Urban Currents

Joint Optimization of E-Scooter and Public Transit Operations

The study develops a bilevel optimization framework to jointly allocate resources for public transit and e-scooters, minimizing operational costs at the upper level while simulating commuter mode choice and congestion costs at the lower level. Tested on various scales in Skövde, Sweden, the model uses a simulated annealing algorithm that reduces computation time by 60% to 70% compared to Gurobi while remaining within 1% of the optimal solution. Scenario analysis shows that as demand increases, the e-scooter mode share rises from 0.18% to 29.14% within the multimodal network.

Why it matters — It provides a computationally efficient method for transit agencies to co-optimize bus schedules, e-scooter fleet sizes, battery allocation, and charging infrastructure, rather than planning micromobility and mass transit in isolation.

Caveat: The framework's performance and demand dynamics were evaluated using a single mid-sized Swedish city, which may not reflect the complexity or transit dynamics of larger metropolitan areas.

Journal of Transportation Engineering Part A Systems · doi

Holistic fairness considerations in facility placement decisions

The study introduces a novel fairness metric combined with network centrality to evaluate facility placement, applying it to census tracts in Spokane County, Washington. Using cluster analysis on public demographic and socioeconomic data, the researchers assessed the distribution of existing facilities. The analysis revealed that while undesirable facilities are inequitably distributed, key desirable facilities like hospitals and parks are optimally located for both fairness and centrality.

Why it matters — It provides a replicable framework for planners to balance spatial efficiency with demographic equity, moving beyond pure centrality optimization which often exacerbates disparate access to services.

Caveat: The empirical findings and facility classifications are demonstrated using a single case study of Spokane County.

npj Urban Sustainability · pdf · doi

Analysis of travel activity patterns and charging choices of BEV users in Japan

This study analyzed the charging choices of 441 battery electric vehicle users in Japan's Chubu and Kanto regions using three structural equation models. The models integrated socio-demographics, acceptable waiting times, and activity-based travel patterns, specifically disaggregating trips into commuting, shopping, and leisure. The analysis revealed that commuting trips strongly align with home charging, shopping trips are highly associated with public fast charging, and higher-income households rely more on residential charging compared to middle-income households.

Why it matters — It demonstrates that electric vehicle charging decisions are not isolated events but are systematically embedded within daily activity routines, showing that infrastructure planning must align with specific trip purposes and regional housing densities rather than just spatial coverage.

Caveat: The findings are based on self-reported survey data from two specific regions in Japan, where local residential densities and parking availability heavily influence the feasibility of home charging.

Transportation · pdf · doi

A directional latent demand index for city links: from mobility traces to service gaps

This study introduces a directional Latent Demand Index that uses a pooled Poisson regression framework on large-scale, anonymized mobility traces to identify unmet travel needs. By combining this index with a schedule-based Transit Service Index derived from GTFS data, the framework maps directional service gaps at the link level. Applying the method to Pittsburgh, Pennsylvania, revealed that 43.7% of high-demand directed edges receive below-median transit service, with these mismatches concentrated in specific, asymmetric corridors.

Why it matters — It provides a scalable, link-level diagnostic tool that captures directional asymmetries like underserved reverse commutes, which traditional place-based metrics and realized ridership data systematically overlook.

Caveat: The empirical findings and identified service gaps are demonstrated using data from a single city, Pittsburgh.

Computational Urban Science · pdf · doi

Controlled environment agriculture as urban infrastructure: advancing a public agenda for spatial equity

This perspective paper outlines a framework to transition Controlled Environment Agriculture (CEA) from a venture-backed, profit-driven model into a public urban infrastructure. It proposes integrating Geospatial Artificial Intelligence (GeoAI), High-Performance Computing (HPC), and urban informatics to optimize site selection for underserved communities, manage crops adaptively, and align indoor farming with broader urban systems.

Why it matters — It shifts the conceptualization of indoor farming from a isolated technological novelty to a public-interest utility, providing a planning pathway to leverage advanced spatial data tools for food justice and equitable infrastructure distribution.

Caveat: The paper presents a conceptual framework and agenda rather than an empirical evaluation or deployment of the proposed GeoAI and HPC tools.

Urban Informatics · pdf · doi

Spatiotemporal machine learning for building demolition risk modelling

This study develops a spatiotemporal machine learning framework to predict building-level demolition risk across Switzerland. The models integrate GIS-based spatial data, building characteristics, surrounding built environment attributes, local demolition dynamics, and socioeconomic indicators, using Shapley Additive Explanations (SHAP) to interpret the key drivers of demolition.

Why it matters — It establishes a predictive benchmark that moves beyond aggregate lifetime distribution functions, allowing urban planners to identify specific, high-risk buildings for targeted material reuse and recycling before they are demolished.

Caveat: The predictive models had to be adjusted to handle an extreme class imbalance due to the very low base rate of actual demolition events across the study area.

