No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Environmental factors for outdoor jogging in Beijing: Insights from using explainable spatial machine learning and massive trajectory data
Abstract
Outdoor jogging, as a physical exercise beneficial for health, has proliferated worldwide. However, understanding the nonlinear and heterogeneous associations between environmental factors and jogging behavior remains challenging. This study established an explainable spatial machine learning framework combining Geographically Weighted-Random Forest (GW-RF) and SHapley Additive exPlanation (SHAP) model to address the nonlinearity, spatial heterogeneity, and interpretability. Using large-scale GPS trajectories and multi-source big data, this study provides the global and local explanations of nonlinear associations in Beijing, China. Our findings highlight that (1) Built Environment (BE) factors play a more important role than visual landscape factors in determining jogging behavior, with sports facilities such as tracks and parks having higher contributions. (2) All environmental variables, including the BE, visual landscape, and social economics, exhibit nonlinear and threshold effects on jogging behavior. (3) Certain factors such as population density, number of parks, Greening View Index, and Sky View Index exhibit differential effects across different periods, regions, and mobility patterns of jogging. (4) Compared with the Ordinary Least Squares model, Geographically Weighted Regression model, and Random Forest model, GW-RF model demonstrates improved performance in modeling and predicting the jogging flow. The findings have important implications for urban planners seeking to create a supportive environment that promotes outdoor jogging.
... Recent studies have explored the relationship between the built environment and jogging [14][15][16][17][18]. Some environmental factors related to jogging have already been studied, such as accessibility [19], temperature [20], air quality [21,22], greenery [23,24], and slope [25]. ...
... The classification of environmental perception and environmental elements was primarily based on the part of speech of the words, while the categorization of factors mainly stems from the study of literature related to jogging. Environmental perception factors involve abstract spatial sensations, predominantly represented by adjectives, including safety [47,48], vibrancy [28], cleanliness [49], slope [25], width [15,50], accessibility [14,19], connectivity [28], air quality [22,51], scenic beauty [14,18], greenery [23,24], lighting [52,53], soundscape [28], and temperature [20,54]. Environmental elements factors primarily include concrete spatial entities, represented mostly by nouns, including vehicles [51,53], pedestrians [14,32], traffic infrastructure [28,55], landscape [14], green space [24,50], waterfront space [26,32], service facilities [14,15], lighting facilities [52,53], pavement [48,56], culture [57], buildings [15,58], and vertical elements [25]. ...
Land use types other than specialized athletic fields provide a variety of jogging environments, addressing the shortage of urban fitness facilities and promoting urban health as well as sustainability. Currently, there is limited research comparing the differences in jogging support among various land use types, which can assist decision-makers in setting priorities and targeted strategies for urban renewal, especially in urban built-up areas with limited land resources. Initially, spatial information, statistical data, and recommendation reason text were extracted from recommended routes in mobile fitness apps and categorized into six land use types. Subsequently, spatial potential was measured through descriptive statistics, buffer area analysis, spatial autocorrelation analysis, and line density analysis. Environmental preferences were gauged by text analysis using jieba word segmentation and grouped word frequency calculation. Finally, the measurement results of different land uses were compared, including scale differences, spatial differentiation, environmental perception, and environmental elements. The research found that streets, residential areas, campuses, parks, and greenways possess significant potential to support jogging, particularly streets. These types of land use exhibit varying spatial potentials and attractions in environmental preferences. Targeted recommendations have been proposed to support the renewal of urban built-up areas and research in related fields.
... 所有模型中都具有显著影响。人口密度、土 地利用混合度、公交站点密度的q值在各个模 型中都较高，说明3个因子对居民户外慢跑 影响产生稳定的显著影响。此外，GVI的解 释力均高于NDVI，表明视觉绿色对于慢跑更 具影响力，这与已有研究结论一致 [5,33] ，其原 表2 交互探测中的因子关系 ...
Urban streets provide environment for road running. The study proposes a non-parametric approach that uses machine learning models to predict road running intensity. The models were developed using route check-in data from Keep, a mobile exercise application, and street geographic information data in Beijing’s core district. The results show that blue space and trail continuity are the most important factors in improving road running intensity. There is an optimum design value for the sky openness and the street enclosure, which need to be balanced with shade while meeting the light of the road. And it is also important to provide appropriate visual permeability. Furthermore, unlike daily activities, it was found that higher function mixture and function density did not have significant positive effects on the road running intensity. This study provides empirical evidence on road running and highlights the key factors that planners, landscape architects, and city managers should consider when design running-friendly urban streets.
Pedestrians' walking duration related to multiple street view (SV) characteristics. • Environment-walking associations differed between weekdays and weekends. • SV environments were primarily associated with weekend walking. • SV-derived people within a threshold may be more optimal in promoting walking. • Walking policy and urban planning need to account for specific street environments. Different aspects of the built and natural environment appear related to people's walking behavior. State-of-the-art transport studies typically incorporate built environmental measures (e.g., density, diversity, design). However, street view (SV) environments capturing how pedestrians perceived their surroundings on site are understudied. Therefore, this study examined possibly non-linear associations between multiple SV-derived environmental features and pedestrians' walking duration in Amsterdam, the Netherlands. We used travel survey data (N = 1,886) between 2014 and 2017. SV-derived environmental measures (e.g., cars and people) were extracted from SV images through a fully convolutional neural network. Covariate-adjusted generalized additive mixed models were fitted to the data. Our results showed that walking-SV features associations differed between weekdays and weekends. On weekdays, pedestrians walked more in neighborhoods with fewer individual standing walls and lower address density. On weekends, pedestrians' walking duration increased with more street greenery, fewer cars, higher address density, pronounced land-use diversity, and further distances to train stations. Non-linear associations were found only in the case of weekday SV-derived people, even after adjusting for other neighborhood characteristics (e.g., address density, land-use mix, and street connectivity). Our findings suggest that SV environmental features complement the typically used built environmental measures to explain pedestrians' mobility. Policy-makers and urban planners are advised to incorporate characteristics of the street environments, also should not only rely on the conventional thinking "the more, the merrier".
Running can promote public health. However, the association between running and the built environment, especially in terms of micro street-level factors, has rarely been studied. This study explored the influence of built environments at different scales on running in Inner London. The 5Ds framework (density, diversity, design, destination accessibility, and distance to transit) was used to classify the macro-scale features, and computer vision (CV) and deep learning (DL) were used to measure the micro-scale features. We extracted the accumulated GPS running data of 40,290 sample points from Strava. The spatial autoregressive combined (SAC) model revealed the spatial autocorrelation effect. The result showed that, for macro-scale features: (1) running occurs more frequently on trunk, primary, secondary, and tertiary roads, cycleways, and footways, but runners choose tracks, paths, pedestrian streets, and service streets relatively less; (2) safety, larger open space areas, and longer street lengths promote running; (3) streets with higher accessibility might attract runners (according to a spatial syntactic analysis); and (4) higher job density, POI entropy, canopy density, and high levels of PM 2.5 might impede running. For micro-scale features: (1) wider roads (especially sidewalks), more streetlights, trees, higher sky openness, and proximity to mountains and water facilitate running; and (2) more architectural interfaces, fences, and plants with low branching points might hinder running. The results revealed the linkages between built environments (on the macro- and micro-scale) and running in Inner London, which can provide practical suggestions for creating running-friendly cities.
