Spatial Distribution of City Tweets and Their Densities

Bin Jiang, Ding Ma, Junjun Yin, Mats Sandberg

Research output: Contribution to journalArticlepeer-review

34 Scopus citations


Social media outlets such as Twitter constitute valuable data sources for understanding human activities in the virtual world from a geographic perspective. This article examines spatial distribution of tweets and densities within cities. The cities refer to natural cities that are automatically aggregated from a country's small street blocks, so called city blocks. We adopted street blocks (rather than census tracts) as the basic geographic units and topological center (rather than geometric center) to assess how tweets and densities vary from the center to the peripheral border. We found that, within a city from the center to the periphery, the tweets first increase and then decrease, while the densities decrease in general. These increases and decreases fluctuate dramatically, and differ significantly from those if census tracts are used as the basic geographic units. We also found that the decrease of densities from the center to the periphery is less significant, and even disappears, if an arbitrarily defined city border is adopted. These findings prove that natural cities and their topological centers are better than their counterparts (conventionally defined cities and city centers) for geographic research. Based on this study, we believe that tweet densities can be a good surrogate of population densities. If this belief is proved to be true, social media data could help solve the dispute surrounding exponential or power function of urban population density.

Original languageEnglish (US)
Pages (from-to)337-351
Number of pages15
JournalGeographical Analysis
Issue number3
StatePublished - Jul 1 2016

All Science Journal Classification (ASJC) codes

  • Geography, Planning and Development
  • Earth-Surface Processes


Dive into the research topics of 'Spatial Distribution of City Tweets and Their Densities'. Together they form a unique fingerprint.

Cite this