Computers Environment and Urban Systems · doi

Family support and the built environment as determinants of active school commuting among adolescents in Chile

A cross-sectional study of 91 parent-child dyads from secondary schools in Chile's Maule region analyzed the factors influencing active school commuting. Using binary logistic regression, the study found that only 23.1% of students walk or cycle to school, with longer distances reducing the likelihood of active commuting (OR = 0.54). Conversely, active commuting was strongly associated with parents who exercise regularly (OR = 4.32) and parents who permit their children to walk to school (OR = 4.73), while individual student health and academic traits showed no significant relationship.

Why it matters — The findings demonstrate that parental behavior and permission, alongside physical distance, are much stronger determinants of active commuting than an adolescent's own physical or mental health profile. This shifts the focus for school mobility interventions in Latin American contexts from targeting individual student habits to addressing parental trust and family-level physical activity.

Caveat: The study relies on a small sample size of 91 dyads from a single region in Chile, which limits the generalizability of the findings to other middle-income or Latin American contexts.

Journal of Transport & Health · doi

Urban ponds as nature-based solutions for heat mitigation and temperature regulation in highly urbanised regions

This study analyzed the thermal impact of urban ponds in Japan's East Harima region between 1985 and 2022 using remote sensing imagery and spatial statistics. Out of 337 ponds identified in 1985, 90.8% provided cooling with an average reduction of 2.95 °C, whereas by 2022, only 274 ponds remained, with 75.18% providing cooling at an average reduction of 3.70 °C. While increased surrounding impervious surfaces enhanced the local thermal gradient and cooling intensity of surviving ponds, fragmentation and the installation of floating solar panels degraded some ponds into localized heat islands.

Why it matters — It demonstrates that small urban water bodies can offer enhanced localized cooling as cities densify, but warns that the physical degradation and modern industrial uses of these ponds, such as floating solar installations, can actively reverse their microclimate benefits.

Caveat: The study relies on remote sensing imagery from early summer to estimate surface temperature regulation, which may not fully capture diurnal variations or air temperature dynamics experienced by citizens.

Landscape and Urban Planning · doi

Urban green space percolation: From connectivity to ecological management mapping

The study develops a percolation-based framework to analyze dynamic urban green space connectivity and translate it into spatial management priorities. Applied to Nanjing, China, the model constructs a least-cost network, identifies critical transitions and merger hierarchies as edge costs increase, and delineates five distinct management zones, a three-level edge hierarchy, and a four-type node typology. Robustness testing demonstrated that protecting these prioritized nodes and edges effectively slows the decline of the network's global efficiency under simulated removals.

Why it matters — It moves beyond static connectivity thresholds by mapping the exact sequence in which isolated green spaces merge, allowing planners to target specific cross-barrier links and stepping-stone nodes that are mathematically proven to maintain system-wide ecological cohesion.

Caveat: The framework's planning utility depends on local calibration of the cost-weighted distances, and its performance was validated using simulated removal rules rather than empirical ecological flow data.

Sustainable Cities and Society · doi

Revealing land-use functional patterns and associated factors in China’s high-speed rail station areas: Based on self-organizing maps and explainable machine learning algorithms

This study analyzed 1,570 high-speed rail station areas across China using self-organizing maps and XGBoost-SHAP models to classify their land-use structures. The areas were categorized into four distinct functional patterns: public facility-dominated, balanced multifunctional, industry-dominated, and commercial service-dominated. The analysis identified land-use diversity, bus accessibility, road network length, industrial structure, and distance to the city center as the primary variables driving these functional differences.

Why it matters — It establishes a national-scale taxonomy of high-speed rail station areas, moving beyond single-city case studies to demonstrate how regional industrial bases and network structures shape local land-use patterns.

Caveat: The study identifies associations and nonlinear relationships between spatial variables and station patterns, but does not establish direct causal links.

Sustainable Cities and Society · doi

Consumer-product attachment framework on electric motorcycle conversion adoption: an integrated choice and hierarchical latent variables

This study analyzed the adoption of electric motorcycle conversion (EMC) using a choice experiment and an Integrated Choice and Hierarchical Latent Variables model applied to motorcyclists in Bali. The model incorporates psychological factors from a consumer-product attachment framework—including indispensability, irreplaceability, and self-extension—alongside traditional attributes like conversion time, battery capacity, charging duration, and costs.

Why it matters — It demonstrates that psychological attachment to an existing vehicle is a key determinant of conversion adoption, showing that owners who view their current motorcycles as indispensable are more likely to adopt EMC, with this effect being particularly strong and consistent among low-income riders and owners of older vehicles.

Caveat: The findings are based on a choice experiment conducted in a single region, Bali, which may limit generalizability to areas with different cultural or economic relationships to motorcycle ownership.

Transportation Research Part A Policy and Practice · doi

Arctic shipping route design with CO2-black carbon trade-offs

A two-stage stochastic multi-objective optimization model was developed to minimize expected costs, CO2, and black carbon (BC) emissions under sea ice uncertainty, solved using a Criterion-Space Benders Decomposition algorithm. Testing on 16 synthetic and three AIS-informed instances revealed that optimizing for CO2 alone is an inadequate proxy for BC, as bi-objective models produced 20% to 128% more BC at matched cost or CO2 levels. The model demonstrates that 56.5% to 84.6% of BC can be mitigated through operational adjustments alone, while near-complete elimination requires cost premiums up to 500%.

