Urban Currents

Methane Hazard in a Post-Mining Urban Environment: A Probabilistic Risk Assessment for Lupeni, Romania

This study analyzed methane hazards in Lupeni, Romania, following unexpected gas emissions during a pipeline installation in August 2024. Using a database of 123 observations across 41 fixed monitoring points over three field campaigns, the researchers found methane concentrations averaging 5.65% vol., with a maximum of 54% vol. They developed a probabilistic screening model using a Beta posterior and 50,000 Monte Carlo simulations to map risk severity and recurrence, revealing a localized but severe hotspot-envelope risk score of 19.40.

Why it matters — It provides a reproducible, data-driven screening workflow for post-mining cities to identify localized gas hazards and target interventions without requiring complex, expensive subsurface geological characterization.

Caveat: The risk assessment relies on a small monitoring dataset of three field campaigns over a short time period, which may not capture long-term seasonal variations in gas migration.

Urban Science · doi

Neoliberal urban governance and local empowerment: Unpacking the paradoxes of the BIP/ZIP programme in Lisbon

This study evaluated 13 editions of Lisbon's BIP/ZIP program from 2011 to 2023, analyzing data on stakeholders, total investments, benefited areas, and project themes. The analysis reveals a mismatch between the socio-spatial problems identified in designated priority areas and the actual projects approved for funding, with investments heavily concentrated in central neighborhoods and a small group of recurring partner associations.

Why it matters — It demonstrates how local empowerment and active citizenship programs under neoliberal governance can inadvertently lead to the capture of social capital by a concentrated group of actors, rather than distributing resources to the most deprived areas.

Caveat: The analysis relies on program-level administrative data regarding investments and partners, rather than direct field evaluations of the projects' local social impacts.

Cities · doi

A causal network for strategic foresight of inclusive climate actions through co-creation of nature-based solutions

This study develops a digital decision-support tool that models the relationships between Inclusive Climate Actions, such as Nature-based Solutions, and pre-defined Key Performance Indicators. The tool uses a causal network to map synergies and trade-offs among these indicators, estimating relative changes both numerically and spatially based on user-selected actions.

Why it matters — It provides a real-time, automated estimation of how climate interventions impact broader social and economic urban systems, enabling collaborative scenario-building during participatory workshops with citizens and public administrations.

Sustainable Cities and Society · doi

Less parking, more life? Promises and challenges of urban parking-space repurposing. Evidence from a controlled before-after quasi-experimental study in Berlin

This study evaluated the conversion of approximately 350 on-street parking spaces (13% of the local total) into green spaces, outdoor classrooms, shared-mobility stations, and delivery zones in a Berlin neighborhood in 2023. Using a controlled before-after quasi-experimental design with 2,027 baseline and 1,770 follow-up survey respondents, the analysis compared treatment areas, comparable control areas, and adjacent neighborhoods. The intervention did not significantly reduce private car use, but it improved perceived road safety and outdoor physical activity while slightly increasing parking search times and visitor noise.

Why it matters — It provides rare quasi-experimental evidence on the immediate trade-offs of neighborhood-level parking removal, demonstrating that while public health and safety perceptions improve, car ownership habits persist in the short term and local policy acceptance declines among the most directly affected residents.

Caveat: The post-intervention evaluation was conducted only six months after the parking spaces were repurposed, which may not capture long-term behavioral adaptations in car ownership or travel mode choices.

Transportation Research Part A Policy and Practice · doi

The impact of delayed urban rail projects on land prices in Hanoi, Vietnam

This study analyzed the effects of construction completion and subsequent operational delays of Hanoi's first urban rail line on local land prices. Using a spatial difference-in-differences model combined with machine learning, the analysis found that completed construction initially boosted land prices by 3.76% in treatment zones, with the highest increases occurring 150 to 300 meters from construction sites. However, during the prolonged delay in starting operations, land prices within 750 meters of the stations fell by 3.45%.

Why it matters — The research measures the financial penalty of infrastructure delays, demonstrating that prolonged uncertainty can actively erode the land value premiums generated during the construction phase.

