Urban Currents

Analyzing the daily flows: Exploring shared micro-mobility factors in Venice

The study analyzed 158,401 origin-destination (OD) daily observations over a two-year period across 50 spatial zones in Venice to understand how weather and air quality affect shared micro-mobility. Using a generalized additive mixed model (GAMM), the researchers modeled the non-linear relationships between daily temperature, rainfall, PM10 concentrations, and trip demand. The findings reveal a seasonal, non-linear relationship for temperature, a significant drop in trips during moderate to heavy rain, and a season-dependent, non-monotonic response to PM10 levels.

Why it matters — It demonstrates that micro-mobility demand is not just a function of vehicle availability but is highly sensitive to fluctuating environmental conditions, revealing complex behaviors like season-specific pollution avoidance and localized seasonal demand spikes.

Caveat: The findings are based on a single, highly unique island city context (Venice and the Lido Islands), which may limit generalizability to more typical inland urban environments.

arXiv · pdf

A CAV-based perimeter-free regional traffic control utilizing existing parking infrastructure

This study designs and simulates a regional traffic management strategy that directs a subset of connected and autonomous vehicles (CAVs) with the longest remaining travel distances to temporarily wait at existing parking facilities during congestion. The strategy was tested via simulations under various parking locations, capacities, and CAV penetration rates to measure its impact on travel times.

Why it matters — It introduces a perimeter-free traffic control method that utilizes existing parking infrastructure to mitigate local queue accumulation, demonstrating that holding vehicles with long remaining trips can reduce travel times even for some of the delayed vehicles.

Caveat: The performance of the strategy is demonstrated solely through simulations, and the abstract does not specify the exact scale, city network, or empirical parameters used in the model.

Transportmetrica B Transport Dynamics · doi

Community thermal energy storage, a cost-effective pathway for the decarbonisation of suburban cooling in Australia

This study simulated six neighbourhood-scale cooling and storage configurations for an Australian suburb using a TRNSYS model driven by high-resolution meteorological, load, and tariff data. The analysis compared individual, district, and community-shared systems using either battery or phase-change material thermal energy storage (TES) combined with household solar photovoltaics. A district cooling system with a centralised battery achieved the highest primary energy savings at 98.6% but cost USD 1.36 million, while a community system with centralised TES delivered 86.1% energy savings at a much lower life cycle cost of USD 0.55 million.

Why it matters — It establishes that community-scale thermal energy storage is the most cost-effective architecture for decarbonising suburban cooling, offering a viable alternative to expensive centralized battery systems while still drastically reducing grid dependence.

Caveat: The findings are based on a simulation model of a single Australian suburb, which may limit direct generalisation to regions with different tariffs, climates, or building standards.

Sustainable Cities and Society · pdf · doi

<scp>DIFFERENTIATED INFRASTRUCTURAL CITIZENSHIP</scp> : Claims‐Making and the Limits to Transformative Urbanization in a Fast‐Growing Small City

This study introduces the concept of differentiated infrastructural citizenship (DIC) through an empirical analysis of water access, caste-based land tenure, and political claims-making in Tiruppur, a rapidly growing small city in India. It examines how a hybrid waterscape and institutionally sparse local governance structures shape how residents negotiate and access infrastructure.

Why it matters — It demonstrates that physical proximity to local state offices does not guarantee successful civic claims, showing instead that everyday political participation in small, rapidly urbanizing cities often yields only incremental, unequal improvements rather than systemic, transformative change.

Caveat: The findings are based on a single, specialized industrial and agrarian small city in India, which may limit direct generalizability to larger metropolitan areas or different national governance contexts.

International Journal of Urban and Regional Research · doi

GIS-based multi-criteria and network accessibility analysis for identifying suitable areas for urban leisure green spaces in Sintra, Portugal

This study models pedestrian accessibility to leisure green spaces in Sintra, Portugal, using network-based service area analysis with a 15-minute walking threshold that accounts for topography and street connectivity. The analysis reveals that 21% of the population (approximately 82,000 residents) lacks walkable access to these spaces, particularly in dense eastern neighbourhoods. To address this, a GIS-based multi-criteria decision analysis was deployed to identify 96 highly suitable candidate areas, selecting five priority sites ranging from 14 to 27 hectares.