Why it matters — It proves that greenhouse gas reduction strategies do not automatically minimize short-lived climate pollutants like black carbon in Arctic shipping, establishing a quantitative framework to negotiate the steep economic trade-offs of polar emission governance.

Transportation Research Part D Transport and Environment · doi

A general data-driven methodology for predicting future air traffic distributions around airports

This study presents a data-driven methodology that uses unsupervised machine learning and statistical analysis on historical flight tracking data to generate future low-altitude aircraft movements. Tested at Stockholm Arlanda Airport using 2022 tracking data to forecast 2024 traffic, the model requires only traffic intensity and a timeframe as inputs. The resulting simulated day-evening-night noise levels (L_DEN) matched actual 2024 measurements closely, overestimating the contour areas of 45 dB(A) or higher by less than 6%.

Why it matters — It provides a simplified forecasting capability for airport noise and emissions that requires minimal input parameters, bypassing the need for complex operational scheduling models by extracting patterns directly from historical tracking data.

Caveat: The model's predictive accuracy is sensitive to unexpected shifts in airport operations between the baseline and forecast years.

Transportation Research Interdisciplinary Perspectives · doi

Analyzing Spatial Variability of the Urban Heat Vulnerability Index in the Face of Climate Change Employing Geospatial Technology in Halifax, Canada

This study constructed a Heat Vulnerability Index (HVI) for Halifax, Canada, across four time points (2006, 2011, 2016, and 2021) by integrating remote sensing and socioeconomic datasets. The index combines 16 normalized variables representing exposure, sensitivity, and adaptive capacity using an equal-weight approach. The analysis revealed that heat vulnerability peaked in 2021, with the regional center consistently identified as a high-risk zone across all analyzed years.

Why it matters — It establishes a multi-decade spatial baseline of heat vulnerability at the local level, allowing planners to pinpoint persistent high-risk urban zones and track how vulnerability shifts over time in response to climate change.

Caveat: The index relies on an equal-weighting scheme for its 16 variables, which assumes all socioeconomic and environmental factors contribute equally to heat vulnerability.

Journal of Urban Planning and Development · doi

The Role of Large-Scale Developments in Defining Urban Image: The Case of Mina District, Doha, Qatar

This study evaluates how large-scale urban developments shape city image by analyzing Doha's Mina District regeneration project across six dimensions: skyline, public spaces, architecture, festivals, heritage, and public art. Using survey data and spatial analysis, it measures the alignment between the government's projected tourism brand and the actual perceptions of local residents.

Why it matters — It demonstrates how master-planned cultural districts successfully translate official branding into public perception, providing a framework for emerging cities to evaluate whether mega-projects achieve their intended identity-building goals.

Caveat: The study relies on a single case study of a highly specific, government-backed waterfront district in Doha, which may limit its applicability to organic or less-regulated urban developments.

Journal of Urban Planning and Development · doi

Revitalizing Public Urban Places through Cultural and Political Memory: A Technological Approach with LLMs and Augmented Reality

This paper presents a conceptual methodology that integrates digital twins, virtual reality, and Large Language Models (LLMs) with spatial computing hardware like the Apple Vision Pro to reconstruct historical urban landscapes. The framework combines semantic image search and visual storytelling to overlay cultural, political, and environmental timelines onto physical public spaces.

Why it matters — It establishes a theoretical pathway for using immersive spatial computing to preserve collective memory and visualize urban transformation, allowing practitioners to design interactive heritage experiences directly within physical urban environments.

Caveat: The methodology is conceptual and lacks a concrete empirical case study, deployment evaluation, or user testing to validate its technical feasibility and real-world impact.

arXiv · pdf

Spatio-Temporal Wildfire Spread Prediction in Canada using a Video Swin-Hybrid-U-Net and Satellite Imagery

A deep learning framework combining a Video Swin Transformer encoder with a convolutional U-Net decoder was developed to predict next-day wildfire incidence maps. The model processes three-day sequences of meteorological and environmental variables using a curated dataset of major Canadian wildfire events spanning 2014 to 2023.

Why it matters — It establishes a scalable, spatio-temporal forecasting capability for Canadian landscapes using exclusively public data from Google Earth Engine, bypassing the scalability limits of previous models.

arXiv · pdf

Optimising Temporary Accommodation Placement Across London with AI-Powered SaaS in E-Governance Systems

Researchers built and piloted DOMUS, a cloud-native, AI-enabled decision-support system designed for the London Borough of Newham to manage temporary housing placements. The system integrates household case records, affordability rules, and live private-rental listings, combining rule-based filtering with large language model-assisted search to match households to properties. A pilot deployment in Newham's secure environment demonstrated reduced search times, improved adherence to placement constraints, and high staff satisfaction compared to manual workflows.

Why it matters — It demonstrates a scalable, ethically governed e-governance architecture that successfully automates complex, rule-bound public administration tasks while preserving officer discretion and statutory compliance.

Caveat: The system's performance and utility metrics are based on a pilot deployment within a single London borough.

arXiv · pdf