Caveat: The findings are based on a single transit line in Hanoi, which may limit generalizability to cities with different institutional contexts or real estate dynamics.

Transport Policy · doi

A decade of change: Urban green space users’ shifting perceptions and practices towards biodiversity

This study analyzed changes in how urban green space users interact with and perceive biodiversity by comparing survey data from 1,217 visitors across five socially diverse parks in Paris between 2010 and 2023. The analysis reveals that park users increasingly link their personal well-being to the perceived richness of local biodiversity, show a higher demand for ecologically oriented green spaces, and express a growing preference for community-led conservation over public policy initiatives.

Why it matters — It documents a decade-long shift in public expectations, demonstrating that urban residents' definition of green space quality has evolved from simple recreational access to active ecological richness, while their trust in municipal environmental management has simultaneously eroded.

Caveat: The findings are based on self-reported survey data from a single European capital, which may limit generalizability to cities with different socio-political contexts or green space planning histories.

Urban forestry & urban greening · doi

Moving well, ageing well: How built and social environments shape mobility and well-being in later life

This study analyzed survey data from 1,115 older adults aged 65 to 93 in Ghent, Belgium, using structural equation modeling to trace how built and social environments influence well-being. The model evaluated both direct effects and indirect pathways mediated by travel modes including walking, cycling, public transit, and driving. The findings reveal that frequent travel of any mode supports well-being, with the built environment exerting a stronger and more consistent influence than the social environment.

Why it matters — It demonstrates that maintaining general trip-making ability is itself a key driver of well-being for older adults, shifting the focus from promoting specific active modes to supporting overall mobility through a combination of walkable infrastructure, transit quality, and car accessibility for those with limited mobility.

Caveat: The findings are based on self-reported survey data from a single European city, which may limit generalizability to regions with different urban densities or transport cultures.

Journal of Urban Mobility · doi

Artificial intelligence to environmental sustainability in urban mobility: A scoping review and a conceptual framework

This scoping review synthesizes 71 journal articles published between 2015 and 2025 to map artificial intelligence applications in urban mobility and their environmental impacts. The analysis uses thematic coding and causal loop diagrams to categorize machine learning and deep learning techniques across transport modes, focusing on demand prediction, traffic optimization, and emissions monitoring.

Why it matters — It demonstrates that while AI-driven traffic and fleet optimization improve energy efficiency, substantial emissions reductions are not achieved by technology alone and require a structural modal shift toward public transit and active transportation.

Caveat: The synthesized evidence is predominantly correlational and heavily concentrated in large cities within high-income nations, limiting its immediate applicability to other contexts.

Journal of Urban Mobility · doi

Linking species distributions and urban governance through social ecological fit in blue green infrastructure

The study evaluates the social-ecological fit of blue-green infrastructure (BGI) in Zürich, Switzerland, by combining habitat suitability and connectivity models for declining amphibians with social network analysis of multi-sector actor surveys. Using social-ecological networks (SENs), it maps areas of alignment and misalignment between physical BGI assets and the actors who govern them.

Why it matters — It demonstrates that while actor collaboration is stronger around amphibian biodiversity hotspots, critical connector habitats like natural springs and cemeteries are frequently neglected due to fragmented governance. This provides a framework to identify specific institutional blind spots in urban conservation, showing which sectors act as bridges and which remain peripheral despite their physical impact on ecosystems.

Caveat: The empirical findings and actor network dynamics are derived from a single mid-sized European city, which may limit direct generalizability to other urban governance contexts.

npj Urban Sustainability · pdf · doi

Urban heat risk assessment: a machine learning analysis of multidimensional morphology in 293 Chinese cities

This study developed a three-dimensional heat risk framework evaluating hazard, exposure, and vulnerability across 293 prefecture-level cities in China using machine learning and multi-source data. The analysis revealed that 64% of these cities face moderate to high heat risk, with higher risks concentrated in the northwest and lower risks in the southeast. Among the landscape metrics analyzed, average building volume, average height, and the building structure index emerged as key factors, accounting for 17%, 11%, and 10% of the association with heat risk respectively, through distinct nonlinear relationships.