Why it matters — It provides a replicable, equity-oriented spatial framework that moves beyond simple buffer-based proximity measures to help planners implement 15-minute city goals. The proposed interventions demonstrate a measurable potential impact, showing how five targeted site selections could extend walkable green space access to an additional 68,200 residents.

Discover Cities · pdf · doi

Taking refuge: Prioritising park climate adaptation strategies based on urban design and planning professionals' insights

A survey of 87 urban design and planning professionals from local government, state government, and private practice in Western Australia evaluates the perceived effectiveness of park climate adaptation strategies. The study focuses on local parks under hot daytime conditions across three distinct climate zones: Hot semi-arid (BSh), Hot desert (BWh), and Mediterranean (Csa).

Why it matters — It establishes a professional consensus favoring nature-based solutions for park cooling, while simultaneously highlighting a critical planning conflict between these irrigation-dependent strategies and declining regional rainfall.

Caveat: The findings rely entirely on the perceived effectiveness of strategies by professionals rather than empirical microclimate measurements.

Cities · doi

Stories of control: Digitalization narratives in smart city strategy papers of German municipalities

This study analyzed the smart city strategy documents of 12 German municipalities using the Narrative Policy Framework and the capability approach. The mixed-methods analysis identified six distinct digitalization narratives across three public problem areas, focusing on how citizens are characterized and how digital technology is projected to affect their capabilities. Cluster analysis was then used to group the municipalities into narrative profiles and examine how these profiles relate to local structural characteristics such as city size, centrality, and socioeconomic growth.

Why it matters — It reveals how local governments discursively construct the relationship between technology and citizen well-being, showing that smart city strategies are shaped by specific local structural contexts rather than a uniform digital agenda.

Caveat: The findings are based on a small sample of 12 municipalities within a single national context, which may limit how well these specific narrative profiles generalize to other countries.

Cities · doi

Informal transportation and health equity: Air pollution exposure in Latin American MTW microenvironments using neural networks

This study measured PM 2.5 and Black Carbon (BC) exposure across 21 origin-destination pairs in Soledad, Colombia, comparing buses, private cars, and informal motorized three-wheeled vehicles (MTWs). Using portable sensors, GPS, and a Bayesian Neural Network Classifier, the researchers found that buses and MTWs had the highest mean PM 2.5 concentrations at 33.0 and 27.5 Ιg/m³ respectively, with peak MTW exposures reaching 959 Ιg/m³. The predictive neural network model achieved a 70% accuracy rate in identifying high-exposure scenarios based on route, mode, and travel time.

Why it matters — It provides the first empirical evidence of air pollution exposure inside informal motorized three-wheelers in Latin America, revealing a stark gender disparity where women inhale up to 68 times more PM 2.5 and BC than men across the studied transit modes.

Caveat: The study relies on a cross-sectional design in a single Colombian city, which may limit the direct generalizability of the specific exposure levels to other Latin American urban contexts.

Journal of Transport & Health · doi

Does cycling infrastructure expansion lead to gentrification? Empirical insights from Berlin

This study analyzed the relationship between cycling infrastructure expansion and gentrification in Berlin from 2014 to 2021 using annual OpenStreetMap data and socioeconomic indicators mapped to a 1 km² grid. The analysis measured the bike path to street length ratio across on-street and off-street networks, finding that lower socioeconomic status areas actually had higher baseline bicycle-path coverage. Longitudinal and linear probability models revealed no robust citywide association between socioeconomic upgrading or gentrification and the expansion of cycling infrastructure, which was instead primarily driven by proximity to the city center.

Why it matters — It provides empirical European evidence to a debate previously dominated by North American case studies, demonstrating that cycling infrastructure investments do not inherently trigger neighborhood gentrification at a citywide scale.

Caveat: The findings are limited to citywide patterns in Berlin and may not capture highly localized, neighborhood-specific gentrification dynamics driven by individual cycling projects.

