Large-scale urban building function mapping by integrating multi-source web-based geospatial data

Authors: CHEN, WEI - Iowa State University; ZHOU, YUYU - Iowa State University; STOKES, ELEANOR - Universities Space Research Associaton; Zhang, Xuesong

Submitted to: Geo-spatial Information Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 9/25/2023
Publication Date: 10/31/2023
Citation: Chen, W., Zhou, Y., Stokes, E.C., Zhang, X. 2023. Large-scale urban building function mapping by integrating multi-source web-based geospatial data. Geo-spatial Information Science. 1-15. https://doi.org/10.1080/10095020.2023.2264342.
DOI: https://doi.org/10.1080/10095020.2023.2264342

Interpretive Summary: Buildings are a major energy consumer and greenhouse gases emitter. The lack of building functional types (e.g., working, living, and shopping) at large spatial scales presents a major challenge for urban planning and management. Here, we developed web- and map-crawlers to extract points of interest (POIs), roads, and land use parcels from Tripadvisor.com and Google Maps. Next, we identified residential and non-residential buildings and their functional types (e.g., hospital, hotel, school, shop, restaurant, and office) by leveraging a machine learning method and different building and land use parcel maps. The proposed method was tested in 50 U.S. cities and achieved high accuracy (94%). The method can be easily transferred to other cities across the globe to support studies that examine sustainability of rural-urban areas, such as the Chesapeake Bay Watershed where both agriculture and cities contribute to water quality concerns and greenhouse gas emissions.

Technical Abstract: Morphological (e.g., shape, size, and height) and function (e.g., working, living, and shopping) information of buildings is highly needed for urban planning and management as well as other applications such as city-scale building energy use modelling. Due to the limited availability of socio-economic geospatial data, it is more challenging to map building functions compared to building morphological information, especially over large areas. In this study, we proposed a novel framework to map building functions in 50 U.S. cities by integrating multi-source web-based geospatial data. First, web crawler and map crawler were developed to extract points of interest (POIs), roads, and land use parcels from Tripadvisor.com and Google Maps, respectively. Second, an unsupervised machine learning algorithm named OneClassSVM was used to identify residential buildings based on landscape features derived from Microsoft building footprints. Third, type ratio of POIs and area ratio of land use parcels were used to identify six non-residential functions (i.e., hospital, hotel, school, shop, restaurant, and office). The accuracy assessment indicates that the proposed framework performed well with an average overall accuracy of 94% and kappa coefficient of 0.63. With the worldwide coverage of Google Maps and Tripadvisor.com, the proposed framework is transferable to other cities over the world. The data products generated from this study are of great use for quantitative city-scale urban studies, such as building energy use modelling at the single building level.