Computers Environment and Urban Systems最新文献

Individual trajectories, capturing significant human-environment interactions across space and time, serve as vital inputs for geospatial foundation models (GeoFMs). However, existing attempts at learning trajectory representations often encoded trajectory spatial-temporal relationships implicitly, which poses challenges in learning and representing spatiotemporal patterns accurately. Therefore, this paper proposes a joint spatial-temporal graph representation learning method (ST-GraphRL) to formalize structurally-explicit while learnable spatial-temporal dependencies into trajectory representations. The proposed ST-GraphRL consists of three compositions: (i) a weighted directed spatial-temporal graph to explicitly construct mobility interactions over space and time dimensions; (ii) a two-stage joint encoder (i.e., decoupling and fusion), to learn entangled spatial-temporal dependencies by independently decomposing and jointly aggregating features in space and time; (iii) a decoder guides ST-GraphRL to learn mobility regularities and randomness by simulating the spatial-temporal joint distributions of trajectories. Tested on three real-world human mobility datasets, the proposed ST-GraphRL outperformed all the baseline models in predicting movements' spatial-temporal distributions and preserving trajectory similarity with high spatial-temporal correlations. Furthermore, analyzing spatial-temporal features in latent space, it affirms that the ST-GraphRL can effectively capture underlying mobility patterns. The results may also provide insights into representation learnings of other geospatial data to achieve general-purpose data representations, promoting the progress of GeoFMs.

Overlapping structures, often overlooked, are crucial in shaping comprehensive urban development and broader megaregional strategies. To address the gap, this study conducts the overlapping communities analysis in the Pearl River Delta (PRD), a megaregion in South China, using big geospatial data from 2018. A novel Overlapping Community Detection based on Density Peaks (OCDDP) is employed to generate multiple communities with diverse functions for different nodes in the commuting network of 60 sub-city divisions. We identify eight overlapping communities in PRD characterized by two categories of communities predominantly centered around Shenzhen and Guangzhou, revealing a bicentric spatial structure. Notably, central sub-cities are characterized by a low-overlap attribute, while peripheral sub-cities manifest a high-overlap tendency. Furthermore, the study investigates the driving forces behind these communities through ridge regression to analyze the impacts of various spatial flows, including policies, investment amount and times, branch funding and number, travel cost, and travel distance, co-patenting, and search index. This part found that four Shenzhen-centric communities are primarily driven by travel cost, co-patenting, branch funding, and number, while the four Guangzhou-centric communities are influenced by co-patenting, investment amount, and times. This study emphasizes differentiated functional linkages and the need for precise policy positioning and resource allocation, paving the way for a coordinated and holistic approach to megaregional development.

Mega city regions (MCRs) have emerged in many countries in the process of urbanisation. Understanding the spatio-temporal dynamics of MCRs is crucial for sustainable urban development. However, the spatial scales and boundaries of these MCRs remain poorly defined, and their temporal dynamics have received limited attention. To address these gaps, we propose a new framework and GSMA algorithm that considers inter-city flows to identify MCRs' central cities, ranges and hinterlands. By utilising comprehensive data of over 30 million inter-city flow records covering 369 cities from Amap and Tencent, calibrated with official data from the Ministry of Transport, we identify 10 MCRs and 16 central cities in China, providing a clearer understanding of the spatial ranges and core areas of MCRs. We find that MCR ranges show relative stability during routine activities and expansions during holiday periods. Compared with previous methods, the proposed framework and algorithm have two prominent advantages. First, our methodology incorporates the directional characteristics of flows into the identification of MCRs' central cities. Second, we strike a balance between enlarging regional influence and tightening the internal connections in MCR delineation. In addition, by incorporating temporal changes in inter-city flows, the study reveals the temporal dynamics of MCRs which reflects the intricate interplay between human activities and urban system dynamics.

