Population spatialization and synthesis with open data

TL;DR

This paper presents an automated method for population spatialization and synthesis using open data sources like OpenStreetMap, crowd-sourced POIs, and online check-in records, validated with ground truth data in Beijing.

Contribution

It introduces an integrated open-data-based approach for population spatialization and synthesis, addressing data scarcity in developing countries.

Findings

Effective parcel delineation using open street data

Accurate identification of residential parcels with online check-ins

Validated population synthesis results with ground truth data

Abstract

Individuals together with their locations & attributes are essential to feed micro-level applied urban models (for example, spatial micro-simulation and agent-based modeling) for policy evaluation. Existed studies on population spatialization and population synthesis are generally separated. In developing countries like China, population distribution in a fine scale, as the input for population synthesis, is not universally available. With the open-government initiatives in China and the emerging Web 2.0 techniques, more and more open data are becoming achievable. In this paper, we propose an automatic process using open data for population spatialization and synthesis. Specifically, the road network in OpenStreetMap is used to identify and delineate parcel geometries, while crowd-sourced POIs are gathered to infer urban parcels with a vector cellular automata model. Housing-related online Check-in records are then applied to distinguish residential parcels from all of the identified urban parcels. Finally the published census data, in which the sub-district level of attributes distribution and relationships are available, is used for synthesizing population attributes with a previously developed tool Agenter (Long and Shen, 2013). The results are validated with ground truth manually-prepared dataset by planners from Beijing Institute of City Planning.