Journal of Transport Geography · pdf · doi

Heat and cold at work: Occupational accidents and apparent temperature in the primary and construction sectors across five spanish provinces

This ecological time-series study analyzed the relationship between daily occupational accidents and Apparent Temperature (AT) over a 17-year period across five Spanish provinces, focusing on the primary and construction sectors. Using a Generalized Additive Model combined with a Distributed Lag Non-linear Model, the researchers found that occupational accidents in the primary sector increase with AT in northern coastal provinces, peaking at 34°C in Cantabria (RR 1.73) and 38°C in Pontevedra (RR 1.48). In construction, a statistically significant association was found only in Seville, where the risk of accidents peaks at 40°C (RR 1.38).

Why it matters — The study demonstrates that the relationship between thermal stress and workplace accidents is highly localized and sector-specific, showing that northern agricultural workers face elevated risk at high temperatures while southern construction workers are highly vulnerable to extreme heat.

Caveat: The study relies on an ecological time-series design, which analyzes aggregate provincial data rather than individual-level exposure and circumstances.

Sustainable Cities and Society · pdf · doi

Modelling the effects of travel time use on mode choice and the value of travel time savings: tackling self-selection and endogeneity bias

This study models how travel-based activity preferences influence mode choice and the value of travel time savings (VTTS) using an attitude-based approach. The model controls for endogeneity and self-selection by incorporating traveller characteristics and travel context variables. The findings show that travel time use preferences increase the probability of choosing rail for trips longer than 5 km and car driving for trips under 5 km, while travel-based activities showed non-significant effects on VTTS once these preferences and context variables were controlled.

Why it matters — It demonstrates that accounting for travel-time-use preferences and context variables can effectively eliminate endogeneity bias in VTTS estimates, providing a more accurate framework for predicting mode shifts toward public transport or autonomous vehicles.

Caveat: The study relies on self-reported attitudes and preferences to capture travel time use, which may introduce measurement errors compared to direct observations of in-transit behavior.

Transportation Research Part A Policy and Practice · doi

Visualizing the future of the trucking industry by identifying strategic driving variables through MICMAC analysis

This study analyzes the future of the trucking industry toward 2030 by evaluating 13 strategic variables across political, economic, social, technological, legal, and environmental domains. Using cross-impact matrix multiplication applied to a ranking (MICMAC) analysis, the research models the interrelations and indirect influences among these variables, finding that digitalization and artificial intelligence are highly influential drivers, while company economics remains the most dependent variable and policy influence is projected to decline.

Why it matters — It establishes a structured, systemic hierarchy of the forces shaping freight transport, demonstrating that technological adoption will supersede policy drivers in dictating the economic viability of trucking firms over the next decade.

Caveat: The analysis relies on a structural matrix methodology (MICMAC) which models perceived relationships among variables rather than empirical, real-world operational data.

Transportation Research Interdisciplinary Perspectives · doi

Unlocking public support to scale urban agriculture for sustainable cities

This study adapts the Value-Belief-Norm framework to analyze the social, environmental, and psychological drivers of public support for urban agriculture. It maps public backing across four distinct behaviors: financial donations, active participation, purchasing produce, and policy advocacy. The analysis reveals that support operates through distinct pathways, showing that donations are driven by social factors, participation by personal experience, and both purchasing and policy advocacy by a sense of responsibility.

Why it matters — It demonstrates that public support for urban agriculture is not a uniform attitude but a set of distinct behaviors requiring tailored mobilization strategies, shifting the focus from individual consumption to civic and institutional advocacy.

Caveat: The findings are based on self-reported behavioral intentions and psychological frameworks rather than observed, real-world participation or policy outcomes.

npj Urban Sustainability · pdf · doi

Circular Economy Synergies and Trade-offs in Data Centres

This report analyzes data center sustainability metrics, identifying structural flaws in the Power Usage Effectiveness (PUE) metric, which conflates cooling and power provisioning while misallocating server fan and transformation losses. It maps the operational trade-offs between on-site water use and upstream electricity-related water consumption, proposes an alternative waste heat recovery metric based on avoided energy rather than recovered volume, and applies the 9R circularity framework to data center product design, materials, and grid interactions.

