Computers Environment and Urban Systems最新文献

Prediction of high-risk areas for urban crime is of great significance for maintaining public safety and sustainable development. However, existing approaches are deficient in spatiotemporal sensitivity and perceptivity, which make it difficult to extract the spatiotemporal dependency from uneven and sparsely distributed data. To address this problem, the novel multi-scale neural network models, namely ST-HGNet and ST-HGNet(a) with attention, were proposed. It is dedicated to further exploring spatiotemporal patterns and improving hotspot location prediction accuracy for sparse types of crimes. First, multi-scale conception and attention mechanisms were introduced to address the receptive field range fixed problem. It enhanced representation of captured information by exposing spatial “scale” dimension and assigning weight relationships. Then, novel multi-scale hierarchical gating architecture was designed that has two forms of whether to add attention or not, to enhance the sensitivity of features and the perception of sparse features by filtering the valid information at different scales. Ultimately, the periodic temporal components were used to capture different time-trend dependencies. The proposed model adopted well-known Chicago assault crime dataset as a case study. Compared with five common benchmark models, the results show that the ST-HGNet model outperformed other baseline models and achieved higher prediction accuracy at multiple level spatial resolution. In particular, ST-HGNet(a) with self-attention achieved the greatest improvement at 1000 m, with a mean hit rate of more than 84%.

Urbanization-induced land cover changes significantly impact ecological environments and socioeconomic growth. Vector-based cellular automata (VCA) models are an advanced cellular automata (CA) method that use irregular cells and perform well in simulating land use changes within urban areas. However, the applicability and parameter setting of VCA models for land cover change simulation are still challenging for researchers. To address this issue, this study applied a VCA model and two raster-based models, i.e., a pixel-based CA model and a patch-based CA model, to simulate and compare their performance in simulating land cover changes. The results show that VCA and patch-based CA were superior, with VCA's FoM being 39.74% higher than pixel-based CA and 11.00% over patch-based CA. VCA effectively tracks construction land expansion in rapidly developing areas, while patch-based CA excels in central urban and suburban shifts, fitting broader study scopes. Additionally, a spatial scale sensitivity analysis of the VCA model revealed that a smaller VCA cell size improves accuracy but introduces a risk of spatial pattern errors. Notably, the scope of study impacts VCA accuracy more than cell size. These findings bolster land cover change modeling theory and offer insights for precise future land cover change simulations and decision-making.

Streets are a crucial part of the built environment, and their layouts, the street patterns, are widely researched and contribute to a quantitative understanding of urban morphology. However, traditional street pattern analysis only considers a few broadly defined characteristics. It uses administrative boundaries and grids as units of analysis that fail to encompass the diversity and complexity of street networks. To address these challenges, this research proposes a machine learning-based approach to automatically recognise street patterns that employs an adaptive analysis unit based on street-based local areas (SLAs). SLAs use a network partitioning technique that can adapt to distinct street networks, making it particularly suitable for different urban contexts. By calculating several streets’ network metrics and performing a hierarchical clustering method, streets with similar characters are grouped under the same street pattern. A case study is carried out in six cities worldwide. The results show that street pattern types are rather diverse and hierarchical, and categorising them into clearly demarcated taxonomy is challenging. The study derives a set of new morphometrics-based street patterns with four major types that resemble conventional street patterns and eleven sub-types to significantly increase their diversity for broader coverage of urban morphology. The new patterns capture urban structural differences across cities, such as the urban-suburban division and the number of urban centres present. In conclusion, the proposed machine learning-based morphometric street pattern to characterise urban morphology has an enhanced ability to encompass more information from the built environment while maintaining the intuitiveness of using patterns.

