Street trees are vital to urban livability, providing ecological and social benefits. Establishing a detailed, accurate, and dynamically updated street tree inventory has become essential for optimizing these multifunctional assets within space-constrained urban environments. Given that traditional field surveys are time-consuming and labor-intensive, automated surveys utilizing Mobile Mapping Systems (MMS) offer a more efficient solution. However, existing MMS-acquired tree datasets are limited by small-scale scene, limited annotation, or single modality, restricting their utility for comprehensive analysis. To address these limitations, we introduce WHU-STree, a cross-city, richly annotated, and multi-modal urban street tree dataset. Collected across two distinct cities, WHU-STree integrates synchronized point clouds and high-resolution images, encompassing 21,007 annotated tree instances across 50 species and 2 morphological parameters. Leveraging the unique characteristics, WHU-STree concurrently supports over 10 tasks related to street tree inventory. We benchmark representative baselines for two key tasks—tree species classification and individual tree segmentation—based on 18 major species and an “Others” category. Extensive experiments demonstrate that while multi-modal fusion yields improvements over uni-modal baselines, it currently presents performance gaps compared to strong 3D-only methods, indicating that effective fusion remains a challenging open problem requiring further research. In particular, we identify key challenges and outline potential future works for fully exploiting WHU-STree, encompassing multi-modal fusion, multi-task collaboration, cross-domain generalization, spatial pattern learning, and Multi-modal Large Language Model for street tree asset management. The WHU-STree dataset is accessible at: https://github.com/WHU-USI3DV/WHU-STree.

The top-down method is widely used to estimate China's CO2 emissions at the county level. However, studies have relied on a single indicator of regional total nighttime light brightness as an instrumental variable for prediction, leading to the assumption that there is a positive correlation between CO2 emissions and total nighttime light brightness in all regions within the same province. This assumption overlooks other heterogeneous relationships and does not correspond to reality. Therefore, this study constructed a dataset of potential feature variables based on multisource data (improved and calibrated nighttime light data, urban and rural human settlement data, and socioeconomic indicator data based on statistical yearbooks). After the main feature variables were identified, a hybrid regression algorithm combining deep neural networks and CatBoost was constructed to generate instrumental variable for predicting CO2 emissions. Compared with the total nighttime brightness, it has a stronger linear relationship with CO2 emissions. Using the top-down algorithm, we estimated China's monthly CO2 emissions at the county level from 2013 to 2021. This dataset provides a solid foundation for predicting the achievement of China's county-level "dual carbon" strategy. The methods used in this study can be generalized to other global regions.

Rapid urbanization has substantially increased the complexity of urban underground spaces. This complexity leads to frequent road collapse incidents that pose significant threats to the safety and property of urban residents. Therefore, accurate methods of performing early road collapse risk assessments are crucial for preventing these incidents and emergency preparedness. In this study, road collapse incident data for 2016–2021 were collected for Foshan, Guangdong Province, a city in southern China. Utilizing InSAR time-series data from Sentinel-1 satellites, ground subsidence maps were generated, and the publicly accessible Ground Subsidence Trend-Based Urban Road Collapse Risk Dataset (GSTURCRD) was constructed. A novel risk assessment method for urban road collapse based on an extended long short-term memory (xLSTM) network was proposed. This method introduces two new LSTM variants, the scalar LSTM (sLSTM) and the matrix LSTM (mLSTM), incorporating exponential gating and an innovative matrix memory structure. These variants are integrated using residual connections to form a comprehensive network architecture that enables effective learning and representation of the temporal features. The experimental results from the dataset demonstrate that the proposed method significantly outperforms the original LSTM network and traditional machine learning methods regarding assessment capability (its accuracy was 0.886, and its recall was 0.857). Furthermore, the method's effectiveness was validated by an analysis of actual incidents that occurred in Foshan; thus, its ability to generate accurate and timely detections and provide early warnings for high-risk road sections in urban areas was confirmed.

