Urban Currents

Mechanisms and spatial spillover effects of artificial intelligence industrial agglomeration on regional innovation level

This study analyzed panel data from 287 prefecture-level cities in China from 2011 to 2023 to evaluate how artificial intelligence industrial agglomeration (AIIA) affects regional innovation. The researchers mapped a spatial pattern of coastal concentration diffusing toward inland nodes, finding that AIIA significantly boosts regional innovation, particularly in eastern regions, cities outside major urban agglomerations, and non-resource-based cities. The positive impact is strongest in the upstream foundational and downstream application layers of the AI industrial chain, operating through human capital and digital economy development channels.

Why it matters — It establishes empirical evidence that AI industrial clustering has a positive spatial spillover effect and a critical-scale threshold, demonstrating that local AI concentration actively enhances the innovation capacity of neighboring regions.

Caveat: The findings are based entirely on the institutional and economic context of China, which may limit their direct applicability to countries with different industrial planning and regional development structures.

Applied Geography · doi

Geographies in flux - how time and location influence shrinking medium-sized cities

This study analyzes urban shrinkage in medium-sized cities (populations 20,000 to 100,000) across Germany and Poland over a 15-year period (2003–2018 for Germany; 2006–2021 for Poland). Using a harmonized shrinkage score built from six demographic and socioeconomic indicators, the researchers applied Moran's I, k-means clustering, and multivariate regression to classify these cities into six distinct typologies. The analysis reveals that proximity to large or metropolitan cities correlates with lower shrinkage, though regional structures and governance contexts explain a larger share of the variation.

Why it matters — It demonstrates that urban shrinkage is not a uniform process, showing that Polish medium-sized cities face intensifying decline concentrated in the south and east, whereas German patterns are more stable and geographically selective. This provides planners with a comparative, multi-national framework that proves regional and institutional contexts are more decisive for a medium-sized city's trajectory than simple geographic proximity to major metropolitan areas.

Cities · doi

Spatial exclusion on public land: Fragmented governance and Vancouver's golf courses

This study analyzes spatial exclusion on public land in Vancouver, British Columbia, focusing on municipal golf courses which represent only about 1% of park sites but occupy 13% to 19% of the city's total park area. Using a mixed-methods design, the researcher combined tract-level spatial inequality models with qualitative governance analysis, including interviews with former mayors, councillors, and Park Board commissioners alongside planning documents, focusing on the Langara Golf Course as a primary case study.

Why it matters — It identifies three specific governance mechanisms—fragmented authority, legacy zoning, and arm's-length administrative structures—that actively preserve exclusive, high-barrier recreational uses on scarce public land during an intense housing crisis.

Caveat: The empirical findings and governance pathways are derived from a single-city case study of Vancouver, which may limit direct generalizability to cities with different municipal administrative structures.

Cities · doi

Consumer reactions to AI enabled surveillance of drivers. the case of Pay-how-you-drive insurance products

This study analyzes consumer reactions to the transition from reward-only to sanction-based algorithmic driver monitoring. Researchers conducted 19 semi-structured interviews with German drivers aged 30 to 50 who use HUK-COBURG's Telematik Plus product, evaluating their responses to a hypothetical sanction-based system extension across six dimensions of concern.

Why it matters — The study identifies a structural 'perception-implementation gap' where users believe they are being socio-demographically discriminated against by the algorithm, even when the system is mathematically designed to exclude those specific variables. This suggests that technical fairness in algorithmic governance does not translate to perceived fairness by the public.

Caveat: The findings are based on a small, qualitative sample of 19 drivers within a single age bracket (30 to 50) reacting to a hypothetical scenario rather than an active, deployed sanctioning system.

Travel Behaviour and Society · pdf · doi

The impact of fuel prices and supply availability on user behavioural change in Australian household travel amidst global turmoil

This study surveyed 1,000 residents of New South Wales, Australia, in April 2026 to analyze how passenger car users adapted to a 30-50% spike in retail fuel prices and supply disruptions caused by Middle East conflict. The survey evaluated short- and long-term behavioral responses—including modal switching, vehicle kilometers traveled, car sharing, telecommuting, and electric vehicle adoption—across several scenarios of varying fuel prices and rationing levels.

Why it matters — It demonstrates that while fuel pricing serves as an effective short-term signal and rationing tool, long-term household resilience to energy shocks requires structural adaptations like vehicle technology transitions and access to travel alternatives.

Caveat: The behavioral responses are based on self-reported survey intentions under hypothetical scenarios rather than observed, long-term empirical travel data.

Transportation Research Part A Policy and Practice · doi

Autonomous vehicle deployment decisions under uncertainty: The roles of parking fees, sharing revenues, and attitudes

This study models how private autonomous vehicle owners choose to deploy their idle vehicles among parking, cruising, or sharing under uncertain fees and revenues. Using stated preference survey data from Japan, the researchers estimated a hybrid choice model based on cumulative prospect theory to capture how latent attitudes, such as trust in safety and privacy concerns, influence these decisions.

Why it matters — The model quantifies how non-monetary factors shape AV deployment, showing that trust in safety increases sharing while privacy concerns suppress it, and positive technology attitudes encourage traffic-generating cruising behavior. This provides a framework for planners to anticipate how subjective user attitudes, rather than just pricing, will impact urban traffic congestion and parking demand.

