Urban Currents

Semantic enrichment of 3D city models via roof material classification for urban greening and heat island mitigation

The researchers developed an end-to-end pipeline that classifies roof materials into five categories using high-resolution RGB imagery masked by OpenStreetMap building footprints, and writes these attributes back into CityGML datasets. This classified data was integrated into an urban heat island screening workflow to simulate cool-roof and green-roof retrofits across Hamburg, Paris, and Madrid, finding potential average roof temperature reductions of 0.83 K, 0.16 K, and 0.6 K respectively under the most favorable scenarios.

Why it matters — It establishes a reproducible, data-efficient method to automatically enrich 3D city models with roof material attributes, enabling city-scale microclimate simulations and greening scenario planning without manual material surveys.

Sustainable Cities and Society · pdf · doi

Measuring residents’ cognition of residential energy retrofits: scale development and validation

This study develops and validates the Residential Building Energy Retrofit Cognition Scale (RBERCS), an 18-item instrument designed to measure how residents perceive and evaluate home energy upgrades. The scale covers five distinct domains: personal status, pre-retrofit building status, retrofit measure performance, social norms, and governance mechanisms. Psychometric testing, including exploratory and confirmatory factor analyses, confirmed the scale's reliability, construct validity, and nomological validity through positive associations with overall attitudes toward retrofits.

Why it matters — It provides a standardized, psychometrically validated tool to measure resident perceptions, allowing policymakers and program designers to segment markets and tailor energy efficiency campaigns based on specific cognitive barriers or drivers.

Journal of Housing and the Built Environment · pdf · doi

Temporal Regimes of Environmental Sustainability in G7 Countries: A Causal Decision Support Framework Based on MiniROCKET

This study analyzes annual environmental and macroeconomic data for G7 countries from 1997 to 2021 using a decision-support framework that combines MiniROCKET temporal feature extraction, Principal Component Analysis (PCA), and Double Machine Learning (DML). The framework evaluates how dominant temporal regimes associate with three distinct sustainability indicators: CO2 emissions, Ecological Footprint, and Load Capacity Factor. The first five principal components explained 89.92% of the temporal feature variance, revealing negative directional associations with all three indicators, with the strongest negative effects found in Italy and positive effects in the United Kingdom.

Why it matters — It provides a multidimensional evaluation framework that moves beyond single-indicator CO2 studies to capture consumption-based ecological pressures and ecosystem carrying capacity simultaneously, revealing that sustainability dynamics are highly heterogeneous across advanced economies and cannot be addressed with one-size-fits-all policies.

Caveat: The bootstrap confidence intervals for the main DML specifications include zero, meaning the identified relationships should be treated as indicative decision-support evidence rather than definitive causal proof.

Applied Spatial Analysis and Policy · pdf · doi

Decomposing the impact of structural change on GDP and transport volume: Evidence from Mexico and USA

This study applies the Logarithmic Mean Divisia Index (LMDI) methodology to decompose transport volume into economic activity, economic structure, and transport intensity across Mexico and the United States. It analyzes how primary, secondary, and tertiary sectors drive or constrain the movement of goods and passengers in both countries. The results show that primary economic activity drives transport in Mexico, whereas tertiary activity dominates in the US, with mining and professional services acting as key sectoral drivers.

Why it matters — It quantifies how structural economic shifts offset efficiency gains, demonstrating that overall transport volume growth driven by economic expansion will outpace the reductions achieved through improved transport intensity.

Research in Transportation Economics · doi

Travel socialization within households: A longitudinal analysis

This study analyzed travel socialization within Dutch households using a two-wave cross-lagged panel model from the Netherlands Mobility Panel. It tracked how travel attitudes and behaviors are transmitted over time among partners and children in two-partner and family households. The analysis revealed that sustainable travel modes show bidirectional influence between partners, mothers act as the primary influencers of sustainable travel attitudes and behaviors for both children and partners, and car use exhibits mutual influence between parents and children.

Why it matters — It demonstrates that household travel behavior is not merely a collection of individual choices but a dynamic, gendered socialization process. This allows planners to design mode-specific interventions that target the most influential household members, such as focusing on mothers to promote cycling or children to reduce family car use.

Caveat: The findings are based on longitudinal data from a single country, the Netherlands, which has a highly distinct cycling and transit infrastructure that may limit generalizability to other national contexts.