Place-level human mobility reflects the collective movement patterns of individuals and groups within defined geographic areas for specific mobility patterns. The COVID-19 pandemic has underscored the pressing concerns on mobility vulnerability during urban crisis. While socioeconomic disparities in mobility disruptions have been thoroughly documented, the impact of the built environment during the pandemic remains inadequately explored. Moreover, the vulnerability of collective human mobility in specific places, considering dual-directional patterns of both incoming and outgoing behaviors, is not well understood. This study utilizes extensive mobile phone data to investigate human mobility vulnerability across U.S. cities at the census block group (CBG) level during COVID-19, focusing on both incoming and outgoing mobility patterns. By integrating socioeconomic and built environment factors, we aim to identify the determinants that influence place-level mobility vulnerability in response to the pandemic. We assess year-over-year disparities in bidirectional mobility density, dwell time, and distance between 2019 and 2020 to evaluate their vulnerabilities in CBGs. The results reveal significant roles of built environment variables on the vulnerability and robustness of various mobility patterns. Our findings underscore the pronounced advantages and drawbacks of the built environments such as developed open space, retail density, employment diversity, job-worker balance, walkability, and transit service frequency on specific patterns of incoming and outgoing mobility vulnerability. Furthermore, interventions in the built environment aimed at promoting sustainable mobility should also consider the potential threats associated with mobility vulnerability. These insights provide practical implications for post-pandemic planning initiatives designed to enhance resilience.

Collective human mobility is an important phenomenon within defined geographic areas for specific mobility patterns. Studies have been conducted to understand human mobility patterns and the determinants, including the built environment, and socioeconomic factors. However, there is a dearth of systematic investigation focusing on the dual-directional nature of collective human mobility in places, referring to both incoming and outgoing mobility behaviors. This study addresses this gap by analyzing mobility density, dwell time, and trip distance of both incoming and outgoing travel behaviors using mobile phone big data in 165,181 census block groups in U.S. cities in 2019. As the results show, built environment features usually demonstrate greater explanatory power than socioeconomic variables, highlighting their vital roles in shaping human mobility. Specifically, built environment characteristics, including developed open space, population density, employment diversity, street intersections, walkability, transit service, and destination accessibility, are noticeably associated with collective mobility patterns in terms of incoming and outgoing density, dwell time, and distance. Socioeconomic variables, such as the proportion of older adults, Black individuals, household income, commuting mode choices are also significantly linked to specific mobility patterns. We also suggest policy implications for built environment interventions to support sustainable place-level human mobility.

Physical disorder is associated with negative outcomes in economic performance, public health, and social stability, such as the depreciation of property, mental stress, fear, and crime. A limited but growing body of literature considers physical disorder in urban space, especially the topic of identifying physical disorder at a fine scale. There is currently no effective and replicable way of measuring physical disorder at a fine scale for a large area with low cost, however. To fill the gap, this article proposes an approach that takes advantage of the massive volume of street view images as input data for virtual audits and uses a deep learning model to quantitatively measure the physical disorder of urban street spaces. The results of implementing this approach with more than 700,000 streets in Chinese cities-which, to our knowledge, is the first attempt globally to quantify the physical disorder in such large urban areas-validate the effectiveness and efficiency of the approach. Through this large-scale empirical analysis in China, this article makes several theoretical contributions. First, we expand the factors of physical disorder, which were previously neglected in U.S. studies. Second, we find that urban physical disorder presents three typical spatial distributions-scattered, diffused, and linear concentrated patterns-which provide references for revealing the development trends of physical disorder and making spatial interventions. Finally, our regression analysis between physical disorder and street characteristics identified the factors that could affect physical disorder and thus enriched the theoretical underpinnings.

Channelization is the most common hydraulic modification of urban rivers. Here, we assessed the effects of urban river morphology on benthic communities by analyzing the characteristics of benthic communities at various sites in channelized and natural rivers of the Longgang River system in southern China. We detected four Clitellata species, five Oligochaeta species, one Polychaeta species, 10 Gastropoda genera/species, two Bivalvia genera/species, two Crustacea genera/species, and 14 Insecta genera/species. Insecta and Oligochaeta were the dominant classes in the wet and dry seasons, and Chironomus plumosus was the most dominant species. The density of Clitellata was significantly lower in channelized rivers (0–0.74 ind/m2) than in natural rivers (0.61–4.85 ind/m2). The Shannon’s diversity index was significantly lower in channelized rivers (0.66–1.04) than in natural rivers (0.83–1.28) in the wet and dry season. NH3.N was positively correlated with Shannon’s diversity index, and chemical oxygen demand and river width were negatively correlated with Shannon’s diversity index. When the concentration of total phosphorus (TP) was low (<3 mg/L), it was positively correlated with Shannon’s diversity index. Our findings indicate that river channel morphology affects benthic faunal structure and diversity, but the effects varied among seasons. Minimized channelization will prevent the loss of aquatic biodiversity in subtropical urban rivers, as will preservation of natural rivers.

