Exposure to environmental stressors such as heat and air pollution is rising in urban areas, placing soft mobility users (e.g., pedestrians, cyclists, and scooter riders) at particular risk—a challenge that will only intensify with climate change. This study develops a comprehensive multi-dimensional exposure assessment framework — BREEZE (Bioclimatic Route Evaluation for Environmental haZard avoidancE) — for urban soft mobility routes that integrates environmental stressor modeling, dynamic population weighting, and citizen-oriented navigation tools. The methodology addresses dual objectives: supporting evidence-based urban planning decisions through exposure hotspot identification and enabling personalized route optimization for individual citizens. The BREEZE framework is designed to be transferable across different urban contexts while accommodating variations in data availability and local environmental conditions. The Brussels-Capital Region (BCR) case study demonstrates the framework’s application across different levels and time spans. At the city level and over the annual time span, incorporating dynamic population data — distinguishing between daytime and nighttime populations — enables more accurate identification of exposure hotspots compared to static census-based approaches, thereby supporting targeted intervention planning. At the individual level and over a daily time span, integrating environmental modeling into route optimization shows that personalized path selection based on hazard avoidance criteria can reduce strong heat or air pollution exposure during daytime or nighttime travel by up to 20% and 40%, respectively. The framework demonstrates how environmental modeling can be translated into actionable insights for both strategic urban planning and individual mobility decision-making, bridging the gap between scientific assessment and practical public health applications.
Urban spatial models often assume that nearby borrowing improves prediction across target variables, but this assumption is rarely tested across distinct urban feature families within a city. We evaluate this question on a citywide hexagonal grid of Cincinnati, Ohio (2016 cells, 60 features in 14 feature families) using multimodal masked reconstruction. We compare four models spanning linear, local nonlinear, shared-gate graph, and selective-gate graph architectures under full leave-one-neighborhood-out evaluation across 50 neighborhoods, with primary summaries for the 36 neighborhoods having at least 20 cells. Median held-out-neighborhood R2 is the primary comparison metric. Cross-neighborhood transfer is family-specific, and no single borrowing strategy dominates. Ridge performs best for street structure, indicating that some spatially regular families are best reconstructed from local linear combinations. The local nonlinear masked autoencoder (MLP-MAE) holds the highest median for most families with positive transfer, though these differences are not statistically distinguishable from the graph models, indicating that local nonlinear multimodal fusion is often sufficient. Graph-based models help most clearly for topography, while the selective-gate graph model wins many individual neighborhood comparisons for several families despite lower medians, reflecting different error profiles. Graph models also produce more neighborhood-structured representations: contextual embeddings from the selective multimodal spatial autoencoder (SMSA) recover neighborhood structure better than raw principal components or local-only embeddings. Learned gates are stable across folds and provide interpretable diagnostics of context reliance, but not a universal ranking of borrowing regimes. These results show that the appropriate spatial borrowing strategy is target-specific rather than universal. For neighborhood-scale interpolation, imputation, and urban monitoring, one spatial modeling recipe can systematically misrepresent some feature families even within the same city.
Urban noise has emerged as a critical environmental stressor affecting public health in densely populated cities. Most existing studies, however, rely on temporally aggregated approaches that obscure how noise-environment relationships vary across seasons and times of day, limiting analytical insight and spatial coverage.This study presents an integrated spatiotemporal framework for urban noise analysis in Seoul, South Korea, using five years of continuous IoT sensor data from 762 monitoring locations. Temporal contexts were defined through Circular K-means clustering, yielding 28 season-by-time-of-day combinations, with meteorological effects explicitly controlled to isolate urban structural contributions. Independent ExtraTrees models were trained for each context and applied to unmeasured locations, with results aggregated to administrative units for neighborhood-scale exposure assessment.Results reveal that noise-environment relationships are not temporally invariant — while road infrastructure consistently ranked as the primary predictor, its relative importance was systematically displaced by commercial activity indicators during evening hours and by topographic features during summer contexts, patterns that a temporally pooled model would fail to detect. Districts with similar mean noise levels exhibited substantially different temporal profiles, underscoring that location alone is insufficient to characterize noise burden. Spatial analysis identified significant neighborhood-scale clustering, with central areas exhibiting both higher exposure and greater temporal variability.These findings demonstrate that context-stratified modeling reveals dimensions of urban noise dynamics that static approaches systematically obscure. The framework is transferable to other data-rich cities seeking evidence-based noise management.
Public disability-inclusive awareness is a critical yet underexamined dimension of inclusive urban development. While existing studies has extensively addressed physical accessibility, services, and institutional arrangements, far less is known about how disability is perceived and discussed in daily urban life. This study investigates the spatiotemporal dynamics of public disability-inclusive awareness across more than 300 Chinese cities using large-scale geotagged social media data. Drawing on nearly 700 million geotagged Weibo posts from 2016 to 2022, we identified 459,694 disability-related posts using a Bidirectional Encoder Representations from Transformers (BERT)-based model. City-level indicators were constructed to capture three interrelated dimensions of public awareness: attention, attitudes, and discourse. The results reveal substantial spatial and temporal heterogeneity in disability-inclusive awareness across Chinese cities. Contrary to conventional expectations, public attention and positive attitudes toward disability do not increase with city size. Instead, many smaller cities in central and western China exhibit higher attention levels and more positive attitudes, whereas several large eastern megacities show comparatively lower engagement. Nationally, disability-related discussions expanded over time in both intensity and thematic breadth, increasingly encompassing diverse disability types, social inclusion, and (in)accessible environments. However, local urban discourse became more concentrated around volunteering and service-oriented themes, accompanied by declining topic diversity across cities. By providing the first large-scale, city-level assessment of public disability-inclusive awareness in China, this study demonstrates the value of geosocial media data for capturing the socio-spatial dynamics of disability-related perceptions and contributes new empirical evidence to support more inclusive urban research and governance.
Aging in place depends on a barrier-free neighborhood travel environment, yet quantitative assessment of the “last kilometer” barrier-free accessibility has not been adequately addressed. This study proposes an assessment framework based on handheld LiDAR point clouds. Near ground obstacles are first detected using elevation interquartile range (IQR) values to construct an obstacle cost map. An eight-neighbor reverse Dijkstra procedure is then used to compute accessible areas, geometric and equivalent travel distances for key neighborhood nodes. Experiments across three sites (including both flat and hilly residential neighborhoods, as well as a school) demonstrate that: 1) The elevation IQR-based obstacle detection method, evaluated at the adopted 0.5 m grid resolution, achieved a Precision of 88.2%, Recall of 83.6%, and F1-score of 85.8% across five manually labelled subsets. 2) Geometric and equivalent distances provide complementary information on wheelchair accessibility. Under the prescribed optimized scenarios, geometric path distance and equivalent travel distance across five functional nodes in the flat terrain neighborhood decreased by 4.24%–76.17% and 17.12%–76.86%, respectively, while the accessible area in the hilly neighborhood increased from 14,501 to 68,761 m2. The proposed framework provides a quantitative means for diagnosing accessibility deficiencies and assessing potential barrier-free modifications in the examined neighborhood-scale environments. To foster further research, the handheld LiDAR point cloud dataset used in this study is publicly available at: https://github.com/c175044/Handheld-LiDAR-Neighborhood-Accessibility-Dataset.
