2026-07-21

How built environment factors shape demand for demand-responsive transport services: spatiotemporal patterns from a German case study

This study analyzed the influence of built environment characteristics on demand-responsive transport (DRT) demand using 2024 trip data from southern Hamburg, Germany. Spatial regression models were applied to small-scale grid cells across five distinct time-of-week periods to capture temporal variations. The analysis revealed that rail stations, hospitals, and population density consistently drive demand, while schools, workplaces, and retail locations show highly time-dependent effects.

Why it matters — It demonstrates exactly when and where DRT acts as a complement to conventional transit rather than a competitor, showing that poor fixed-route transit access only boosts DRT demand during off-peak hours when regular service is weak.

Caveat: The findings are derived from a single DRT service in southern Hamburg, which may limit their direct applicability to cities with different transit structures or spatial layouts.

Case Studies on Transport Policy · doi

Pathways to adopting microtransit: Examining the role of older adults' usability perceptions and willingness to use in Mississippi

This study analyzed survey data from 193 respondents aged 55 and older in Mississippi to evaluate how built environment, socio-economic, and travel characteristics influence their willingness to use on-demand microtransit. Using a structural equation model, the research measured how factors like smartphone adaptation, car expenses, and current transit usage mediate usability perceptions. The model revealed that residents in rural and suburban areas experience fewer difficulties with their current transportation modes than urban peers, and that males reported higher public transit usage than females.

Why it matters — It identifies the specific behavioral and environmental mediators that dictate microtransit acceptance among older adults in a predominantly rural state, offering a framework to design targeted pilot programs and pricing strategies for aging populations.

Caveat: The findings are based on self-reported survey data from a relatively small sample of 193 older adults within a single US state, which may limit generalizability to other regions.

Journal of Transport & Health · doi

Quantifying the accuracy of OpenStreetMap for longitudinal cycling-infrastructure change: A reproducible validation framework using Google Street View

The study presents an open-source workflow to validate temporal changes in cycling infrastructure using OpenStreetMap (OSM) snapshots and historical Google Street View imagery, using Barcelona (2015-2023) as a case study. Geometric differencing of OSM networks revealed an 80.3% increase in cycleway length from 130.7 km to 235.7 km. Validation of these changes showed high recall (0.96) and precision (0.82) for infrastructure additions, but very low precision (0.20) for infrastructure removals.

Why it matters — It establishes that while OSM history is highly reliable for tracking network expansions, apparent infrastructure removals are frequently mapping artifacts rather than physical losses, providing a calibration framework that prevents researchers from misinterpreting database edits as real-world changes.

Caveat: The low precision and high uncertainty for infrastructure removals are based on a very small sample size of five detected cases.

Journal of Transport Geography · doi

Transnational hubs and domestic hinterlands: Busan and the limits of port consolidation policies in Japan

This study analyzes how global container shipping networks shape domestic port hinterlands in Japan by combining a longitudinal analysis of inland distance friction from 1985 to 2018 with cross-sectional spatial interaction models for 2018 using CCFS and AIS data. It finds that feeder connectivity to the foreign transhipment hub of Busan, South Korea, is associated with strong but spatially bounded hinterland capture, particularly for exports, whereas trunk-line connectivity shows weak and inconsistent effects.

Why it matters — It demonstrates that global maritime network integration reshapes rather than eliminates spatial frictions, showing that regional feeder networks to foreign hubs can be more influential for domestic cargo capture than direct trunk-line connections.

Caveat: The empirical analysis is restricted to the Japanese port system and its specific relationship with the Busan hub, which may limit generalizability to other national port-hinterland configurations.

Journal of Transport Geography · pdf · doi

Unveiling the effects of activity-travel chaining on peak and off-peak travel: workers vs. non-workers

This study analyzes how the propensity to chain activities affects motorized person miles traveled (PMT) during peak and off-peak hours, comparing workers and non-workers. Using a structural equation modeling framework, the researchers evaluate direct and indirect effects through four mediators: travel time savings ratio, activity space area, average daily trips, and non-car mode usage. The findings show that activity-travel chaining increases peak-hour motorized PMT for both groups across most activity types, though non-workers chain maintenance activities during off-peak hours to avoid peak-period travel.

Why it matters — It demonstrates that consolidating trips into chains does not automatically reduce peak-period road demand, challenging the common planning assumption that efficient chaining behavior naturally mitigates peak congestion.

Caveat: The abstract does not disclose the dataset, sample size, or geographic location used to estimate the structural equation models.

Travel Behaviour and Society · pdf · doi

Learning-Enabled Output–Feedback Design for Lane Change Control: A Comparative Study

This study evaluates an estimator-based output-feedback learning controller for autonomous lane changes using a small-scale remotely controlled car platform. The controller, which learns an optimal steering policy online without direct access to full system states, was experimentally compared against a model-based model predictive control (MPC) and a learning-based state feedback controller under noisy and limited sensor conditions.

Why it matters — It demonstrates that learning-enabled output-feedback control can maintain robust and smooth lane-changing performance under realistic sensor constraints, reducing the necessity for expensive, high-precision sensing hardware in autonomous vehicles.

Caveat: The control strategies were tested on a scaled, small-scale RC car platform rather than a full-sized vehicle in real-world traffic.