The urban environment affects human behavior and health. Most studies on the feelings of street spaces have not considered a specific kind of realistic scene, such as running. To overcome this limitation, we explored the relationship between the urban environment and the pleasure of running. We collected 8260 street view images from 153 running routes in Beijing and invited more than 400 volunteers of different genders and ages to rate their sense of pleasure in street view images of the urban running environment through an online survey. Then, the proportion of visual elements in street images was extracted based on semantic segmentation, and the landscape was divided. Finally, a linear mixed model was used to predict the pleasure scores of different gender and age groups for different landscapes. The results show significant differences in the pleasure scores for different landscapes and age groups. Middle-aged people’s sense of pleasure was lower than that of the young and the elderly. More greenery was associated with a higher pleasure score, while the proportion of urban elements such as buildings was negatively correlated with the pleasure score. The results indicate that running in a natural landscape is pleasurable and beneficial for mental health.
A healthy acoustic environment is an essential component of sustainable cities. Various noise monitoring and simulation techniques have been developed to measure and evaluate urban sounds. However, sensing large areas at a fine resolution remains a great challenge. Based on machine learning, we introduce a new application of street view imagery — estimating large-area high-resolution urban soundscapes, investigating the premise that we can predict and characterize soundscapes without laborious and expensive noise measurements. First, visual features are extracted from street-level imagery using computer vision. Second, fifteen soundscape indicators are identified and a survey is conducted to gauge them solely from images. Finally, a prediction model is constructed to infer the urban soundscape by modeling the non-linear relationship between them. The results are verified with extensive field surveys. Experiments conducted in Singapore and Shenzhen using half a million images affirm that street view imagery enables us to sense large-scale urban soundscapes with low cost but high accuracy and detail, and provides an alternative means to generate soundscape maps. R2 reaches 0.48 by evaluating the predicted results with field data collection. Further novelties in this domain are revealing the contributing visual elements and spatial laws of soundscapes, underscoring the usability of crowdsourced data, and exposing international patterns in perception.
In recent years, the impact of the urban environment on residents’ physical activity (PA) has received extensive attention, but whether this impact has differences in the jogging preferences of residents in different footpath environments and different genders requires further research. Therefore, based on jogging trajectory data, this paper uses the grouping multiple linear regression model to study the different influencing factors of different footpath environments on the jogging of residents of different genders. The results show that (1) jogging activities (JA) were mainly concentrated in the community footpath environment, and its peak was reached at night; (2) the rise and fall of elements in built environments, social environments, and natural environments significantly affected the relative jogging distance of residents; (3) Residential land density (RLD) has a positive impact on the JA of community and green land footpaths and has a negative impact on the JA of urban footpaths. However, arterial road density (ARD) and bus distance density (BDD) have opposite significant effects on the JA of communities and green land footpaths; (4) ARD has the significant opposite effect on the JA for residents of different genders on urban footpaths and community footpaths. Facilities diversity (FD), population density (PD), and bus stop density (BSD) also had significant opposite effects on the JA of residents of different genders on green land footpaths. In general, we put forward a method theory to identify the footpath environment and provide references for improving the layout and construction of different gender residents for different footpath environment elements.
The walkability of urban streets (WoUS) benefits public health and urban livability, but there is still no consensus on the quantitative and comprehensive evaluation of walkability. However, emerging deep-learning, sensor-network, and GIS technologies suggest the possibility of overcoming previous limitations. Therefore, this study explores a new approach for multicriteria evaluation of WoUS, including physical and perceived walkability. After capturing street-view images, traffic-flow data, and environmental-sensor data covering the streets to be analyzed, we automatically calculate physical walkability from four criteria (walkability index, street quality, traffic flow, and physical environment). We first integrate 2-dimensional analysis of walk score, noise, and light using GIS, then perform 3-dimensional composition calculations of greenery, enclosure, and relative walking width using semantic segmentation, vehicle density using multi-object tracking, and indicator weighting distributions using an analytic hierarchy process. Perceived walkability is evaluated by a hierarchy of walking needs model comprising feasibility, accessibility, safety, comfort, and pleasurability. As an empirical study, we conduct an experiment within Anonymous University for physical and perceived walkability evaluations. The results indicate that with perceived WoUS as a benchmark, the automated physical WoUS results reveal the strengths in data integration and processing that are feasible, reliable, and cost-effective. This was an initial attempt at constructing an integrated research methodology for WoUS from various walking-related variables. This framework could support designers as an assistive tool for generating both quantitative and qualitative judgments from various criteria and recommending walkable pedestrians paths to motivate people to walk further, longer, and more often, rather than providing the shortest paths based on a path-finding algorithm.
Buses and metros are two main public transit modes, and these modes are crucial components of sustainable transportation systems. Promoting reciprocal integration between bus and metro systems requires a deep understanding of the effects of multiple factors on transfers among integrated public transportation transfer modes, i.e., metro-to-bus and bus-to-metro. This study aims to reveal the determinants of the transfer ratio between bus and metro systems and quantify the associated impacts. The transfer ratio between buses and metros is identified based on large-scale transaction data from automated fare collection systems. Meanwhile, various influencing factors, including weather, socioeconomic, the intensity of business activities, and built environment factors, are obtained from multivariate sources. A multivariate regression model is used to investigate the associations between the transfer ratio and multiple factors. The results show that the transfer ratio of the two modes significantly increases under high temperature, strong wind, rainfall, and low visibility. The morning peak hours attract a transfer ratio of up to 57.95%, and the average hourly transfer volume is 0.94 to 1.38 times higher at this time than in other periods. The intensity of business activities has the most significant impact on the transfer ratio, which is approximately 1.5 to 15 times that of the other independent variables. Moreover, an adaptative geographically weighted regression is utilized to investigate the spatial divergences of the influences of critical factors on the transfer ratio. The results indicate that the impact of a factor presents spatial heterogeneity and even shows opposite effects (in terms of positive and negative) on the transfer ratio in different urban contexts. For example, among the related socioeconomic variables, the impact of the housing price on the downtown transfer ratio is larger than that in the suburbs. Crowd density positively influences the transfer ratio at most stations in the northern region, whereas it shows negative results in the southern region. These findings provide valuable insights for public transportation management and promote the effective integration of bus and metro systems to provide enhanced transfer services.
Population migration, accompanied by urbanization, has led to an increase in the urban-settled population. However, quantitative studies on the distribution of urban-settled population, especially at fine scale, are limited. This study explored the relationship between characteristics of human perceived environment and the distribution of settled population, and proposed a quantitative method to predict the distribution trend of settled population. Through the semantic segmentation of street view images and accessibility calculation based on traffic isochrone and points-of-interest, we determined human perception factors. The influence of human perception factors was quantified using the geographic detector method, and the settlement intention index (SII) was constructed combining the analytic hierarchy process to predict the distribution trend of settled population. The results indicated the following. (1) Human perception was one of the important factors influencing the distribution of urban-settled population, and the cycling accessibility to traffic facilities was closely related to the distribution of settled population. (2) The accessibility and visibility of green space with low independent influence portrayed a strong enhancement on the interactive effect of other perception factors. (3) The SII mapping of Beijing showed that the SII was reliable. This study analyzes the role of human perception in shaping the environment, and provides reference for population-related urban planning problems.