Over the past 15 years, environmental criminologists have explored the application of agent-based models (ABMs) of crime events and various theoretical frameworks applied to understand them. Models have supported criminological theorising and, in some cases, been applied to make predictions about the impact of interventions devised to reduce crime. However, decision-making frameworks utilised in criminological ABMs have typically been implemented through traditional techniques such as condition-action rules. While these models have provided significant insights, they neglect a crucial component of theoretical accounts of offending, the notion that offenders are learning agents whose behavioural dynamics change over time and space. In response, this article presents an ABM of residential burglary in which offender agents utilise reinforcement learning (RL) to learn behaviours. This solution enables offender agents to learn from individual-level perceptions of the environment and, given these perceptions, develop behavioural responses that benefit themselves. The model includes conceptualisations of the Routine Activity Theory (RAT), Crime Pattern Theory (CPT) and a utility function, Target Attractiveness, which acts as a behavioural mould to nudge offender agents to learn behaviours in keeping with the Rational Choice Perspective (RCP). Trained behaviours are then tested by introducing crime prevention interventions into the model and examining the reactions of offender agents. In keeping with empirical studies of offending, experimental results demonstrate that offender agents utilising RL learn to offend at targets where rewards outweigh risks and effort, offend close to home, frequently victimise high-rewarding targets, and conversely learn to avoid offending in areas associated with high levels of risk and effort.

Agent-based models (ABMs) of human systems are often parameterized using real-world data. For some ABMs this is not possible because the reality upon which the models are based does not exist or is not generalizable from one setting to another. In this paper we implement an online decision game to parameterize an agent-based model of pedestrian and cyclist route choice decisions in a neighbourhood. Our conceptual framework is to use an experimental game to log decision-making behaviour, summarize this behaviour into a decision model, and then transfer this model to an ABM. The product of this framework is an ABM with agents informed by human decision making made within the game, rather than the real world. The results of our analysis suggest that the decision model is consistent with some general theory about decision making, but the ABM illustrates some unique and contextually specific patterns of flow. ABMs parameterized with game data may be useful for forecasting the effects of change on urban transportation infrastructure.

The field of urban morphology, crucial for understanding the evolutionary trajectories of cityscapes, has traditionally depended on manual classification methods. The surge in deep learning and computer vision technologies presents an opportunity to automate and enhance urban typo-morphology studies. This research addresses three critical shortcomings in the current body of work: the neglect of urban fabric's three-dimensional qualities, the homogeneity of spatial scales in dataset creation and the dependence on a single-perspective for urban fabric classification. A novel deep learning-based model, UrbanClassifier, is introduced, trained on a substantial dataset that encapsulates the three-dimensionality of urban fabric along with morphological types and development periods. Extensive experimentation across four European cities highlights the model's ability to incorporate diverse spatial scales and viewpoints in urban fabric analysis. The UrbanClassifier exemplifies a method integrating features from various scales and perspectives, thus laying the groundwork for scalable and accessible urban typo-morphology studies, aiding practitioners in discerning the spatio-temporal evolution of urban fabric.

Interpretations of the urban cellular automata (CA) model aim to ensure that its predictive behaviors are consistent with real-world processes. Current urban CA interpretations have revealed the impacts of driving factors on land development suitability, or neighborhood effects and random perturbation on simulation results. However, three limitations remain unresolved: (1) the interpretations of deep learning (DL)-based urban CA are seldom integrated with the prerequired feature selection, (2) the input features from different urban CA modules are still explained by separate approaches, and (3) the interpretation results are rarely derived at the cell level to uncover spatially varying urban land development patterns. This study proposes a SHapley Additive exPlanations (SHAP)-based urban CA interpretation framework to address these challenges and improve urban CA. This framework uses model-level SHAP importance to identify dominant features from different modules for constructing the final simulation model. Then, cell-level SHAP importance is used to uncover spatially varying driving forces of urban expansion. The framework's effectiveness is rigorously tested and confirmed using a convolution neural network CA (CNN-CA) model for Dongguan City. The experimental results demonstrate that (1) SHAP-based model interpretation improves feature selection for DL-based urban CA. The figure of merit for CNN-CA calibrated using SHAP-based important features improves by 3%, outperforming the tested baseline methods. (2) SHAP measures the impacts of each feature from different CA modules in a whole. In this case, physical factors are much more important at the model level than proximity and accessibility factors, while neighborhood effect is the second most crucial factor. (3) Cell-level SHAP interpretations uncover spatially different urban land development patterns. For example, due to the extensive industrial land development in the northern Songshan Lake Zone, in the CNN-CA model, proximity to major roads within this region is associated with positive SHAP-based contribution share on cell-level urban expansion.