The issue of jobs-housing balance concerns the sustainable development of cities and the well-being of residents. Conventional measurement approaches, however, often fall short due to the zoning problem (as a subproblem of the modifiable areal unit problem), leading to inconsistent and inaccurate results depending on the spatial partitioning scheme applied. This paper discusses the application and advantages of the local colocation quotient method in jobs-housing balance measurement. A case study of Nanjing, China, is selected, and mobile location data are used to obtain the jobs and housing locations of workers. Then, the adjusted jobs-workers ratio and the local colocation quotient values that reflect the degree of jobs-housing balance are calculated and compared by category. The results show that on the one hand, due to the zoning effect, when points are aggregated into spatial units, some points with different spatial characteristics are masked by the dominant value of the units; on the other hand, the local colocation quotient method can solve the zoning problem and obtain more fine-scale and accurate results, thus providing a new analytical tool and perspective for this field.

African cities are growing rapidly and more than half of their populations live in deprived areas. Local stakeholders urgently need accurate, granular, and routine maps to plan, upgrade, and monitor dynamic neighborhood-level changes. Satellite imagery provides a promising solution for consistent, accurate high-resolution maps globally. However, most studies use very high spatial resolution images, which often cover only small areas and are cost prohibitive. Additionally, model transferability to new cities remains uncertain. This study proposes a scalable and transferable approach to routinely map deprived areas using free, Sentinel-2 images. The models were trained and tested on three cities: Lagos (Nigeria), Accra (Ghana), and Nairobi (Kenya). Contextual features were extracted at 10 m spatial resolution and aggregated to a 100 m grid. Four machine learning algorithms were evaluated, including multi-layer perceptron (MLP), Random Forest, Logistic Regression, and Extreme Gradient Boosting (XGBoost). The scalability of model performance was examined using patches of the different deprived types identified through visual image interpretation. The study also tested the ability of models to map deprived areas of different types across cities. Results indicate that deprived areas have heterogeneous local characteristics that affect large area mapping. The top 25 features for each city show that models are sensitive to the spatial structures of deprived area types. While models performed well on individual cities with XGBoost and MLP achieving an F1 scores of over 80%, the generalized model proves to be more beneficial for modeling multiple cities. This approach offers a promising solution for scaling routine, accurate maps of deprived areas to hundreds of cities that currently lack any such map, supporting local stakeholders to plan, implement, and monitor geotargeted interventions.

Despite the rapid advancement and extensive applications of online Volunteered Geographical Information (VGI) projects such as OpenStreetMap (OSM), the persistence of data inequality remains a significant challenge, compromising the global reliability of their data products. This study examines the influence of contributions made by organizations, which have notably risen within the OSM community, on data inequality. The Gini coefficient is utilized to quantify data inequality, while a suite of statistical methods, including spectral analysis and robust correlation analysis, is applied to evaluate the distribution and impact of organizational efforts across various nations. Our findings indicate that organizations predominantly allocate their resources to nations with less complete data and surpass collective efforts of average contributors in mitigating OSM data inequality. Furthermore, the phenomena appears to be particularly significant for NGOs or corporations with humanitarian visions.

Most studies of urban morphology rely on morphometrics, such as building area and street length. However, these methods often fall short in capturing visual patterns that carry abundant information about the configuration of urban elements and how they interact spatially. In this study, we introduce a novel method for learning morphology features based on figure-ground maps, which leverages recent developments in computer vision. Our method facilitates discovering and comparing urban form types in a fully unsupervised manner. Specifically, we examine building fabrics by 1 km patches. A visual representation learning model (SimCLR) casts each patch into a latent embedding space where similar patches are clustered while dissimilar patches are dispelled, thus generating morphology representations that entail the layout of building groups. The learned morphology features are tested in urban form typology clustering and comparison tasks in four diverse cities: Singapore, San Francisco, Barcelona, and Amsterdam, with data sourced from OpenStreetMap. Clustering results show effective identification of typical urban morphology types corresponding to urban functions and historical developments. Further analyses based on the representations reveal inner- and cross-city morphological homogeneity relating to socio-economic drivers. We conclude that this method is a promising alternative for effectively describing urban patterns in morphology analysis.