Machine learning and artificial intelligence (ML/AI), previously considered black box approaches, are becoming more interpretable, as a result of the recent advances in eXplainable AI (XAI). In particular, local interpretation methods such as SHAP (SHapley Additive exPlanations) offer the opportunity to flexibly model, interpret and visualise complex geographical phenomena and processes. In this paper, we use SHAP to interpret XGBoost (eXtreme Gradient Boosting) as an example to demonstrate how to extract spatial effects from machine learning models. We conduct simulation experiments that compare SHAP-explained XGBoost to Spatial Lag Model (SLM) and Multi-scale Geographically Weighted Regression (MGWR) at the parameter level. Results show that XGBoost estimates similar spatial effects as those in SLM and MGWR models. An empirical example of Chicago ride-hailing modelling is presented to demonstrate the utility of SHAP with real datasets. Examples and evidence in this paper suggest that locally interpreted machine learning models are good alternatives to spatial statistical models and perform better when complex spatial and non-spatial effects (e.g. non-linearities, interactions) co-exist and are unknown.
This paper considers the nature of social surveillance through the physical activity tracking app MapMyRun and examines how this was experienced during the COVID-19 pandemic during the UK and USA summer 2020 lockdowns. In contributing to debates in digital geographies around the entanglements of the fleshy and digital body, the paper responds to calls for research to recognise the increasing sociality of self-tracking (Coutre, 2021), specifically considering how, during the COVID-19 pandemic, these apps offered a form of connection during a time of isolation. Using data from email and video interviews, I argue that whilst a Foucauldian account of surveillance can be used as a point of departure, it is limited in accounting for the social aspects of self-tracking. I therefore propose that applying Robinson’s (2000) concept of ‘noisy surveillance’ to self-tracking is useful for understanding the messiness of surveillance in terms of the complications and noisiness involved in interactions in digital spaces, as well as the opportunities for performance management online particularly during lockdown.
The random forest model has been frequently used for house sale price prediction because researchers have shown that it can create better results with smaller prediction errors than conventional statistical models. However, house sale prices tend to be spatially autocorrelated due to shared neighborhood environments and similar physical characteristics, which have been rarely considered and accounted for in the random forest model. A failure to address spatial autocorrelation may result in over‐ or underpredicted house sale prices, especially when these sale prices are highly spatially dependent. This research proposes an extended random forest approach, introducing additional proxy variables that furnish geographic proximity measures of house locations in order to capture spatial autocorrelation that is unexplained by covariates. An application of the extended random forest approach for house sale prices in Fairfax County, Virginia, confirms the effectiveness of the proxy variables with an improvement of house sale price prediction accuracy. The introduction of the proxy variables can correct over‐ and under‐predicted house sale prices that show an underlying spatial structure.
To reexamine the relationship between leisure-time physical activity (LTPA) and the built environment (BE), this paper takes advantage of the massive amount of data collected by an accelerometer and GPS-based fitness mobile app. Massive LTPA data from more than 3 million users were recorded by Codoon in 500m by 500m grid cells and aggregated to 742 natural cities in mainland China. Six BE indicators were quantified using GIS at the city scale. Robust regression analysis was used to estimate the correlation between LTPA and BE. Five of six BE indicators—connectivity, road density, land use mix, points of interest density, and density of parks and squares—were significantly, positively, independently, and linearly related to LTPA in the regression analysis. The study obtains findings that are consistent with the previous literature but also provides novel insights into the important role of POI density in encouraging LTPA, as well as how the relationship between LTPA and BE varies by time of day. The study also sheds light on the embrace of new technology and new data in public health and urban studies.
The health and welfare of older adults have raised increasing attention due to global aging. Cycling is a physical activity and mode of transportation to enhance the mobility and quality of life among older adults. Nevertheless, the planning strategies to promote cycling among older adults are underutilized. Therefore, this paper describes the nonlinear associations of the built environment with cycling frequency among older adults. The data were collected from the Zhongshan Household Travel Survey (ZHTS) in 2012. The modeling approach was the eXtreme Gradient Boosting (XGBoost) model. The findings demonstrated that nonlinear relationships exist among all the selected built environment attributes. Within specific intervals, the population density, the land-use mixture, the distance from home to the nearest bus stop, and the distance from home to CBD are positively correlated to the cycling among older adults. Additionally, an inverse “U”-shaped relationship appears in the percentage of green space land use among all land uses. Moreover, the intersection density is inversely related to the cycling frequency among older adults. These findings provide nuanced and appropriate guidance for establishing age-friendly neighborhoods.
Urban parks are essential components of urban ecosystems, providing recreation and relaxation places to residents. Measuring the spatial accessibility to urban parks serves as an initial step in urban planning and developing urban development strategies to improve social and environmental justice. This study aims to evaluate measures of spatial accessibility to urban parks by comparing three geographic information systems (GIS)-based approaches, accounting for network complexity, transport modes, distance thresholds, and destination choices. Taking Ipswich City (Australia) and Enschede (the Netherlands) as two testbeds, we examine the spatial patterns of a total of 21 accessibility measures in the two cities and conduct a correlation and principal component analysis to unravel the interrelationship between these measures. The results suggest that among all measures under the three approaches, the selection of distance thresholds and transport modes matter more to accessibility measures than the destination choices. Furthermore, when distance threshold and transport mode are held constant, the network-based and entrance-based methods provide more realistic accessibility measures than other methods. We also discuss the generality of the entrance-based method we propose and suggest ways to choose the most appropriate accessibility measure for use in different contexts.
Objectives
To conduct a longitudinal ecological analysis of the distance to and participation in free weekly outdoor physical activity events (parkrun) in England from 2010 to 2019, and related socioeconomic and ethnic inequalities, to inform policies to support participation in physically active community events.
Methods
We calculate distance to the nearest parkrun event for each English Lower Layer Super Output Area (LSOA) each month from January 2010 to December 2019. We then report the trends in distance to and participation in parkrun by Index of Multiple Deprivation quintile. We also report trends in the Relative Index of Inequality (RII) by deprivation for participation and distance to nearest event. We go on to investigate trends in LSOA level determinants (e.g. deprivation and ethnic density) of parkrun participation between 2010 and 2019, using multivariable Poisson regression models.
Results
Mean distance to the nearest parkrun event decreased from 34.1 km in 2010, to 4.6 km in 2019. Throughout the period, parkrun events tended to be situated closer to deprived areas compared to less deprived areas. Participation rates increased superlinearly (greater than linear increase) from 2010 to 2013 before slowing to linear growth. Participation over the period exhibits a clear socioeconomic gradient, with people from deprived areas having consistently lower participation rates over the period. parkrun participation rates became more equal between 2010 and 2013 (RII improved from 189 to 39), before stabilising at an RII between 32.9 and 39.6 from 2014 to 2019. The results of the Poisson regression model validate this finding; the coefficients on IMD score initially increased from −0.050 in 2010 to −0.038 in 2013, and then remained relatively stable to 2019 (−0.035).