Large-scale text data from public sources, including social media or online platforms, can expand urban planners' ability to monitor and analyze urban conditions in near real-time. To overcome scalability challenges of manual techniques for qualitative data analysis, researchers and practitioners have turned to computer-automated methods, such as natural language processing (NLP) and deep learning. However, the benefits, challenges, and trade-offs of these methods remain poorly understood. How much meaning can different NLP techniques capture and how do their results compare to traditional manual techniques? Drawing on 90,000 online rental listings in Los Angeles County, this study proposes and compares manual, semi-automated, and fully automated methods for identifying context-informed topics in unstructured, user-generated text data. We find that fully automated methods perform best with more-structured text, but struggle to separate topics in free-flow text and when handling nuanced language. Introducing a manual technique first on a small data set to train a semi-automated method, however, improves accuracy even as the structure of the text degrades. We argue that while fully automated NLP methods are attractive replacements for scaling manual techniques, leveraging the contextual understanding of human expertise alongside efficient computer-based methods like BERT models generates better accuracy without sacrificing scalability.

Mobile phone network data is a vital source for unveiling human mobility characteristics in accordance with its large-scale spatiotemporal trajectory information. However, mobile phone network data usually records location at the level of cell towers, lacking accurate individual locations. Therefore, the authenticity and credibility of the conclusions drawn from such data are often questioned due to the spatial uncertainty. In this paper, we evaluate the location differences between users and the cell towers during connection establishment. Furthermore, we delve into the representation and contributing factors of spatial uncertainty, including cell tower density, antenna status, and user mobility. Our analysis is based on one-month mobile signaling data and taxi GPS data collected in Foshan (a prefecture-level city in China), which represent two forms of data on the mobility of the same individual. We conclude that to estimate user positions, areas significantly larger than the nearest cell tower are necessary. The influence of tower density and antenna load on connection accuracy does not exhibit a straightforward linear dependency; instead, it fluctuates once a threshold is reached. Connection accuracy is typically higher when users are stationary than when they are in motion. Our findings together indicate that it should carefully assess the accuracy of position estimation when mapping from cell tower location to user location.

Building characteristics, such as number of storeys and type, play a key role across many domains: interpreting urban form, simulating urban microclimate or modelling building energy. However, geospatial data on the building stock is often fragmented and incomplete. Here, we propose a novel and easily adaptable method to predict building characteristics in diverse cities, which attempts to mitigate such data gaps. Our method exploits the geospatial connectivity between street-level urban objects and building characteristics by employing graph neural networks, as they can model spatial relationships and leverage them for predictions. We apply this approach in three representative cities (Boston, Melbourne, and Helsinki) that offer a variety of building features as prediction targets (storeys, types, construction period and materials) and diverse urban environments as predictors. Overall, the magnitude of errors is acceptable for a series of use cases. In the prediction of building storeys, an average of 81.83% buildings in three cities have less than one-storey prediction error. We also find that the prediction of building type achieves an average of 88.33% accuracy across three cities. Meanwhile, an average of 70.5% of buildings are correctly classified by construction period in Melbourne and Helsinki, and the building material prediction accuracy is 68% in Helsinki. The results confirm that our approach is adaptable across different urban environments because comparable performance is achieved in the other two cities. Further, we assess the impact of varying local data availability on model performance. Our findings underscore the feasibility of the method in scenarios with sparse building data (10%, 30% and 50% availability). Our graph-based approach advances research on filling in incomplete building semantics from existing datasets, and showcases the potential to enable 3D city modelling. Given the broad applicability of the approach to predicting many building characteristics, diverse downstream applications exist, such as enhancing contemporary urban studies (e.g. exploring streetscapes) and facilitating the development of 3D GIS (e.g. maintaining and updating 3D building settings).