Bus-based mobile sensing has emerged as a cost-effective approach for collecting high spatio-temporal air quality data by leveraging the mobility of buses. However, when selecting an optimal subset of buses from a large fleet for deploying a limited number of sensors, existing studies have primarily focused on assessing the coverage of the study area by buses, disregarding the temporal gap between consecutive coverage at specific locations. It is worth noting that pollutant concentrations exhibit smooth variations over time, rendering data collected at very short intervals redundant. Therefore, this study first identified five key criteria for evaluating the air quality monitoring importance in various locations. Then two bus selection models that consider both the spatiotemporal coverage of the study area and the temporal gap between sensing data are proposed. Specifically, the maximal spatio-temporal coverage bus selection model (MaxCoverage) maximizes overall spatio-temporal coverage with a guaranteed time interval between consecutive sensor measurements, and the minimal fleet size model (MiniSize) selects the minimum number of buses based on based on specified requirements for monitoring time interval and counts. Experimental validation using a real-world bus trajectory dataset from Shenzhen, China demonstrates the effectiveness of the proposed models. The results show that the MaxCoverage_TC1 model has time intervals 2.7 timeslots longer than the baseline, and the MiniSize_TC1 model has an average time interval that is 1.4 timeslots longer.

Urban forests provide many benefits for municipalities and their residents, including air quality improvement, urban atmosphere cooling, and pluvial flooding reduction. Monitoring the trees is one of the tasks among the several urban forest assessment procedures. Trees with a risk of falling may threaten the locals and the infrastructure of the cities, thereby being an immediate concern for forestry managers. In general, a set of measures and aspects are collected from field survey analysis to estimate whether the trees represent a risk to the safety of the urban spaces. However, gathering the tree's physical measures in fieldwork campaigns is time-consuming and laborious considering the massive number of trees in the cities. Therefore, there is an urge for new computational-based methodologies, especially those related to the latest advances in artificial intelligence, to accelerate the assessment of trees in the municipality areas. In this sense, this work aims at using several machine learning-based methods in the context of tree condition inspection. Particularly, we present the prediction of the tree failure probability by using several aspects collected over time from fieldwork campaigns, with a special focus on external physical measures of the trees. Further, we provide the samples with their respective tree failure probability values as a new open dataset for further investigations on tree status monitoring. We also present a novel dataset composed of images of trees with bounding boxes delineations of the tree, trunk, and crown for automating the tree monitoring tasks. Regarding the tree failure probability estimation, we compared several regression algorithms for estimating the tree failure likelihood. Moreover, we propose a stacking generalization approach to enhance forecast accuracy and minimize prediction errors. The results showed the viability of the proposed method as an auxiliary tool in tree analysis tasks, which attained the lowest average Mean Absolute Error of 5.6901$\pm$1.1709 yielded by the stacking generalization model.

A comprehensive understanding of population distribution is critical for assessing socio-economic issues. However, the widely used dasymetric mapping method relies on models built at a coarse administrative scale and estimates population at a fine-gridded scale. This difference in scale between the training and estimating domains results in significant heterogeneity in data distribution. To address this issue, we proposed a scale heterogeneity-avoided method based on artificial neural networks that can take population density as an independent variable and gridded properties, including remote sensing images, digital terrain models, road networks, building footprints, and land use, as dependent variables. Our experiments in Hong Kong in 2016 and 2021 showed significant advantages of the proposed method. Compared to commonly used methods, our approach demonstrated a 19.4% improvement in the root mean square error. Furthermore, the advantages of our method became more apparent at larger census units, and the accuracy of the pre-trained model for directly estimating population in other temporal phases was satisfactory. Among the geospatial data variables, land use was the most significant in accurately estimating population. Replacing land use data with random numbers led to a decrease in accuracy by over 89.0%, while other properties only resulted in decreases of 2.7% to 13.9%. We further investigated spatiotemporal changes in population distribution from 2016 to 2021, finding that population growth mainly occurred in new built-up areas, while larger population decreases occurred in old towns. Throughout the study period, the population tended to concentrate more, as the average population density increased while the median population density decreased.