Conclusions
Over the past 10 years, geodesic distance to the nearest parkrun decreased from a mean of 34 km to 5 km. In 2010, there was equality between the least and most deprived areas but by 2017 the distance of the most deprived areas was 29% that of the least deprived. Participation was shown to have increased over the past 10 years which can be split into two distinct phases: from 2010 to 2013 participation increased super-linearly and inequality in participation fell dramatically; from 2013 to 2019 participation increased linearly, and inequality in participation remained stable. Despite parkrun's ambitions of creating inclusive events and engaging with deprived communities, the socioeconomic gradient in participation rates remained high and stable since 2013. Gaining a better understanding of the reasons why parkrun grew so quickly may be useful for other physical activity movements, while further analysis of the relatively lower participation rates in areas with higher socioeconomic deprivation is important for developing initiatives to encourage physical activity in these communities.
Although previous studies have assessed the relationships between visual space indicators and urban residents' psychological perceptions, systematic research on the relationship between the urban visual space and residents' psychological perceptions is still rare. The purpose of this study is to explore the correlation between the urban visual space and residents' psychological perceptions, analyze the influence of visual space indicators on residents' subjective perceptions, and focus on whether blue-green space is influenced by walkability, enclosure, openness, imageability, and traffic flow in the mechanism of its effect on urban residents' psychology. In this study, the Fully Convolutional Network (FCN-8s) is used to segment the street view images of Wuhan. An urban perception data set is obtained by coupling the human-machine adversarial framework and the Random Forest algorithm. Based on the street view data, we design seven kinds of visual space indicators. We conduct multiple linear regression analysis on urban residents' perceptions and the mediating effect analysis of blue-green space on residents' psychology using the above five visual space indicators as mediator. The results show that there is a strong relationship between the urban visual space indicators and residents' psychological perceptions, the visual space indicators such as greenness and enclosure have a strong influence on residents' psychological perceptions and significant mediating effects of some visual space indicators occur in the process of blue-green space influencing residents' psychology. This study comprehensively analyzes the relationship and mechanism between visual space indicators and residents’ psychological perception, and makes forward-looking work to further explore the complex mechanisms between visual space and residents’ psychology.
There is limited evidence on the gender differences and location-specific built-environment factors associated with bicycling in Latin American cities. This study aimed to assess commuting in Bogotá by (1) analyzing the gender-specific trend of the standardized number of bicycle commuters during 2005–2017; and (2) assessing the socio-demographic, community, built-environment and natural factors associated with bicycle commuting stratified by gender. This secondary-data analysis included data from the Household Travel Surveys and Multipurpose Surveys to calculate the number of bicycle commuters per habitant from 2005 to 2017 by gender. We assessed the socio-demographic and built-environment factors fitting generalized additive models stratified by gender using the 2015 Household Travel Survey. Although both women and men increased the standardized number of bicycle commuters, male commuters show a steeper trend than women, evidencing the widening gender gap in bicycle commuting over time. Bicycle commuting was negatively associated with household motor vehicle ownership, steeper terrain slope, longer commute distance, and scarce low-stress roads at trip origin and route. Among women, the availability of bike paths at the trip destination was positively associated with bicycling, while age and being a student were negatively associated with bicycling. Among men, living in areas with the lowest socio-economic status was positively associated with bicycling, while having a driver's license and living close to bus rapid transit stations were negatively associated with bicycling. In conclusion, bicycle and transport infrastructure play different roles in commuting by bicycle by gender and trip stages (origin – route – destination).
Recreational road running is growing in popularity and has been linked to numerous mental and physical health benefits. However, we know little about what environmental preferences or concerns runners have regarding participation in the sport, and whether differences exist across age and gender. We conducted a cross-sectional survey on recreational road runners to investigate the type of built and natural environments road runners prefer, as well as the safety and health concerns that may affect runners’ choice of environment. Responses were analyzed by age and gender. A total of 1228 road runners responded to the survey; 59.6% of respondents were women and 32.1% of respondents were men. Most respondents preferred to run on asphalt or sidewalk surfaces, and preferred well-lit, tree-lined routes. Major concerns for both men and women include animals and dangerous road conditions. Men and women differed significantly in their responses to the importance of running around others and their primary concerns while running. Results of this study serve to deepen our understanding of recreational road runners’ environmental preferences and concerns, providing valuable information for public health officials and city planners alike. This information must be considered if we are to continue to encourage uptake of running as a sport and reap its health effects.
Urban vibrancy can facilitate human activities and social interactions, attract capital and talent, enhance competitiveness and creativity, maintain resilience, and finally achieve a sustainable urban development. Theoretically, transit-oriented development (TOD) is beneficial for vibrancy. In practice, TOD implementation does not always lead to vibrant community life. To foster vibrancy around stations, we need to understand relationships between TOD-ness (i.e., the degree to which the current (physical) conditions of station areas meet the standards of TOD) and vibrancy. Empirical studies on this topic are scarce, despite numerous studies on built environment effects on vibrancy. Moreover, these studies are likely to overestimate/underestimate the effects as most of them neglect the pervasive nonlinearity and synergism. This research contributes to the understanding of nonlinear and synergistic effects of TOD on vibrancy by constructing gradient boosting decision tree model using multi-source data from 166 metro station areas in Shenzhen, China. Local explanations for the model indicate that (1) Sufficient bus services, horizontal built-up coverage, and mixed-use buildings are dominant contributors to vibrancy around metro stations; (2) TOD have nonlinear influences on vibrancy; (3) Evident synergistic effects exist among/within TOD dimensions. Practical implications of the findings, such as targeted policies with nuanced planning/design criteria, are further discussed.
Population aging is a conspicuous demographic trend shaping the world profoundly. Walking is a critical travel mode and physical activity for older adults. As such, there is a need to determine the factors influencing the walking behavior of older people, especially in the era of population aging. Streetscape greenery is an easily perceived built-environment attribute and can promote walking behavior, but it has received insufficient attention. More importantly, the non-linear effects of streetscape greenery on the walking behavior of older adults have not been examined. We therefore use readily available Google Street View imagery and a fully convolutional neural network to evaluate human-scale, eye-level streetscape greenery. Using data from the Hong Kong Travel Characteristic Survey, we adopt a machine learning technique, namely random forest modeling, to scrutinize the non-linear effects of streetscape greenery on the walking propensity of older adults. The results show that streetscape greenery has a positive effect on walking propensity within a certain range, but outside the range, the positive association no longer holds. The non-linear associations of other built-environment attributes are also examined.
Running is one of the most popular forms of physical activity. To date, no literature explores association between the features of the built environment and running. A limited amount of literature uses walkability as a proxy for runnability, thereby misjudging the differing goals of walking and running: leisure, sport or commute. We create a rough runnability index using an affordance-based framework as a conceptual and methodological attempt to quantify features of the built environment that facilitate or hinder running as a form of leisure or sport activity. Three indices are created in the City of Surrey, British Columbia, Canada using pixelated edges. We find that areas in the downtown core and near high traffic routes show low safety and general runnability, whereas areas near parks and in low traffic, residential areas show higher safety for runners. Representing runnability using pixelated edges allows for sub-block level analysis of runnability as experienced by runners.
