Urban Currents

Understanding Long-Term Dynamics of Individual Metro Usage: A Hidden Semi-Markov State Framework with Survival Analysis

The study developed a state-based lifecycle modeling framework combining Hidden Semi-Markov Models with discrete-time survival analysis to track how individual transit usage changes over time. Using four years of smart card data from the Shanghai metro system (2021-2024), the model identified five distinct mobility states and mapped how passengers transition between them, as well as their likelihood of stopping or resuming metro use.

Why it matters — It reveals that passenger disengagement and return are governed by entirely different temporal rules: the risk of a user stopping transit use depends on their current mobility state rather than how long they have been in it, whereas their likelihood of returning decays sharply the longer they remain inactive. This allows transit operators to predict when users are at risk of churning and time their retention interventions more effectively.

Caveat: The empirical findings and identified mobility states are based on a single metro system in Shanghai and may reflect local transit dynamics rather than universal passenger behaviors.

arXiv · pdf

From optimization to prediction: Transformer-based path-flow estimation to the traffic assignment problem

The study introduces a Transformer-based deep learning model to predict equilibrium path flows directly, bypassing traditional mathematical optimization. Evaluated on four networks including Sioux Falls, Anaheim, Eastern Massachusetts, and a synthetic Manhattan-like grid, the model estimates path-level traffic flows in multi-class networks and adapts to changing demand and network topologies without recalculation.

Why it matters — It shifts traffic assignment from computationally expensive iterative optimization to near-instantaneous prediction, enabling rapid 'what-if' scenario testing and real-time traffic management on large-scale urban networks.

Caveat: The model's performance and speedups were demonstrated on standard synthetic and regional benchmark networks rather than a deployed, real-time metropolitan traffic system.

Transportation Research Part C Emerging Technologies · doi

Correction to ‘ EMBODYING AND RESISTING URBAN HEAT INJUSTICE: Migrant Vulnerabilities and Radical Adaptations in El Raval, Barcelona’

This document is a formal correction to a previously published paper on urban heat injustice in Barcelona, specifically updating the acknowledgments section to include a missing Horizon Europe funding grant.

Why it matters — It ensures proper funding attribution and administrative accuracy for the research project, though it contains no new empirical findings or methodological changes.

Caveat: This is a correction notice and does not contain any scientific data or research findings.

International Journal of Urban and Regional Research · doi

Demand-agnostic assessment of on-demand pooled transit services

The study develops a simulation-based method to evaluate pooled on-demand transit feeder services in areas lacking historical demand data, using resident address points to generate synthetic demand across varying adoption rates. The framework evaluates service viability using three key performance indicators to identify the optimal combination of service area and transit hub. The method was demonstrated by ranking 12 pre-selected feeder zones in Krakow, Poland.

Why it matters — It provides municipal planners with a structured pre-deployment tool to identify viable transit zones and hubs before investing in physical infrastructure or service contracts, removing the requirement for pre-existing travel demand datasets.

Caveat: The evaluation relies on synthetic demand generated from static residential address points, which may not fully capture actual dynamic travel behaviors or non-residential trip generators.

European Transport Research Review · pdf · doi

Modeling driver behaviour and the interplay of demographic and environmental factors in traffic violations: A bayesian network approach

This study developed a Bayesian Network model to analyze the combined effects of driver demographics, behaviors, and environmental conditions on crash severity. The model was trained and tested using a two-year dataset of traffic violations and accidents from Tempe, Arizona, identifying speeding, impaired driving, and low-visibility conditions as critical risk factors.

Why it matters — It demonstrates how Bayesian Networks can map the complex, non-linear interactions between driver age, specific traffic violations, and environmental hazards, allowing planners to run predictive scenarios and design age-specific road safety interventions.

Caveat: The model's predictive relationships and risk patterns are derived from a single mid-sized American city, which may limit its direct applicability to different urban layouts or international driving cultures.

Transportation Research Interdisciplinary Perspectives · doi

Urban congestion relief experiments through routing-app interventions

The study conducted large-scale empirical experiments across approximately 100 highly congested road segments in 10 major US cities, rerouting a small share of Google Maps trips to alternative routes of equivalent road classes. This marginal routing intervention achieved a city-average increase in vehicle speeds of 2% on the targeted segments, improved travel times by 0.7%, and showed potential to reduce annual CO2-equivalent emissions by over 1,000 tons per city in most locations.

Why it matters — It provides real-world empirical evidence that minor, app-based routing adjustments affecting only a small fraction of drivers can measurably improve overall urban road network efficiency without requiring infrastructure changes.

