interactions between human activities and city land-use structures by utilizing free-accessible socially sensed data sources, and building upon recent research trend and technologies in geographical information science, urban study, and computer science. This dissertation addresses three main questions related to human dynamics: 1) how human activities in an urban environment are shaped by socioeconomic status and the intra-city land-use structure, and how in turn, the knowledge of socioeconomic status-activity relationships can contribute to understanding the social landscape of a city; 2) how different types of activities are located in space and time in three U.S. cities and how the spatiotemporal activity patterns in these cities characterize the activity profile of different neighborhoods in the cities; and 3) how recent socially sensed information on human activities can be integrated with widely-used remotely sensed geographical data to create a novel approach for discovering patterns of land use in cities that are otherwise lacking in up to date land use information. This dissertation models the associations between socioeconomics and mobility in the Washington, D.C. metropolitan area as a case study and applies the learned associations for inferring geographical patterns of socioeconomic status (SES) solely using the socially sensed data. This dissertation also implements a semi-automated workflow to retrieve activity details from socially sensed Twitter data in Washington, D.C., the City of Baltimore, and New York City. The dissertation integrates remotely-sensed imagery and socially sensed data to model the dynamics associated with changing land-use types in the Washington, D.C.-Baltimore metropolitan area over time.