Human-perceived landscape and amenities are not equivalent to the objective physical world. Traditional big data research that only uses objectively measured environments based on point-of-interest data has difficulty capturing the effect of human environmental perceptions on walking behavior. We proposed a conceptual framework to examine the effect of human perception on walking behavior by integrating travel trajectories from the “big data” derived from mobile phones and “small data” obtained from a questionnaire survey of residents living in Chengdu, China. Results show that the human subjective perceived environment is inconsistent with the objectively measured built environment. The walking behavior of individuals is completely regulated by their personal perception of the surrounding environment. Moreover, enhancing the environmental design of pathways leading to retail shops, recreational areas, and shopping centers could significantly enhance perceived accessibility and foster a more pedestrian-friendly neighborhood.
Green spaces may have beneficial impacts on children's cognition. However, few studies explored the exposure to green spaces beyond residential areas, and their availability, accessibility and uses at the same time. The aim of the present study was to describe patterns of availability, accessibility, and uses of green spaces among primary school children and to explore how these exposure dimensions are associated with cognitive development. Exposures to green space near home, school, commuting route, and other daily activity locations were assessed for 1607 children aged 6-11 years from six birth cohorts across Europe, and included variables related to: availability (NDVI buffers: 100, 300, 500 m), accessibility (proximity to a major green space: linear distance; within 300 m), and use (play time in green spaces: hours/year), and the number of visits to green spaces (times/previous week). Cognition measured as fluid intelligence, inattention, and working memory was assessed by computerized tests. We performed multiple linear regression analyses on pooled and imputed data adjusted for individual and area-level confounders. Availability, accessibility, and uses of green spaces showed a social gradient that was unfavorable in more vulnerable socioeconomic groups. NDVI was associated with more playing time in green spaces, but proximity to a major green space was not. Associations between green space exposures and cognitive function outcomes were not statistically significant in our overall study population. Stratification by socioeconomic variables showed that living within 300 m of a major green space was associated with improved working memory only in children in less deprived residential areas (β = 0.30, CI: 0.09,0.51), and that more time playing in green spaces was associated with better working memory only in children of highly educated mothers (β per IQR increase in hour/year = 0.10; 95% CI: 0.01, 0.19). However, studying within 300 m of a major green space increased inattention scores in children in more deprived areas (β = 15.45, 95% CI: 3.50, 27.40).
Outdoor jogging is a beneficial and replicable physical activity (PA). The nonlinear effects of the built environment (BE) on jogging fitness received little attention, compared with the widely-concerned linear effects of BE on walking and cycling PA. We explore nonlinear effects at trip and origins/destinations (OD) levels in the case of Chengdu using Random Forest, based on large-scale jogging trajectory data recorded by a fitness app. The major findings include: (1) BE factors exert diverse nonlinear effects on jogging at trip and OD levels. (2) Quantity and accessibility of facilities contribute largely to model predictive power. (3) Nonlinear effects are symmetrical for O/D of jogging, unlike long-distance travel. Distance to park, distance to track, and population density show Ushaped effects on OD volume. (4) Effective ranges and thresholds in nonlinear effects vary across trip/OD levels. The findings call for environmental intervention to promote PA.
With the surging usage of e-scooters worldwide, there is a growing interest in understanding different aspects of e-scooters trips and their impact on urban mobility. Further, the emergence of this new mode of transportation has led to questions regarding the spatial accessibility of e-scooters and understanding how the built environment and urbanism characteristics affect riders' abilities to reach certain destinations. In this study, initially, a data-driven approach was proposed to construct the service areas for dockless e-scooter using origin-destination trip data. Service areas are defined as spatial areas that riders are regularly able to reach via an e-scooter. E-scooter service areas were constructed for traffic analysis zones in Louisville, KY, using agglomerative hierarchical clustering and convex hull algorithms. Then, the relationship between various built environments and urbanism characteristics and the e-scooter service areas was examined using principal component analysis and random forest regression. The results showed that percent of residential properties, length of the block, Walk Score®, Transit Score ®, and Dining and Drinking Score contributed most to the size of the e-scooter service area. The findings of this research offer a transferable method to estimate e-scooter service areas to quantify access to goods and services. Further, the study discusses how the built environment and urbanism characteristics might affect the size of the service areas.
Few studies have investigated the influences of the multilevel built environment (BE) factors on jogging, compared with walking or cycling. This study explored neighborhood and street-level BE factors' associations with fitness jogging via multi-source urban data. Empirical analysis of Chengdu, China showed the following: (1) BE factors, including sky view factor, bus stop density, presence of waterscapes, and geographic location significantly, impact jogging activity. (2) The significance and effect of BE factors vary across time. Jogging activities are more sensitive to BE on weekends than on weekdays. (3) Jogging trips are more closely related to BE factors in urban areas than in the suburbs. Jogging activities are mainly affected by the artificial environment in urban areas and the natural environment in suburban areas. Moreover, the effects of multiple BE factors become obvious as trip distance increases. These findings call for urban planning and infrastructure provision strategies to promote jogging activities.
The built environment is found to relate to running behaviors. However, the impacts of the street environment on running were less addressed due to the lack of running data in large geospatial urban regions, while the potential of semi-open data sources like Strava Heatmap for running studies is rarely verified. Moreover, how objective features and subjective perceptions of the street environment are related to running is still largely unknown. We hypothesize that the eye-level subjective and objective streetscapes may complement the macro-scale built environment factors to better inform running amount prediction. Therefore, we evaluated the associations between running and street attributes by applying multi-sourced data, street view imagery (SVI) and artificial intelligence (AI) technologies, taking Boston as an example. We found that, first, the street environment is significantly correlated with running. Accounting for the spatial effects, the collective strength of street attributes was almost the same as the counterpart of the built environment, validating the value of including subjective and objective streetscapes measures in running studies. Second, street factors can complement built environment factors, indicating the necessity of using both macro-scale and eye-level environmental features to interpret running. Third, in addition to higher accessibility and more public transportation, the safer, wider and relatively open streets with more natural views, street lights, amenities and furniture, could promote running, while the enclosed environment, dense and overwhelming buildings, and excessive interruptions on streets might hinder running. Our study provides an important example of using semi-open running data and integrating multi-sourced data and AI to bring new insights into running and urban environmental studies. The findings could provide instructive suggestions for the establishment of a running-friendly urban environment and ultimately help to improve public health.
The COVID-19 pandemic challenged emergency management in cities worldwide. Many municipalities adopted restrictive, one-size-fits-all spatial regulations such as lockdowns without fully considering the inhabitants' daily activities and local economies. The existing epidemic regulations' unintended detrimental effects on socioeconomic sustainability necessitate a transition from the "lockdown" approach to more precise disease prevention. A spatially and temporally precise approach that balances epidemic prevention with the demands of daily activities and local economies is needed. Thus, the aim of this study was to propose a framework and key procedures for determining precise prevention regulations from the perspectives of the 15-minute city concept and spatiotemporal planning. Alternative regulations of lockdowns were determined by delineating 15-minute neighborhoods, identifying and reconfiguring facility supplies and activity demands in both normal and epidemic conditions, and performing cost-benefit analyses. Highly adaptable, spatially- and temporally-precise regulations can match the needs of different types of facilities. We demonstrated the process for determining precise prevention regulations in the case of the Jiulong 15-minute neighborhood in Beijing. Precise prevention regulations-which meet essential activity demands and are adaptable for different facility types, times, and neighborhoods-have implications for long-term urban planning and emergency management.