Journal of Transportation Engineering Part A Systems · doi

Adaptable Cities and Temporary Urbanism

This theoretical paper synthesizes the historical evolution of temporary urbanism and adaptable cities, tracing how planning debates have shifted from rigid, permanent structures toward flexible, transient spatial uses.

Why it matters — It provides a conceptual framework that helps planners reconcile the tension between long-term master planning and the immediate, shifting needs of urban populations.

Caveat: The paper is a theoretical synthesis and does not present empirical data, quantitative measurements, or case study evaluations.

Journal of the American Planning Association · doi

The European Union as a Global Cultural Power

This study evaluates the European Union's international cultural relations strategy a decade after the launch of its 2016 Joint Communication. It analyzes how the EU's cultural diplomacy framework has been implemented globally and assesses its effectiveness in projecting cultural power.

Why it matters — It provides an empirical assessment of a major multilateral cultural policy framework, showing how strategic policy intentions translate into actual diplomatic practice and global influence.

Caveat: The analysis is limited by the brief abstract, which does not disclose the specific methodology, sample size, or geographic datasets used to measure this cultural influence.

Cultural Trends · doi

Multiscale space-time varying coefficient modeling using generalized additive models: determining model form

This study introduces a method for multiscale space-time varying coefficient modeling by extending Generalized Additive Models (GAMs) into the temporal domain. The approach generates multiple competing GAMs with different space-time interaction specifications for each predictor, selecting the optimal model using the Akaike Information Criterion (AIC). When tested on simulated data and an empirical case study, the selected GAM outperformed Multiscale Geographically and Temporally Weighted Regression (MGTWR) in model fit.

Why it matters — It provides a systematic way to determine the exact nature and scale of space-time dependencies in predictor-to-response relationships, relaxing rigid assumptions about how spatial and temporal variations interact.

International Journal of Geographical Information Systems · pdf · doi

GKTLF: a geographical knowledge-guided transfer learning framework for cross-regional DEM super-resolution

The researchers developed the Geographical Knowledge-guided Transfer Learning Framework (GKTLF) to improve cross-regional digital elevation model (DEM) super-resolution. The framework incorporates terrain skeleton lines and slope features, and uses a terrain similarity metric to handle spatial heterogeneity across regions. Tested across four regions in China, the GKTLF-enhanced ESRGAN model improved accuracy by up to 35.8% under full-sample pre-training and up to 31.2% when using only 40% of pre-training samples compared to standard ESRGAN baselines.

Why it matters — It provides a method to generate high-resolution elevation models in target regions that have highly different terrain characteristics from the training regions, reducing the performance drop typically caused by spatial heterogeneity.

Caveat: The framework's performance gains varied widely across the four test regions, with accuracy improvements ranging from a high of 35.8% down to just 4.6% under full pre-training.

International Journal of Geographical Information Systems · doi

Factors influencing bike-sharing travel behaviour: a Beijing structural equation modelling case study

This study analyzed the factors influencing bike-sharing travel behavior in Beijing using structural equation modeling with four latent variables: behavioral attitude, subjective norm, perceived behavioral control, and shared bicycle satisfaction. The model was estimated and refined using AMOS software based on survey data collected from local residents, measuring satisfaction specifically through safety, convenience, and economy.

Why it matters — It establishes a structured, empirical relationship between user psychology and bike-sharing usage, demonstrating how subjective norms and perceived control interact with specific service attributes like safety and cost to drive travel choices.

Caveat: The findings are based on survey data from a single city, Beijing, which may limit their generalizability to cities with different urban forms, transit systems, or bike-sharing regulatory frameworks.

Urban Planning and Transport Research · pdf · doi

COSMOLOGICAL MARKINGS AND ERASURES: The Political Theology of State‐Led Gentrification and Heritagization

This study analyzes the reconstruction of the fortress around the Yogyakarta Kraton in Indonesia, a project tied to a UNESCO World Heritage designation. Using a multi-method approach, it examines how the state-led construction of heritage facilitates tourism and gentrification while legitimizing the displacement of local residents.

Why it matters — It demonstrates that state-led gentrification and heritagization can function as tools of political theology, where the physical reconstruction of urban space serves to glorify state power and reinforce the dual cultural and governmental authority of the ruling Sultanate.

Caveat: The empirical evidence is drawn from a single, highly specific political and cultural context in Yogyakarta, which may limit its direct applicability to heritage-led gentrification in non-monarchical systems.

International Journal of Urban and Regional Research · doi

GUIDED Network-Agnostic Feature Initialization for Spatial Transferability in GNN-based Models

This research introduces Geometrically Unconstrained Inductive Demand EmbeDding (GUIDED), a network-agnostic initialization layer that represents travel demand as scalar attributes on virtual links rather than node-specific features. Evaluated using a Heterogeneous Graph Attention Network (HetGAT) across multiple urban topologies, the framework maintains predictive accuracy for traffic assignment while reducing training time per epoch by approximately 50% through optimized scatter operations.

Why it matters — It overcomes the spatial generalization gap of standard Graph Neural Networks, allowing traffic assignment models to transfer to entirely new urban environments and out-of-distribution demand patterns without requiring artificial input homogenization.

arXiv · pdf