Why it matters — It exposes how the industry's primary efficiency metric (PUE) misdirects sustainability efforts by focusing on infrastructure rather than exploding compute energy, while providing a clearer framework for balancing water-energy trade-offs and grid integration challenges.

arXiv · pdf

How Infrastructure and Streetscape Shape E-Scooter Route Choice: Evidence from Washington, DC

This study analyzed e-scooter route choices in Washington, DC, using GPS trajectory data and a Path Size Logit model. The model combined roadway infrastructure data with visual streetscape features, such as tree, building, and wall coverage, extracted from Google Street View images using computer vision. The analysis revealed that protected bicycle facilities are necessary to attract riders on major roads, whereas both protected and designated lanes suffice on minor roads; it also found that sidewalk riding is common but only asphalt-paved sidewalks yield positive utility.

Why it matters — It demonstrates that the effectiveness of cycling infrastructure is highly dependent on the surrounding roadway context, and proves that while visual streetscape features like summer tree canopy influence route choice, physical roadway infrastructure remains the primary determinant of where e-scooter users ride.

Caveat: The findings are based on a single city, Washington, DC, where specific local regulations and sidewalk paving materials may limit generalizability to other urban contexts.

arXiv · pdf

Modeling E-Bike Route Choice in Washington, DC: A Path Size Logit Approach

This study modeled shared e-bike route choices in Washington, DC, using GPS trajectory data from the Capital Bikeshare system. The researchers estimated a Path Size Logit model using a hybrid choice set of observed routes and shortest paths, combining GIS-based infrastructure data with computer vision-derived street-level greenery features from Street View images. The model reveals that e-bike riders prioritize routes that minimize conflicts with vehicles and pedestrians, and that bicycle facilities have a significantly larger positive impact on route choice along major roads than along minor roads.

Why it matters — It demonstrates that incorporating street-level visual features like tree canopy improves route choice models, and quantifies how the behavioral influence of cycling infrastructure varies depending on the underlying roadway hierarchy and trip length.

Caveat: The findings are based on shared e-bike trips from a single municipal bikeshare system, which may not fully represent the route preferences of private e-bike owners.

arXiv · pdf

Workplace dependence in urban economies

Using hourly population data paired with detailed company records in a European city, this study measures physical workplace attendance across different levels of COVID-19 restrictions to identify the drivers of workplace dependence. The analysis reveals that while industry type and firm productivity dictate physical attendance overall, the influence of income and gender depends heavily on proximity to the city center.

Why it matters — It demonstrates that remote-work inequalities are spatially structured, revealing a 'service trap' where female-majority and income-diverse areas near the urban core remain highly dependent on physical workplaces due to a concentrated, place-bound service economy.

Caveat: The empirical findings are derived from a single, unnamed European city, which may limit generalizability to urban areas with different spatial layouts or economic structures.

arXiv · pdf

Evaluating Machine Learning Models for Post-Wildfire Debris-Flow Prediction

The study evaluated 15 machine learning models, including the Tabular Prior-Data Fitted Network (TabPFN), to predict post-wildfire debris flows using basin-scale observations from the western United States. TabPFN achieved the highest unaugmented performance with a threat score of 0.637, while SHAP analysis revealed that short-duration rainfall intensity and storm accumulation were the primary predictive features. Augmenting the limited training data with TabPFN-generated synthetic samples improved the performance of almost all models, yielding a maximum threat score increase of 0.041 for deep learning architectures.

Why it matters — It establishes a highly accurate predictive framework for post-wildfire hazards and demonstrates that synthetic data augmentation can successfully overcome the chronic shortage of physical debris-flow observations.

arXiv · pdf

School network reorganization under educational and spatial constraints using classical and quantum optimization

An optimization framework for school network reorganization was developed using an Integer Linear Programming formulation that balances demographic trends, territorial accessibility, and educational criteria. The model was validated using a synthetic benchmark generator and applied to a real-world case study of the entire public school network in Italy's Calabria region. Additionally, the formulation was adapted into a constrained quadratic model and executed within a hybrid quantum optimization environment.