Buildings are fundamental components of urban areas and they play a vital role in supporting human activities in daily life. Understanding the actual building functions is essential for many urban applications, such as city management, urban planning, and optimization of transportation systems. Existing studies for inferring building functions are mainly based on a building's own features, and ignore its “geographic context” (e.g., the influences of nearby buildings). This paper introduces a novel geo-aware neural network to infer the functions of individual buildings. To this end, the proposed model integrates information about the built environment and human activity of a target building and its “geographic context”. The model further includes a geo-aware position embedding generator and transformer encoders to better capture the complex relationships between buildings. The evaluation results demonstrate that the proposed model outperforms all baselines and achieves a classification accuracy of 90.8%. Meanwhile, the proposed model works well even with a small amount of training dataset and has a good transferability to another urban area. In summary, the proposed model is an effective and reliable approach for inferring the functions of individual buildings and has high potential for city management and sustainable urban planning.
This study aims to examine the degree to which the results of social equity analysis on public transit reliability are sensitive to the choice of spatial unit of analysis (i.e., modifiable areal unit problem (MAUP)). Using the city of Winnipeg as an example, we investigate the social equity of bus on-time performance (OTP), and pass-up distribution at multiple levels (e.g., stop, route, neighborhood) and compare the results. Neighborhoods are classified as minority vs. non-minority population dominant neighborhoods and transit routes and stops are classified into minority vs. non-minority population serving ones according to the socioeconomics of residents. We conduct an equity assessment by calculating 1) the number of pass-ups, 2) the number of bus deviations (e.g., early arrivals, delays) from the predetermined schedule, and 3) the average deviations time (in seconds) that occurred on minority vs. non-minority serving routes/stops as well as in minority vs. non-minority dominant neighborhoods. We also consider different day types (e.g., weekdays, weekends) and disability status (e.g., wheelchair, regular passengers) in the equity assessment. While the route-level results depict that transit service reliability is equitable, results of the neighborhood and stop-level analysis reveal inequities in the distribution of the OTP and pass-ups in the city. Our findings demonstrate that different levels of spatial aggregation can significantly change the results of social equity analysis of public transit reliability, thereby leading to incomplete conclusions that inadequately capture the inequity due to the existence of MAUP. The results of this paper provide insights to transit authorities, planners, and policymakers for diagnosing the social equity landscape of transit reliability in a more robust manner.
Understanding the influence of built environment on running behaviour is a significant step towards developing related landscape strategies. This paper adopted a volunteered geographic information (VGI) approach to measure urban runnability by quantifying environmental features that encourage or hinder running activities. The GPS-based routes collected from Strava were used to compute the running intensity of street segments in Helsinki, Finland. We applied multilevel regression models to assess the spatial-varied impacts of street environment on running intensity, so as to elicit runner's preferences and investigate their associations with sociodemographic characteristics at higher hierarchical level (neighbourhood). The results showed street greenery assessed by Green View Index (GVI) represented the greenness exposure to runners better than top-down greenness assessed by Normalized Difference Vegetation Index (NDVI), and thus can be considered as a more reliable predictor for running behaviours. Blue space density was the predominant factor and associated with running intensity positively. Running intensity negatively correlated with urban density, connectivity and traffic accidents, and positively correlated with traffic noise and air pollution. Population density and income level were positively associated with running intensity. These results provide profound insights and boost our capacity to design more attractive and sustainable streets for physical activity.
Accurate crash frequency prediction is critical for proactive safety management. The emerging connected vehicles technology provides us with a wealth of vehicular motion data, which enables a better connection between crash frequency and driving behaviors. However, appropriately dealing with the spatial dependence of crash frequency and multitudinous driving features has been a difficult but critical challenge in the prediction process. To this end, this study aims to investigate a new Artificial Intelligence technique called Geographical Random Forest (GRF) that can address spatial heterogeneity and retain all potential predictors. By harnessing more than 2.2 billion high-resolution connected vehicle Basic Safety Message (BSM) observations from the Safety Pilot Model Deployment in Ann Arbor, MI, 30 indicators of driving volatility are extracted, including speed, longitudinal and lateral acceleration, and yaw rate. The developed GRF was implemented to predict rear-end crash frequency at intersections. The results show that: 1) rear-end crashes are more likely to happen at intersections connecting minor roads compared to major roads; 2) a higher number of hard acceleration and deceleration events beyond two standard deviations in the longitudinal direction is a leading indicator of rear-end crashes; 3) the optimal GRF significantly outperforms Global Random Forest, with a 9% lower test error and a substantially better fit; and 4) geographical visualization of variable importance highlights the presence of spatial non-stationarity. The proposed framework can proactively identify at-risk intersections and alert drivers when leading indicators of driving volatility tend to worsen.
As carsharing is promoted in many cities, studying the factors related to station-level demand under multi-mode operating carsharing systems will help better develop carsharing programs. In this paper, the number of transactions is calculated to represent the carsharing demand based on the data from a one-way and round-trip hybrid operation carsharing system in Beijing. The generalized additive model is then conducted with carsharing demand as the dependent variable and five groups of independent variables: station attributes, public transit, built environment, weather, and consumption features. The main results are: (1) the one-way station has a substitution effect on the round-trip station; (2) the one-way and round-trip stations not only have considerable demand in places well served by transit but also can play different and complementary roles in areas underserved by transit; (3) the percentage of residential POIs, the POI-mixed entropy, and the number of cars per household are the most statistically significant factors; (4) weather conditions such as precipitation, high temperatures, and poor air quality play a positive role in carsharing usage; (5) personalized incentives help attract more usage, while the positive effect is minor; (6) both population density and average consumption are highly nonlinear to carsharing usage. Population densities of around 10,000 people/km² have the largest positive effect on one-way station usage and around 6,500 people/km² have the largest negative effect on round-trip station usage. When the average consumption exceeds 100 RMB, the usage of one-way stations decreases more rapidly compared to round-trip stations. This study can help the operators understand and refine their business strategies, and can provide insights into determining the location and mode of carsharing stations.
Understanding spatio-temporal patterns of tourist movement behaviors is vital for Destination Management Departments (DMDs) on destination planning, marketing, and resource management. This study uses open GPS-trajectory data to analyze the microscopic spatio-temporal patterns of tourists' movement behaviors in Mount Huashan in China. Two major measures, Markov chains and cluster analysis, are used to cluster tourists into groups to show their spatio-temporal movement behaviors within the study site. The Markov chain analysis unveiled three kinds of spatial patterns on microscopic tourists’ movement: “proximity transfer”, “span transfer” and “hub transfer”, while the cluster analysis further demonstrated three kinds of microscopic spatio-temporal patterns: “day-climbing”, “night-climbing”, and “full-day sightseeing”. These results provide vital implications for tourism management of the mountainous scenic areas in general and in Mount Huashan in particular.
