Dynamically Weighted Spatiotemporal Interpolation for Modeling Distribution of Social Media Population

Authors: Dandong Yin*, University of Illinois, Shaowen Wang, University of Illinois
Topics: Geographic Information Science and Systems, Spatial Analysis & Modeling, Cyberinfrastructure
Keywords: GIS, CyberGIS, Agent-based modeling, Emergency Evacuation
Session Type: Paper
Day: 4/6/2019
Start / End Time: 8:00 AM / 9:40 AM
Room: Roosevelt 1, Marriott, Exhibition Level
Presentation File: No File Uploaded


Population distribution modeling is the core of various geospatial applications, and also a critical factor for emergency evacuation. However, the traditional census-survey-based population models struggle to get dynamic population information with desirable spatial resolution. The emerging location-based social media (LBSM) data provides individual trajectory information with enough spatiotemporal resolution, but also posing questions, such as data accuracy, discontinuity, representativeness, and massive volume. In this research, we investigated the major challenges in modeling population distribution from LSBM data, proposed methodologies to mitigate some of these major challenges, and developed a CyberGIS-Jupyter-enabled workflow to obtain an LBSM-based population model with plausible spatiotemporal resolution. By orchestrating a combination of methodologies including space-time-prism and temporal activity curve, the proposed workflow is able to alleviate the data sparsity and of uncertainty in LBSM data to a certain degree. A case study is conducted with 5.1 million geotagged tweets in the city of Miami. The output population distributions for each hour of the day at 30m resolution for the city of Miami. Diurnal dynamics in the achieved results correspond well with local landmarks in Miami city, which illustrates the effectiveness of the model as well as the unique value of LBSM data.

Abstract Information

This abstract is already part of a session. View the session here.

To access contact information login