Why it matters — It establishes a scalable, national-scale assessment framework that quantifies how specific, multi-dimensional urban morphology and building structures nonlinearly influence heat risk, moving beyond single-city studies to inform macro-level urban planning.

npj Urban Sustainability · pdf · doi

Spatial–Temporal Interaction–Aware Multiagent Reinforcement Learning for Traffic Signal Control

The researchers developed a multiagent reinforcement learning algorithm, STIA-MARL, for multi-intersection traffic signal control that incorporates historical traffic flows to capture dynamic spatiotemporal correlations. The model uses a custom graph structure to modify adjacency relationships, which improves how a shared policy network adapts to different intersections. Testing on synthetic and real-world datasets showed at least an 11.5% reduction in average travel time compared to existing state-of-the-art methods.

Why it matters — It demonstrates that incorporating historical flow data directly into the spatial modeling of traffic networks, rather than relying on static spatial relationships, significantly improves the efficiency of coordinated traffic signal control.

Journal of Transportation Engineering Part A Systems · doi

Prospective Teachers’ Beliefs About Inclusive Geography Teaching: Impact of an Intervention

This study evaluated a university-based teacher training course designed to improve inclusive geography teaching beliefs and self-efficacy. Using a pre-post experimental design with an untreated control group, the researchers analyzed changes in prospective teachers' beliefs using repeated measures ANOVA and MANCOVA.

Why it matters — The intervention demonstrates that targeted university coursework can successfully shift teacher attitudes and significantly boost their self-efficacy regarding inclusive geography instruction, providing a tested model for teacher education curricula.

Caveat: The abstract does not disclose the sample size, the duration of the intervention, or whether these changed beliefs translate into actual inclusive classroom practices.

Journal of Geography · doi

Non-synchronism in Global Usage of Research Methods in Library and Information Science from 1990 to 2019

The study analyzed 5,281 research papers from 81 countries published in international Library and Information Science (LIS) journals between 1990 and 2019. The researchers manually annotated a subset of articles to train a deep learning model that automatically classified the research methods used across the entire dataset to compare national methodological profiles.

Why it matters — It demonstrates that methodological preferences for the same research topics vary significantly by country, while also showing that the gap between national and international methodological distributions has steadily narrowed over the past three decades.

Caveat: The classification model's performance metrics and specific accuracy rates are not disclosed in the abstract.

arXiv · pdf · doi

Efficient Waste Sorting for Circular Economy: A Confidence-guided comparison between One-Vs-All and One-Vs-Rest Classification Strategies with Human-in-the-Loop for Automated Waste Sorting

This study evaluates and compares One-Vs-All (OvA) and One-Vs-Rest (OvR) classification strategies for automated waste sorting using a custom dataset aligned with the municipal waste categories of Goslar, Germany. The researchers applied varying confidence thresholds to both models to identify uncertain, high-risk samples and route them to human annotators, balancing classification accuracy against human review effort.

Why it matters — It demonstrates how automated waste sorting tools can be customized to localized municipal regulations while incorporating a human-in-the-loop mechanism to manage classification uncertainty and reduce sorting errors.

Caveat: The evaluation is limited to a single dataset tailored specifically to the waste sorting scheme of one German city, Goslar.

arXiv · pdf

Autonomous discovery of traffic laws with AI traffic scientists

The researchers developed TrafficSci, an agentic AI system that automates the discovery of traffic laws through an iterative workflow of evidence scoping, hypothesis induction, and validation. Across four case studies spanning population, network, control, and trajectory scales, the system successfully rediscovered three established traffic laws and identified a previously unreported temporal memory scale in urban driving behavior using trajectory datasets from eight cities.

Why it matters — It demonstrates that AI-driven scientific discovery can be successfully extended from controlled laboratory environments to highly complex, observational urban systems, offering a systematic way to uncover hidden regularities in city-scale mobility data.

arXiv · pdf