The understanding of pedestrian-level greenery across urban forms in built environment configurations in high-density cities is insufficient. We conducted a citywide investigation of urban greenery from the pedestrian perspective by developing a deep learning technique to extract greenery from fisheye images generated from Google Street View images in Hong Kong. Relying on open-source data, we compared pedestrian-level greenery measurements with the satellite-based normalized difference vegetation index (NDVI) in diverse urban forms represented by local climate zone classes. Street greenery was spatially variant, and low greenery was found predominantly in private residential and commercial/business lands in high-density areas. Pedestrian-level measurement and the NDVI were strongly correlated, but the inconsistency between them increased from high-and mid-rise forms to low-rise forms and from compact forms to open forms. We also demonstrated the idea of integrating nearby street greenery with spatial information on population and urban morphology for inequality analysis. Potential implications for urban planning are provided. The findings linking street greenery with urban morphology are useful for urban and greenery planning in climate-resilient, sustainable, and healthy cities. Our analytical approach using open-source data is transferable to other high-density cities.

Transit-oriented development (TOD) planning strategy has been widely implemented worldwide to formulate dense, mixed-use built environment in the past three decades. The primary goal of TOD is to promote public transit usage including both transit mode share and ridership. Research supports that built environment characteristics around metro stations affect residents' travel behaviors and metro usage. However, the evidence remains inconsistent in different urban contexts. Furthermore, research focusing on mode share such as commuting trips at station level is still scarce. In this study, a rule-based model was used to identify commuting trips using metro service with smart card data (SCD), covering more than 90 percent of all metro passengers in Wuhan, China. Built environment characteristics around metro stations were measured with a 3Ds framework (density, diversity, and design). Results suggest that population density is negatively associated with metro commuting mode share, while street intersection shows a positive relationship. Office-oriented urban function and street intersection are positively correlated with metro ridership. Hence, exploring the fine-grained relationship of metro usage and built environment factors around transit stations in different urban and social contexts warrants further research attention.

Urban spatial structure, which is primarily defined as the spatial distribution of employment and residences, has been of lasting interest to urban economists, geographers, and planners for good reason. This paper proposes a nonparametric method that combines the Jenks natural break method and the Moran’s I to identify a city’s polycentric structure using point-of-interest density. Specifically, a polycentric city consists of one main center and at least one subcenter. A qualified (sub)center should have a significantly higher density of human activity than its immediate surroundings (locally high) and a relatively higher density than all the other subareas in the city (globally high). Treating Chinese cities as the subject, we ultimately identified 70 cities with polycentric structures from 284 prefecture-level cities in China. In addition, regression analyses were conducted to reveal the predictors of polycentricity among the subjects. The regression results indicate that the total population, GDP, average wage, and urban land area of a city all significantly predict polycentricity. As a whole, this paper provides an alternative and transferrable method for identifying main centers and subcenters across cities and to reveal common predictors of polycentricity. The proposed method avoids some of the potential problems in the conventional approach, such as the arbitrariness of thres hold. setting and sensitivity to spatial scales. It can also be replicated rather conveniently, as its input data, such as point-of-interest data, are widely available to the public and the data’s validity can be efficiently checked by field trips or other traditional data sources, such as land-use maps or censuses.

The coronavirus pandemic is an ongoing global crisis that has profoundly harmed public health. Although studies found exposure to green spaces can provide multiple health benefits, the relationship between exposure to green spaces and the SARS-CoV-2 infection rate is unclear. This is a critical knowledge gap for research and practice. In this study, we examined the relationship between total green space, seven types of green space, and a year of SARS-CoV-2 infection data across 3,108 counties in the contiguous United States, after controlling for spatial autocorrelation and multiple types of covariates. First, we examined the association between total green space and SARS-CoV-2 infection rate. Next, we examined the association between different types of green space and SARS-CoV-2 infection rate. Then, we examined forest-infection rate association across five time periods and five urbanicity levels. Lastly, we examined the association between infection rate and population-weighted exposure to forest at varying buffer distances (100 m to 4 km). We found that total green space was negative associated with the SARS-CoV-2 infection rate. Furthermore, two forest variables (forest outside park and forest inside park) had the strongest negative association with the infection rate, while open space variables had mixed associations with the infection rate. Forest outside park was more effective than forest inside park. The optimal buffer dis-tances associated with lowest infection rate are within 1,200 m for forest outside park and within 600 m for forest inside park. Altogether, the findings suggest that green spaces, especially nearby forest, may significantly mitigate risk of SARS-CoV-2 infection.