Dockless bike-sharing (DBS), with its advantages of flexibility, environmental friendliness, and good physical fitness, is regarded as an effective approach to relieve urban traffic issues. As such, various studies have been conducted to explore the impact of the built environment on DBS usage. However, few studies have investigated the nonlinear effects at the street level. Especially, existing studies provided no insights on the influence of traffic conditions on DBS trips. By taking the central districts of Shenzhen, China as the case, this study proposed a new method for extracting the street-level DBS trips by considering the bicycle trajectory, and charactered the street-level associated factors by using multi-source big data. Moreover, we further employed an advanced supervised machine learning approach (i.e., gradient boosting decision trees, GBDT) and the machine learning interpretation methods (i.e., relative importance and partial dependence plots) to examine the nonlinear effects of traffic conditions and built environment factors on DBS trips at the street level. The results indicated a significant positive association between vehicle flow and DBS trips. High driving and riding traffic are more likely to occur simultaneously on urban primary arterials. Furthermore, vehicle PM2.5 emissions are positively correlated with DBS trips during peak commuting hours, but negatively correlated during non-peak hours. The effective range and threshold effects of the other factors on cycling were further identified. These findings can inform scientific decisions on the improvement of non-motorized transportation systems and cycling-friendly environments in metropolitan areas.
Characterizing activity pattern such as behavior preferences or habits is crucial in many fields. However, existing studies mainly focus on the spatial-temporal dimensions of raw trajectory, but ignore the context information in multi-aspect trajectory that affects behavior significantly. In this paper, we present a data-driven framework to characterize outdoor jogging activity patterns with massive multi-aspect trajectory data. In our framework, a novel multi-aspect trajectory Latent Dirichlet Allocation (MAT-LDA) model is presented to discover latent activity patterns from multi-aspect trajectories. Specifically, the model inherits from LDA, but extends its topics and words to mine the combined patterns in multi-aspects. Then, clustering analysis is performed to find and characterize the jogger groups with similar preference patterns. Experiments with real jogging GPS tracks recorded by 16,643 users' fitness app show that the MAT-LDA model can efficiently discover the latent activity patterns and quantify the correlations and interdependencies between patterns of multi-aspect attributes. Moreover, many interpretable preferences are discovered at individual level, and jogger groups (e.g., mini groups, jog hobbyist) with common context-aware preferences are revealed to understand fitness jogging. Our method can capture activity behavior preferences of multiple aspects from multi-aspect trajectory data, and our work can enrich outdoor fitness application with interpretable preference patterns.
A growing number of studies show that the uneven spatial distribution of COVID-19 deaths is related to demographic and socioeconomic disparities across space. However, most studies fail to assess the relative importance of each factor to COVID-19 death rate and, more importantly, how this importance varies spatially. Here, we assess the variables that are more important locally using Geographical Random Forest (GRF), a local non-linear regression method. Through GRF, we estimated the non-linear relationships between the COVID-19 death rate and 29 socioeconomic and health-related factors during the first year of the pandemic in the USA (county level). GRF outputs are compared to global (Random Forest and OLS) and local (Geographically Weighted Regression) models. Results show that GRF outperforms all models and that the importance of variables highly varies by location. For example, lack of health insurance is the most important factor in one-third (34.86%) of the US counties. Most of these counties are (concentrated mainly in the Midwest region and South region). On the other hand, no leisure-time physical activity is the most important primary factor for 19.86% of the US counties. These counties are found in California, Oregon, Washington, and parts of the South region. Understanding the location-based characteristics and spatial patterns of socioeconomic and health factors linked to COVID-19 deaths is paramount for policy designing and decision making. In this way, interventions can be designed and implemented based on the most important factors locally, avoiding thus general guidelines addressed for the entire nation.
Studies on the linkages between nature exposure and physical activities often focus simply on the immediate vicinity of home locations, but path-based exercises, such as running and cycling, are continuous activities and cover a broad spatial extent. Thus, the traditional home buffer approach fails to acknowledge the settings where road running actually occurs. This study employed an activity path-based measure approach using public participation GIS (PPGIS) to investigate the associations between running satisfaction and nature exposure. The mapped routes (N=545) that included an assessment of satisfaction level were collected from 249 runners resided in the Helsinki Metropolitan Area, Finland. Logistic regression analyses revealed a positive association between running satisfaction and nature exposure, including eye-level greenness, top-down greenness and blue space density. Top-down greenness was assessed by Normalized Difference Vegetation Index (NDVI) and the eye-level greenness by Green View Index (GVI), the latter one of which uses a deep learning algorithm. Running environment was more satisfying in those routes with more public transport nodes. Other traffic-related factors breaking the momentum of runners such as traffic light density were inversely related to running satisfaction. Demographic characteristics such as education background also played a significant role in the perceived satisfaction with running routes. The positive impacts of nature exposure on running satisfaction further verify the linkages between landscape and public health.
Understanding locally heterogeneous physical contexts in built environment is of great importance in developing preemptive countermeasures to mitigate pedestrian fatality risks. In this study, we aim to investigate the non-linear relationship between physical factors and pedestrian fatality at a location-specific level using a machine learning approach. The state-of-art machine learning algorithm, eXtreme Gradient Boosting (XGBoost), is employed for a binary classification problem, in which nationwide locations where fatal pedestrian accidents occurred for the years from 2012 to 2019 in Korea serve as positive samples (np = 13,366). For negative samples, locations with no pedestrian accidents are selected randomly to the size that is 10 times larger (nn = 133,660) than positive samples. Fifteen features under the categories of road conditions, road facilities, road networks, and land uses are assigned to both the positive and negative sample locations using Geographic Information System (GIS). A method is proposed to avoid the class imbalance problem, and a final unbiased model is utilized to predict fatal pedestrian risks at the negative sample locations. In addition, Shapley Additive Explanations (SHAP) is introduced to provide a robust interpretation of the XGBoos prediction results. It is shown that 21.6% of the negative sample locations have a probability of fatal pedestrian accidents greater than 0.5 (or 78.4% accuracy). Generally, a road segment that lies in many of the shortest routes in a dense residential area with many lively activities from aligned buildings is a potential spot for fatal pedestrian accidents. However, based on the SHAP interpretation, the relationships between the features and pedestrian fatality are found nonlinear and locally heterogeneous. We discuss the implications of this result has for drafting policy recommendations to reduce pedestrian fatalities.
Existing studies have seldom used large-scale trajectory data to analyze jogging activities in urban parks. Most of them have relied on traditional questionnaires and on-site interviews. Therefore, this study aims to uncover the characteristics and the potential influencing factors of jogging activities based on trajectory data recorded by a mobile app, using the case of Chongqing city. The results show that urban parks with high jogging flow are mainly distributed within the inner ring road of Chongqing, whereas urban parks with low jogging flow are newly built outside the inner ring road. The volume of jogging flow in urban parks in the spring and summer is higher than that of autumn and winter, and the volume on weekends is higher than that of weekdays. The peaks of jogging in urban parks vary across space and over time, leading to different spatiotemporal patterns. Urban parks along the subcenters, riversides, and airport corridors have morning (6–7 a.m.) and evening peaks (7–8 p.m.). Urban parks in the newly urbanized areas and industrial zones have evening peaks. The regression models show that walking loops and waterscapes have positive effects on jogging flow. The landscape shape index of urban parks and the distance to the city center negatively affect the jogging flow. Finally, the study indicates the possibility of using large-scale trajectory data to analyze jogging activities, which is helpful for urban park planners and managers to improve the frequency of jogging activity.