Why it matters — It establishes a scalable method for regional school consolidation that can be solved using both classical and emerging quantum computing architectures, allowing planners to evaluate complex policy scenarios and resource allocations across large geographic areas.

arXiv · pdf

An improved car-oriented mean-field theory for stochastic traffic flow models

The study introduces an enhanced mean-field analysis for cellular automata models of single-lane vehicular traffic by merging the Car-Oriented-Mean-Field theory with the 2-site cluster method. This combined approach is designed to capture both short- and long-range spatial correlations, and its performance is demonstrated on the Velocity-Dependent-Randomization model with a maximum velocity of one.

Why it matters — It provides a theoretical framework capable of analyzing traffic models with inhomogeneous stationary states and phase separation, such as those with slow-to-start rules, which classical mean-field theories fail to accurately describe.

Caveat: The analytical improvements are demonstrated only on a simplified traffic model where the maximum velocity is restricted to one.

arXiv · pdf

Joint Spatial and Temporal Generalized Dissimilarity Mixed Modeling (stGDMM) for Beta Diversity

The study introduces the spatio-temporal generalized dissimilarity mixed model (stGDMM), extending previous spatial-only models to capture dynamic changes in beta diversity over both space and time. The model's utility is demonstrated using a biodiversity dataset of Bray-Curtis dissimilarity measures from the Cape Floristic Region in South Africa.

Why it matters — It provides a formal statistical framework to analyze how species composition differences evolve across both spatial distances and temporal intervals simultaneously, resolving stochastic limitations of foundational generalized dissimilarity models.

Caveat: The model's performance and applicability are demonstrated using only a single ecological dataset from South Africa.

arXiv · pdf

Multi-Agent Reinforcement Learning for Online Traffic Scheduling in Time-Sensitive Application

This study develops a multi-agent reinforcement learning framework for online traffic scheduling in Time-Sensitive Networking (TSN) within mobile edge computing environments. Using the Heterogeneous-Agent Proximal Policy Optimization (HAPPO) algorithm, the model treats each TSN queue as an autonomous agent to capture inter-queue dependencies. Simulation results show the framework reduces average frame waiting times by up to 26.8% and worst-case delays by approximately 16.8% compared to existing schedulers.

Why it matters — It introduces a decentralized, adaptive scheduling capability that can handle dynamic, co-located extended reality (XR) traffic flows, moving beyond the static optimization and centralized models that fail when traffic patterns change.

Caveat: The performance improvements and coordination capabilities were evaluated exclusively within a simulated environment rather than a physical mobile edge computing deployment.

arXiv · pdf

Weather- and Location-Aware Agentic Dining Recommendation: Leveraging LLM World Knowledge for Region-Sensitive Contextual Reasoning

The researchers designed and built a prototype dining recommendation system that uses a large language model to orchestrate Google location services and a weather API. Instead of relying on hard-coded rules, the system uses the model's latent cultural knowledge to recommend food that fits both the local weather and regional culinary traditions.

Why it matters — It demonstrates an architectural pattern for context-aware recommendation that replaces brittle, hand-crafted rule tables with natural language reasoning, allowing systems to scale region-specific recommendations without specialized training data.

Caveat: The system's performance has not been validated through a formal user study, and the underlying model carries a risk of cultural stereotyping in its location-based reasoning.

arXiv · pdf

Personalized Federated Sparse Adaptation of Time-Series Foundation Models

This study introduces a personalized federated sparse adaptation framework for adapting time-series foundation models to building energy forecasting. The framework uses a heterogeneous temporal mixture-of-experts adapter with a sequence-level router that maps 168-hour context windows to specialized experts. Evaluated across 50 buildings and three foundation model backbones, the personalized federated learning approach consistently outperformed both global federated learning and purely local training.

Why it matters — It demonstrates that adapting large time-series models for private, distributed building energy data requires a balance between shared global knowledge and client-specific temporal behaviors, proving that adaptation strategies must be tailored to both the specific model backbone and the individual client.

arXiv · pdf