The influence of high-density environment on urban residents is controversial, and its effect varies with specific contexts. Meanwhile, urban planners and policy-makers are increasingly aware that urban greenery may mitigate the detrimental effects of crowded environments on quality of life in high-density cities. However, little empirical evidence is available in the context of China. This study aims to examine the complex relationship between urban density, urban greenery, and older people's life satisfaction, with survey data collected from 1,594 older adults in 129 neighborhoods in Shanghai, China. Urban density was assessed using floor area ratio and building coverage ratio respectively, and urban greenery was measured by street view greenery, greening rate, Normalized Differential Vegetation Index (NDVI), and accessibility to nearest parks. Results from structural equation modeling showed that higher urban density was related to lower life satisfaction, and a reduced sense of community was a significant pathway between higher urban density and lower life satisfaction. Furthermore, eye-level greenery cushioned the negative effect of urban density on life satisfaction. Our findings highlighted the necessity of optimizing high-density neighborhood environments and promoting eye-level greenery in high-density urban areas to create aging-friendly cities.

Many studies have confirmed that the characteristics of the built environment affect individual walking behaviors. However, scant attention has been paid to population-level walking behaviors, such as pedestrian volume, because of the difficulty of collecting such data. We propose a new approach to extract citywide pedestrian volume using readily available street view images and machine learning technique. This innovative method has superior efficiency and geographic reach. In addition, we explore the associations between the extracted pedestrian volume and both macro-and micro-scale built environment characteristics. The results show that micro-scale characteristics, such as the street-level greenery, open sky, and sidewalk, are positively associated with pedestrian volume. Macro-scale characteristics, operationalized using the 5Ds framework including density, diversity, design, destination accessibility, and distance to transit, are also associated with pedestrian volume. Hence, to stimulate population-level walking behaviors, policymakers and urban planners should focus on the built environment intervetions at both the micro and macroscale.

Promoting urban vibrancy is one of the major objectives of urban planners and government officials, and it is linked to various benefits, such as urban prosperity and human well-being. There is ample evidence that built environment characteristics are associated with urban vibrancy; however, the spatiotemporal associations between built environment and urban vibrancy have not been fully investigated owing to the inherent limitations of traditional data. To address this gap, we measured spatiotemporal urban vibrancy in Shenzhen, China, using Tencent location-based big data, which is characterized by fine-grained population-level spatiotemporal granularity. Built environment characteristics were systematically measured using the 5D framework (density, diversity, design, destination accessibility, and distance to transit) with multi-source datasets. We investigated the spatiotemporal non-stationary associations using a geographically and temporally weighted regression (GTWR) model. The results indicated that the GTWR models achieved better goodness-of-fit than linear regression models. Built environment factors such as population density; point of interest (POI) mix; residential, commercial, company, and public service POI; and metro station were significantly associated with urban vibrancy. Time series clustering revealed spatiotemporal clustered patterns of the associations between built environment factors and urban vibrancy. To promote urban vibrancy with urban planning and design strategies, both the spatial and temporal associations between the built environment and urban vibrancy should be considered.

Considering that most working adults spend nearly half their waking time at work, creating a supportive built environment around workplaces could be a feasible approach to maintain adequate levels of physical activity. However, the extent to which the built environment around workplaces influences walking behaviors in working adults remains unclear. Using survey data of 1009 full-time employees in Shanghai, China, this study assessed the nonlinear relationships between the built environment characteristics around workplaces and three domains of walking behaviors (commuting, utilitarian, and recreational walking). Using gradient boosting decision trees, our results showed that the built environment around workplaces is crucial for higher levels of walking behaviors, but built environment features tended to have distinctive associations with different domains of walking behaviors. Specifically, the number of physical activity facilities was positively associated with all three domains of walking behaviors, while a high floor area ratio was negatively associated with different domains of walking behaviors to some extent. Furthermore, several built environment characteristics, such as land use entropy, street view greenery, distance from home to the city center, and distance between the city center and workplaces had distinctive associations with different domains of walking behaviors. The findings of this study could provide nuanced guidance for creating pedestrian-friendly environments around workplaces to promote walking behaviors and overall physical activity levels in the working population.