Nature Cities · pdf · doi

Design Before Disaster: Japan’s Culture of Preparedness

This paper analyzes Japan's historical and contemporary culture of disaster preparedness, focusing on how design, architecture, and urban planning are integrated to mitigate risks from earthquakes, typhoons, and flooding.

Why it matters — It documents a comprehensive national model where disaster resilience is embedded directly into daily spatial design and cultural practices, rather than treated solely as an emergency response measure.

Caveat: The abstract provides a high-level historical and cultural overview rather than empirical, quantitative measurements of specific mitigation outcomes.

Journal of the American Planning Association · doi

African urban planning histories

This study traces the historical evolution of urban planning across African cities, analyzing how colonial legacies, post-independence policies, and rapid demographic shifts shape contemporary metropolitan structures. It synthesizes planning trajectories to contextualize the projected growth of the continent's urban population to 1.4 billion by the mid-twenty-first century.

Why it matters — It provides a historical framework that explains why current planning paradigms often fail to accommodate rapid demographic growth, shifting the focus from treating African urbanization as an anomaly to understanding it through its specific historical precedents.

Caveat: The analysis relies on historical synthesis and demographic projections rather than new empirical datasets or localized spatial modeling.

Planning Perspectives · pdf · doi

Contesting verticality through consumption practices: digital markets as neighborliness in high-rise Santiago

Based on a year-long ethnographic study of high-rise buildings in Santiago, Chile, this research examines how residents use messaging applications to establish localized digital economies. These informal, neighbor-to-neighbor commercial transactions foster social interactions and community relations within highly dense, vertical residential environments.

Why it matters — It demonstrates that digitalized consumption practices can actively counter the social isolation and impersonal nature typically associated with high-rise living, showing how rigid architectural structures are socially reshaped through everyday digital neighborliness.

Caveat: The findings are based on qualitative ethnographic fieldwork within a specific South American city, which may limit generalizability to vertical housing contexts with different socioeconomic or regulatory conditions.

Urban Geography · doi

Measuring causal strengths from spatial cross-sectional data with geographical cross mapping cardinality

The Geographical Cross Mapping Cardinality (GCMC) model was developed to measure causal strengths from spatial cross-sectional data by calculating the intersectional cardinality of neighborhoods in reconstructed state space. The model integrates the DeLong placement method to assess the statistical significance of its estimates and was validated using a simulated three-variable causal benchmark alongside three representative spatial datasets with established causal structures.

Why it matters — It provides a more reliable method for spatial causal inference by successfully capturing weak, moderate, and strong coupling regimes while maintaining a low false-positive rate and remaining robust against observational noise.

International Journal of Geographical Information Systems · pdf · doi

Correction

This correction notice refers to a case study on promoting healthy lifestyles through the participatory design of a residential park in Riyadh, Saudi Arabia.

Why it matters — The document serves as a formal correction to the published case study, ensuring the accuracy of the academic record for this urban park design research.

Caveat: The provided text is a brief correction notice and does not contain the empirical findings, methodology, or data of the original study.

Cities & Health · pdf · doi

The aesthetic politics of memory anchoring in urban redevelopment at Shibati, Chongqing, China

This tracking study analyzes the entire urban redevelopment process of Shibati, a historic neighborhood in Chongqing, China. It traces how socially and culturally constructed memories and counter-memories are converted into calculable, tradable economic capital to optimize market positioning.

Why it matters — The study conceptualizes how memory acts as an aesthetic-political pathway, demonstrating that the politicization of heritage is often neutralized and co-opted into economic value during state-led urban governance.

Caveat: The findings are based on a single, qualitative tracking study of one neighborhood redevelopment project in Chongqing.

Urban Studies · doi

Missing maps and elastic eligibility: Archival securitization and retrospective arbitrage in Saigon and Mumbai

This study analyzes bureaucratic practices of territorial securitization in Ho Chi Minh City and Mumbai, focusing on the disappearance of a ratified planning map in the Thፐ Thim New Urban Area and the manipulation of eligibility lists in Mumbai's Slum Rehabilitation Scheme. It conceptualizes these actions as deliberate archival revisions that generate elastic, unstable categories of land and population rather than fixed property boundaries.

Why it matters — It demonstrates that the destabilization of official archives and property records is not a symptom of administrative failure or corruption, but rather a systematic, patterned technique of urban governance used to secure authority through the overproduction of bureaucratic instruments.

Caveat: The findings are based on qualitative analysis of two specific case studies in Ho Chi Minh City and Mumbai, which may limit direct generalizability to cities with different administrative and legal structures.

Environment and Planning D Society and Space · doi

Special issue ‘Geographies of science fiction’: introduction

This introductory paper establishes a framework for analyzing science fiction through a geographical lens, focusing on how the genre uses fictional displacement and deterritorialization to represent spatial relationships. It introduces a special issue dedicated to exploring the unique spatial affordances of science fiction texts in the context of contemporary planetary crises.

