Evaluating the Relation Between Station Area Design Parameters and Transit Usage for Urban Rail Systems in Ankara, Turkey

Ozgur-Cevher O., Altintasi O., TÜYDEŞ YAMAN H.

INTERNATIONAL JOURNAL OF CIVIL ENGINEERING, vol.18, no.8, pp.951-966, 2020 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 18 Issue: 8
  • Publication Date: 2020
  • Doi Number: 10.1007/s40999-020-00506-7
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Applied Science & Technology Source
  • Page Numbers: pp.951-966
  • Keywords: Station area design, Rail transit ridership, Urban rail transit planning, Regression models, Statistical analysis, LAND-USE, ORIENTED DEVELOPMENT, RIDERSHIP, TRANSPORT, PERFORMANCE, IMPACTS, DENSITY, MODELS, LEVEL
  • Middle East Technical University Affiliated: Yes


Urban rail transit (URT) systems are particularly favoured by planners due to their potential in attracting car users. However, a URT investment must be complemented by land-use characteristics at macroscale (whole city) and design parameters at microscale (vicinity of the station). Despite the common referencing to key concepts (i.e. density, diversity and connectivity), the variability in their definitions and scales causes ambiguity in the determination of their quantified impact on URT ridership. Furthermore, their impact may be different in developing countries, where more mesoscale (corridor-based) effects are expected in the early stages of the URT network. This study aimed to evaluate mathematically the impact of eight selected station design and public transit supply variables on ridership, based on data collected at 14 stations of two existing rail system corridors, both metro (M1) and light rail transit (A1) lines in Ankara, Turkey. Principal component analysis (PCA) indicated three major groups of parameters: (1) land use, (2) public transit supply and (3) connectivity around stations. In single-variable regression models, density and diversity were not found to be statistically significant factors despite the commonly accepted conceptual relation to ridership. A multiple regression model with bus frequency and density (R-2 = 0.902) explained the ridership in Ankara URT systems more significantly at its early development stage. There was a significant difference between developed factor relations for both M1 and A1 lines, suggesting that mesolevel impacts should be considered in the evaluation of URT systems.