Travel Behaviour and Society · pdf · doi

Improving ridepooling reliability using shareability shadows

The study introduces a passenger-specific predictive model that calculates a Shareability Score using shareability shadows and expected demand rates to estimate travel-time unreliability in ridepooling systems. This score is used to apply targeted operational rules, guaranteeing zero travel-time updates for low-score passengers. Simulations using ridepooling data from Manhattan and Utrecht demonstrate that, at a 29% passenger rejection rate, this approach reduces the share of passengers experiencing travel-time updates by up to 47.8% and guarantees a fully reliable trip for 83.9% of passengers at assignment.

Why it matters — It shifts ridepooling reliability management from uniform, system-wide policies to targeted, passenger-level interventions. This allows operators to guarantee travel-time reliability for a large majority of riders without sacrificing overall system efficiency across different fleet sizes and network structures.

Caveat: The performance and reliability improvements are demonstrated through simulations rather than a real-world pilot deployment.

Transportation Research Part C Emerging Technologies · doi

Reinforcement learning in railway research and applications: A review

This systematic scoping review analyzed 162 studies published between 2006 and March 2025 that apply reinforcement learning (RL) to railway systems. The analysis categorized the literature into six domains, including traffic planning, automated driving, and maintenance, evaluating how researchers formulated states, actions, rewards, and validation environments.

Why it matters — It establishes that while RL is a growing research area for sequential railway decisions, the field suffers from highly fragmented task formulations, a lack of shared modeling standards, and an over-reliance on simulation-based testing without real-world deployment evidence.

Caveat: The findings are limited by a lack of comparable benchmarks, generalizability evidence, and explainability across the reviewed studies.

Transportation Research Interdisciplinary Perspectives · pdf · doi

Urban green space and peri-urban forest microclimates in the city of Augsburg, Germany: A seasonal and synoptic perspective

This study analyzed microclimate data collected over two annual cycles across five structurally diverse urban green spaces and one vegetated residential district in Augsburg, Germany. The researchers compared temperature dynamics across mixed forests, pine forests, heathland, and an urban park, finding that mixed forests provided the highest daytime cooling and reduced maximum Modified Physiologically Equivalent Temperature by approximately 6 °C during a heatwave. Mixed-effects modeling revealed that tree density had the strongest overall association with lower air temperatures, while taller trees correlated with daytime cooling but nighttime warming.

Why it matters — It demonstrates that the cooling efficiency of urban vegetation is highly dependent on diurnal cycles and synoptic weather conditions, showing that dense, mixed peri-urban forests are more effective at mitigating heat stress than parks or pine forests.

Caveat: The empirical findings are derived from a single German city, which may limit direct generalization to different climatic zones.

Urban forestry & urban greening · doi

“It’s not the place, it’s the crowd”: an explorative study on queer mobilities and spatial practices in Brussels, Belgium

This study investigates the mobility experiences and spatial practices of queer individuals in Brussels, Belgium, using qualitative data gathered from focus groups. The research details how participants employ coping strategies to navigate public spaces, negotiate safety, and make transport mode choices, revealing that female participants often use cycling as a safety mechanism while male participants view it as an aspirational or sustainable option.

Why it matters — It demonstrates that feelings of safety and urban belonging for queer people are dictated by the social composition of a space rather than just its physical design, establishing a direct link between perceived gender, sexual identity, and daily transit choices.

Caveat: The findings are based on qualitative focus groups within a single European capital, which may limit their direct generalizability to other urban contexts with different transit infrastructures or social dynamics.

Journal of Urban Mobility · pdf · doi

Urban air mobility vertiports: A bibliometric analysis and scoping review of physical site selection, design, and integration with ground transport

This study synthesizes the literature on urban air mobility vertiports by conducting a bibliometric mapping and qualitative scoping review of 81 peer-reviewed papers published between 2015 and 2025. It maps how research has evolved from basic spatial coverage models to more complex analyses incorporating wind, noise, equity, land-use compatibility, and multimodal ground integration.

Why it matters — It provides a unified framework that connects network-level site selection with building-scale design and passenger terminal experience, identifying critical gaps in current research regarding rooftop operations, charging infrastructure layouts, and multi-scale governance.

Caveat: The findings are based on a scoping review of existing literature rather than new empirical data or physical testing of vertiport designs.

Journal of Urban Mobility · doi

Research on Road Traffic State Reconstruction Using License Plate Recognition Data

The study develops an improved least-squares path flow estimation model that relies solely on license plate recognition data, reconstructing missing vehicle trajectories via particle filtering and estimating turning flows through tensor decomposition. The method was validated using 248,798 license plate records from 22 intersections in Kunming, China, during peak traffic periods. The model's estimation errors remained stable at sampling rates above 60%, but degraded rapidly below this threshold.