Understanding the influencing mechanism of the urban streetscape on crime is fairly important to crime prevention and urban management. Recently, the development of deep learning technology and big data of street view images, makes it possible to quantitatively explore the relationship between streetscape and crime. This study computed eight streetscape indexes of the street built environment using Google Street View images firstly. Then, the association between the eight indexes and recorded crime events was revealed with a poisson regression model and a geographically weighted poisson regression model. An experiment was conducted in downtown and uptown Manhattan, New York. Global regression results show that the influences of Motorization Index on crimes are significant and positive, while the effects of the Light View Index and Green View Index on crimes depend heavily on the socioeconomic factors. From a local perspective, the Pedestrian Space Index, Green View Index, Light View Index and Motorization Index have a significant spatial influence on crimes, while the same visual streetscape factors have different effects on different streets due to the combination differences of socioeconomic, cultural and streetscape elements. The key streetscape elements of a given street that affect a specific criminal activity can be identified according to the strength of the association. The results provide both theoretical and practical implications for crime theories and crime prevention efforts.
The relationship between the built environment and the physical activity (PA) of children has been explored; however, the results remain controversial. Moreover, the studies on the non-linear influences of the built environment on PA are very limited. Using the 2019 questionnaire survey data from Xi'an, China, this study applies a gradient boosting decision tree (GBDT) model to explore the non-linear impact and the collective contribution of the urban built environment on the PA of children. The results showed that compared with demographics, the characteristics of the built environment are more important in predicting PA of children (58.8% vs 41.2%). Among the variables of the built environment that affect the PA of children, land-use mix around the school, the distance from home to school, and the distance to the nearest park are the top three important factors. A non-linear relationship exists between the specific built environment characteristics and the PA of children.
It is important to quantify human heat exposure in order to evaluate and mitigate the negative impacts of heat on human well-being in the context of global warming. This study proposed a human-centric framework to examine human personal heat exposure based on anonymous GPS trajectories data mining and urban microclimate modeling. The mean radiant temperature (Tmrt) that represents human body's energy balance was used to indicate human heat exposure. The meteorological data and high-resolution 3D urban model generated from multispectral remotely sensed images and LiDAR data were used as inputs in urban microclimate modeling to map the spatio-temporal distribution of the Tmrt in the Boston metropolitan area. The anonymous human GPS trajectory data collected from fitness Apps was used to map the spatio-temporal distribution of human outdoor activities. By overlaying the anonymous GPS trajectories on the generated spatio-temporal maps of Tmrt, this study further examined the heat exposure of runners in different age-gender groups in the Boston area. Results show that there is no significant difference in terms of heat exposure for female and male runners. The female runners in the age of 45–54 are exposed to more heat than female runners of 18–24 and 25–34, while there is no significant difference among male runners. This study proposed a novel method to estimate human heat exposure, which would shed new light on mitigating the negative impacts of heat on human health.
Studies often rely on home locations to access built environment (BE) influences on physical activity (PA). We use GPS and accelerometer data collected for 288 individuals over a two-week period to examine eight GPS-derived BE characteristics and moderate-to-vigorous PA (MVPA) and light-to-moderate-vigorous PA (LMVPA). NDVI, parks, blue space, pedestrian-orientated intersections, and population density were associated with increased odds of LMVPA and MVPA, while traffic air pollution and noise were associated with decreased odds of LMVPA and MVPA. Associations varied by population density and when accounting for multiple BE measures. These findings provide further information on where individuals choose to be physically active.
Studies have indicated that a sense of community may be shaped by the built environment and has potential mental wellbeing implications. However, few studies have explored this pathway empirically. Moreover, research has rarely differentiated the role of objective and perceived built environment. Based on a survey of 1,553 older adults undertaken between 2015 and 2017 in Hong Kong, we explored the distal mediation pathway from objective built environment to both mental health and subjective wellbeing through perceived built environment and sense of community, using multilevel structural equation modeling. The results showed that perceived built environment and sense of community can fully explain the residential density and subjective wellbeing relationship. The inverted U-shape relationship between street connectivity and mental health was identified. Park-based green space had a protective role for both mental health and subjective wellbeing and was explained by two mediators, but vegetation-based green space was negatively associated with subjective wellbeing. Land use mix had positive total effects on both mental health and subjective wellbeing and was partially mediated by perceived built environment and sense of community. Recreational services showed a protective effect on both mental health and subjective wellbeing, and both were partially mediated by two mediators. The negative direct effect of health services on subjective wellbeing offsets the positive indirect effect through two mediators. The study findings have implications for landscape and urban planning policy and can provide an empirical contribution to the theoretical foundation of aging in place.
Understanding intermodal transit trip generation is essential to increase the share of long-distance transit trips among urban transportation systems. Although many studies have investigated trip generation, the existing literature still has limited evidence about intermodal transit trips and their nonlinear associations with the built environment over space. This study proposes a decision framework to identify the mean relative importance of socioeconomic attributes and built environment elements as well as their effective ranges and threshold effects at the spatial scale. An empirical study was conducted using large-scale smart card data in Nanjing, China. The modeling results indicate the proposed hybrid model can significantly enhance the predictive power, as compared to traditional models. The mean relative importance of the distance to the nearest metro station ranks the highest among all attributes studied, followed by bus route and land use mix. The effective ranges and thresholds of most built environment elements vary spatially with the upper quartile zones being the largest.
Previous research has reported that greenery is an important factor in walking activities, with greenery existing in various forms, including trees, gardens, green walls, and other examples. However, traditional methods of measuring urban greenery involve limitations in coverage of various forms of greenery and do not reflect the actual degree of exposure to pedestrians. Accordingly, this study examined the street Green View Index (GVI) and its associations with walking activities by different income groups using survey data on walking behaviors in 2350 residents in Seoul, Korea. This study utilized Google Street View (GSV) and deep learning to calculate the GVI by semantic segmentation, referring to greenness from the visual perspective of pedestrians. Correlation analyses between traditional greenery variables and GVI were conducted to examine differences, and multiple regression models were applied to identify the relationships between walking time and greenery variables. The results of this study show differences between conventional greenery variables and GVI in terms of specific greenery forms and perspectives. As hypothesized, GVI was more closely associated with walking time than the traditional greenery variables. Also, this study found that the low-income residents generally lived in low GVI neighborhood, but walking time is more sensitive to GVI. These results were because GVI represents the actual greenery exposure to pedestrians, and there was a difference between income groups in the degree of vehicle usage in daily life. The results of this study indicate that, when analyzing the relationship between urban greenness and walking behavior, it is necessary to examine the relationship from multiple angles and to investigate the importance of eye-level street greenery. Our findings provide useful insights for public policies to promote pedestrian walking environments.
A free-floating bike sharing system is an up-and-coming and marketable solution to promote transport flexibility and health benefits, which many people regard as a realistic way of generating more environmentally-friendly trips. Although many studies have investigated the associations between bike sharing usage and built environment, the existing literature has limited evidence about the relative importance of different built environment elements and their threshold impacts on cycling trips. This study contributes to the literature by proposing a modeling framework to explore the nonlinear impacts of built environment on bike sharing demand. A case study is conducted using Mobike bike sharing data in Chengdu. The analytical results indicate that population density and employment density are the two most significant factors that influence bike sharing usage. Total effects of land use variables rank the highest, followed by accessibility variables and transport facility variables. We then analyze the nonlinear impacts of different built environment elements on bike sharing usage to identify their effective ranges and threshold effects. These findings are important for planning departments to boost the share of non-motorized trips and embrace a cyclist-friendly design.