Why it matters — It bridges a gap between literary studies and geography, offering researchers a theoretical basis to use speculative fiction as a tool for analyzing and rethinking human relationships to threatened environments.

Caveat: The text is an introductory essay to a special issue and does not present empirical data or a specific localized case study.

Cultural Geographies · doi

Identifying and characterising urban land use transformation processes using cadastral parcel geometries: A case study of Spanish cities

The study presents a framework that tracks geometric and land use changes in cadastral parcels across 48 Spanish cities between 2022 and 2024. Using spatial metrics, intersection operations, and graph construction, the method categorizes parcel transformations across six land use classes into three primary processes: aggregating, subdividing, and non-impactful. The analysis reveals that subdividing typically converts large, complex undeveloped parcels into simpler, elongated shapes, whereas aggregating merges average-sized, simple parcels into more compact and complex forms.

Why it matters — It establishes a method to measure the physical restructuring of urban land at the parcel level, moving beyond simple land-use classification to quantify how the actual geometry and boundaries of urban properties evolve over time.

Caveat: The analysis is limited to a short two-year observation window between 2022 and 2024, which may capture immediate development actions but not long-term structural trends.

Environment and Planning B Urban Analytics and City Science · doi

Mastering the Twin–Game: Hierarchical Reinforcement Learning in a Digital Twin Sandbox for Adaptive Urban Healthcare Optimization—A Case Study of Wuhan

The researchers developed Twin-Game, a hierarchical reinforcement learning framework that optimizes urban healthcare resources using a digital twin sandbox of Wuhan, China. The system combines a macro-level Deep Q-Network for spatial attention allocation with a micro-level Proximal Policy Optimization model for continuous controls like bed expansion and staff scheduling. The simulation integrates a dynamic gravity model and a two-way referral mechanism, parameterizing Wuhan's facilities, population, and transport networks alongside synthetic patient flows.

Why it matters — It provides a method to optimize complex, city-scale healthcare systems without the dimensionality explosion and unstable convergence that typically plague single-layer reinforcement learning models in massive action spaces.

Caveat: The framework's patient arrivals and disease profiles are synthetically generated due to the unavailability of public individual-level clinical flow data, meaning the findings serve as simulation-based decision support rather than clinically validated outcomes.

ISPRS International Journal of Geo-Information · doi

Public transit gains and spatially uneven travel demand changes after NYC congestion pricing

Using time series foundation models to generate probabilistic counterfactual forecasts, this study evaluated the mobility impacts of New York City's January 2025 congestion pricing program. The analysis of bus, subway, and aggregate trip volume data revealed a significant increase in post-policy transit ridership alongside a modest decrease in overall travel demand. These effects varied spatially, with overall demand reductions concentrated inside the Congestion Relief Zone, while public transit gains extended well beyond Manhattan's core.

Why it matters — It demonstrates a scalable, uncertainty-aware evaluation framework for large-scale urban interventions where traditional, clean control groups cannot be constructed due to system-wide spillovers.

arXiv · pdf

Model-based clustering of compositional trajectories for the analysis of mobility data

The study introduces a state-space statistical framework to model and cluster individual mobility trajectories derived from mobile phone data. By representing uncertain device locations as compositional data in a simplex, the model integrates spatial measurement error with the proportions of surrounding road types. The method was applied to aggregate individual movements into distinct population-level mobility patterns using a dataset from the municipality of Padova, Italy, and its surrounding areas.

Why it matters — It provides a way to analyze noisy mobile phone tracking data by explicitly accounting for spatial uncertainty and matching raw location signals to the underlying road network structure simultaneously.

Caveat: The framework's empirical utility is demonstrated only on a single geographic case study in and around Padova.

arXiv · pdf

Understanding the "Airport" Censorship Circumvention Ecosystem in China

A systematic study of China's subscription-based censorship circumvention proxies, known as 'airports', using surveys from 1,667 respondents, social media analysis, and active network measurements. The researchers identified 3,431 active airports built on open-source toolkits, subscribed to 35 of them to evaluate performance, and analyzed their multi-hop architecture, payment methods, and vulnerability to government takedowns.

Why it matters — It documents and measures the dominant, consumer-facing circumvention ecosystem in China, revealing that these commercial tools outperform academic alternatives but introduce new risks, including centralized payment tracking and airport-level censorship policies.

Caveat: The survey sample of 1,667 respondents may not be fully representative of all internet users in China due to the self-reporting nature of circumvention tool usage.

arXiv · pdf