The creation of walkable environments, and the promotion of walkability for health and environmental benefits have been widely advocated. However, the term "walkability" is often associated with two related but distinct walking behaviors: individual and collective walking behaviors. It is unclear whether spatial disparity exists between them, and whether built environment characteristics have distinctive effects on them. This research was the first to explore the spatial disparity between the two types of walking behaviors. Collective walking behaviors were measured using the citywide pedestrian volume, extracted from 219,248 street view images. Individual walking behaviors were measured form a population-level survey. Spatial mismatches were found between the two types of walking behaviors and built environment elements had stronger associations with collective walking behaviors. Therefore, it is prudent to theoretically differentiate collective and individual walking behaviors, and targeted planning policies must be developed to promote one or both types of walking behaviors.

Urban greenery is closely related to people's behaviour. With the advancement of science and technology in Artificial Intelligence, wearable sensors and cloud computing, the potential for studying the relationship between people and urban greenery through new data and technology is constantly being explored, such as assessing population exposure to urban greenery using multi-source big data. Taking one individual participant as a case study, this paper proposes and validates the effectiveness of using wearable camera (Narrative Clip 2) and machine learning (Applications Programming Interface of Microsoft Cognitive Service) to assess personal exposure to urban greenery. Microsoft API is used to identify urban greenery tags, including "flower", "forest", "garden", "grass", "green", "plant", "scene" and "tree", in personal images taken by the wearable camera. Personal exposure to urban greenery is assessed by calculating the frequency of the urban greenery tags in all the images taken. Furthermore, the overall evaluation and regularity of personal exposure to urban greenery (including "static exposure" and "dynamic exposure") are explored to identify the characteristics of individual's greenery lifelogging. This study makes a brave attempt that may contribute a new perspective in applying personal big data in studying individual behaviour.

In view of the current problems encountered in public safety and emergency management, this study proposes a method for constructing a public safety map. Based on the public safety theory and the existing POI system, a method for constructing a public safety attribute index system is proposed, and a preliminary discussion is made on the collection mechanism of public safety attribute data and the operation mechanism of the public safety map. It is hoped to provide a universal and easy access to emergency space information for emergency management departments and rescue workers.

There is striking racial disparity in the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection rates in the United States. We hypothesize that the disparity is significantly smaller in areas with a higher ratio of green spaces. County level data on the SARS-CoV-2 infection rates of black and white individuals in 135 of the most urbanized counties across the United States were collected. The total population in these counties is 132,350,027, comprising 40.3% of the U.S. population. The ratio of green spaces by land-cover type in each county was extracted from satellite imagery. A hierarchical regression analysis measured cross-sectional associations between racial disparity in infection rates and green spaces, after controlling for socioeconomic, demographic, pre-existing chronic disease, and built-up area factors. We found a higher ratio of green spaces at the county level is significantly associated with a lower racial disparity in infection rates. Four types of green space have significant negative associations with the racial disparity in SARS-CoV-2 infection rates. A theoretical model with five core mechanisms and one circumstantial mechanism is presented to interpret the findings.

AbstractThis study examined the associations between green spaces and one-years’ worth of SARS-CoV- 2 infection rates across all 3,108 counties in the contiguous United States after controlling for multiple categories of confounding factors. We found green spaces at the county level have a significant negative association with infection rates. Among all types of green spaces, forest yields the most consistent and strongest negative association. Sensitivity analyses confirmed the negative association of forest across five urbanicity levels, and the strength of the association increases as disease incidence increases across five time periods. Although forest located in moderately urbanized counties yields the strongest association, the negative pattern of significant associations holds across all five urbanicity levels. A population-weighted analysis revealed that proximity to forest within a moderate walking distance (≤ 1.0–1.4 km) may provide the greatest protection against the risk of infection.