Why it matters — It establishes a concrete empirical threshold of 60% sensor coverage for deploying license plate recognition systems, allowing traffic planners to estimate path flows and origin-destination patterns without relying on sparse mobile sensor data or complex modeling assumptions.

Caveat: The empirical performance threshold was determined using data from a single city during peak hours, which may limit its immediate applicability to off-peak periods or different network topologies.

Journal of Transportation Engineering Part A Systems · doi

Identifying and Assessing Road Infrastructure Development Potential Considering Land-Use Uncertainty: An Example of Developing Highways in Canton Zürich, Switzerland

The study presents a three-step scenario planning methodology that generates multiple infrastructure changes, models diverse future land-use scenarios, and dynamically allocates mobility demand to assess highway developments. The framework evaluates proposals using societal indicators including construction costs, travel time delays, noise externalities, and environmental emissions, and is demonstrated on a highway corridor case study in Dübendorf–Hinwil, Switzerland.

Why it matters — It provides a systematic way to synchronize land-use and transport planning during early, highly uncertain stages, allowing planners to transparently prioritize infrastructure variants before consensus is reached.

Caveat: The methodology is demonstrated on a single highway corridor in Canton Zürich, meaning its performance across different institutional or geographic planning contexts remains to be verified.

Journal of Urban Planning and Development · doi

Cities and political violence in West Africa

This study analyzes the spatial distribution of political violence across 17 West African countries from 2012 to mid-2025, comparing urban agglomerations to their surrounding peripheries. The analysis reveals that while the majority of violence occurs within and around urban areas, large cities experience most incidents inside their boundaries, whereas smaller border towns see more violence just outside their limits. Over the study period, the proportion of urban-centric violence decreased, indicating a gradual shift of conflict toward rural areas.

Why it matters — It demonstrates that West African insurgent movements strategically exploit both urban and rural environments rather than following a simple urbanization-of-conflict trend, showing that security strategies must shift toward mobile, community-based approaches in peripheral and border regions.

arXiv · pdf

Forced Displacement of People Experiencing Homelessness: Housing and Movement Outcomes after Encampment Clearances

This study analyzes longitudinal street outreach data using relational event models to track where unhoused individuals relocate after encampment clearances. The analysis models post-removal outcomes, including tract-to-tract migration, shelter entry, housing placement, and the risk of losing contact with service providers.

Why it matters — It demonstrates that forced displacement fails to transition people indoors and instead disperses them nearby in smaller groups, while significantly increasing the risk that service providers lose track of clients. This directly challenges the efficacy of clearance policies by showing they reduce the visibility of homelessness rather than resolving it.

Caveat: The analysis relies on street outreach data, which may underrepresent displaced individuals who completely lose contact with service networks immediately after a clearance.

arXiv · pdf

Estimating Heterogeneity in Travel Mode Choice Shifts with Causal Forests

This study applies a non-parametric causal forest machine learning model to 802,935 trip records from the 2017 and 2022 waves of the US National Household Travel Survey to measure pandemic-induced shifts in travel mode choice. The model estimates an overall 1.86 percentage point increase in car-mode share, alongside declines of 0.38 percentage points for public transit and 1.57 percentage points for walking. The shift toward driving was most pronounced for trips under one mile, female travelers, and households earning over USD 200,000 annually.

Why it matters — It introduces causal machine learning to travel mode choice analysis, demonstrating how planners can isolate and quantify highly specific, unequal behavioral shifts across diverse demographic groups rather than relying on uniform average effects.

Caveat: The study relies on comparing distinct survey waves from 2017 and 2022 as control and treatment groups to proxy the pandemic's causal impact, which may conflate the pandemic with other temporal changes.

arXiv · pdf

A Review of Vision-Based Vehicle Detection for UAV-Based Traffic Monitoring: Experimental Insights and Future Directions

This review synthesizes recent advancements in deep neural network models for vision-based vehicle detection using unmanned aerial vehicles (UAVs). It evaluates how these models handle varying flight altitudes, motion-induced image variations, and high-resolution data processing, while identifying critical gaps in real-time processing, environmental robustness, and integration with existing traffic control systems.

Why it matters — It establishes a clear set of technical requirements—specifically around edge processing and adaptive control integration—needed to transition UAV-based traffic monitoring from isolated computer vision benchmarks into active, responsive urban traffic management systems.

arXiv · pdf · doi