Caveat: The findings are based on stated preference survey data rather than observed real-world behavior, which may differ as actual autonomous vehicle technology and sharing markets mature.

Transport Policy · doi

Waterlogging tolerance of Rhododendron × pulchrum ‘Oomurasaki’ as a candidate shrub for roadside stormwater management in Japan

This study evaluated the waterlogging tolerance of Rhododendron × pulchrum 'Oomurasaki', Japan's most common roadside shrub, by testing its morphological and physiological responses to four flooding durations (control, 1-day, 3-day, and 5-day) in a greenhouse. The 5-day flooding regime killed the plants, while 3-day flooding significantly reduced biomass, canopy spread, visual quality, and stomatal conductance, whereas 1-day flooding caused only minor, non-lethal effects.

Why it matters — It establishes the precise hydrological limits for a ubiquitous urban shrub, showing it can be safely integrated into roadside rain gardens only if the design guarantees complete drainage within 24 hours.

Caveat: The findings are based on controlled greenhouse conditions rather than a real-world roadside deployment where plants face additional urban stressors.

Urban forestry & urban greening · pdf · doi

Space for people, not cars: a large-scale visual choice experiment on urban street redesigns

A stated choice experiment with 5,991 participants from a representative Swiss mobility panel evaluated fictional urban street redesigns using either static images or navigable 360-degree visualizations. The experiment varied street infrastructure, social spaces, and greenery to measure their impact on public acceptance, willingness to live nearby, and willingness to cycle. The 360-degree format made social features more salient to respondents, while static images heightened sensitivity to environmental and infrastructural elements.

Why it matters — It demonstrates that the choice of visualization medium directly biases how the public evaluates street redesigns, while also showing that adding greenery is a universally effective strategy to boost public acceptance across diverse demographic and political groups.

Caveat: The study relies on stated preferences and fictional street designs rather than real-world behavioral responses to actual street reallocations.

npj Urban Sustainability · pdf · doi

Computer-Assisted Web-Based Personal Interview Survey for Capturing Urban Mobility: A Contextual Approach in Resource-Constrained Settings

The study piloted a computer-assisted web-based personal interview (CAWPI) method in Chattogram, Bangladesh, combining face-to-face interviews with a low-cost web platform using enumerators' own devices. The survey collected demographic, vehicle, and travel data from 550 households across 33 wards, capturing 3,884 trips from 2,104 individuals. The platform integrated OpenStreetMap for geolocation, was optimized for low-bandwidth conditions, and achieved a 96% completion rate.

Why it matters — It demonstrates a viable, low-cost methodology for gathering high-quality, geolocated urban mobility data in resource-constrained Global South cities where traditional agency-provided devices or pure web surveys are financially or logistically infeasible.

Caveat: The system faced operational limitations during deployment, including restricted geospatial search capabilities, occasional connectivity drops, and manual input errors by enumerators.

Journal of Transportation Engineering Part A Systems · doi

Discovering Extreme Commuting through the Lens of the Built Environment: An Interstation Analysis

This study analyzed extreme rail-transit commutes lasting 60 minutes or more during morning and evening peak hours in Shanghai, China, using large-scale smart card data from September 2021. By combining Poisson regression and gradient boosting decision trees, the researchers modeled origin-destination pairs to identify nonlinear associations and threshold effects of built environment variables on extreme commuting.

Why it matters — The findings demonstrate a symmetrical but distinct pattern between morning origin-side variables and evening destination-side variables, proving that extreme rail commuting is jointly driven by the spatial mismatch of living costs, workplace accessibility, and local neighborhood conditions.

Caveat: The analysis is limited to rail-transit commuters within a single Chinese megacity, which may not represent extreme commuting patterns for other transport modes or smaller urban contexts.

Journal of Transportation Engineering Part A Systems · doi

Spatial similarity in socioeconomic data: a wavelet approach for England

The study introduces a method to compare geographic areas based on their internal spatial patterns by embedding areal data into a regular grid and applying a discrete wavelet transform to extract scale- and direction-specific coefficients. This approach was applied to England's 2025 Index of Multiple Deprivation at the Lower Layer Super Output Area (LSOA) level to group contiguous units with similar spatial structures.

Why it matters — It demonstrates that areas sharing highly similar internal spatial structures are distributed across different regions, levels of urbanization, and average deprivation levels, showing that standard geographic classifications are poor proxies for local spatial patterns.

arXiv · pdf

A Generative Framework for the Creation of Multi-Attribute Geographically-Explicit Synthetic Population

A hierarchical diffusion-based generative framework was developed to synthesize populations with realistic, region-specific joint distributions of five attributes (age, gender, employment, education, and income) alongside explicit home and work locations. The model was applied across 50 U.S. states and Washington, D.C., generating a nationwide synthetic population of 332,387,543 individuals. Evaluation against Iterative Proportional Fitting (IPF) and a one-shot diffusion baseline demonstrated superior reconstruction of joint distributions while preserving realistic residential and workplace spatial patterns.

Why it matters — It provides a scalable method to generate highly detailed, geographically explicit synthetic populations from aggregated data, allowing agent-based models and micro-simulations to incorporate more realistic human behaviors and interactions at both regional and national scales.

arXiv · pdf