Computers Environment and Urban Systems最新文献_第2页

Designing and testing microtransit routes to improve social inclusion: A pilot study in a suburban area 设计和测试微交通路线以提高社会包容性：郊区的试点研究

IF 8.3 1区地球科学 Q1 ENVIRONMENTAL STUDIES

Computers Environment and Urban Systems

Pub Date : 2026-04-01 Epub Date: 2025-12-23 DOI: 10.1016/j.compenvurbsys.2025.102394

Alessandro Emilio Capodici , Martina Citrano , Gabriele D'Orso , Marco Migliore , Leonardo Minaudo , Riccardo D'Angelo

Suburbs are often characterized by a scarcity of mobility options to access services. Introducing microtransit is a promising way to improve public transport in suburbs, ensuring greater social inclusion and connecting isolated areas to main transit hubs. The paper aims to develop a multi-step GIS-based methodology for designing semi-flexible stop-based microtransit, having fixed routes and flexible routes (detours) and operating with real-time ride bookings (zero lead time). We considered a suburban area in Palermo, Italy, as study area. The identification of fixed and flexible routes was based on the forecasted passenger flows, through the estimate and the assignment of the daily origin-destination matrix for microtransit, also considering safety, spatial, and technical constraints. A small-scale pilot was carried out between November and December 2022 to test microtransit routes and the reliability of a mobile application to operate the service. A customer satisfaction and a willingness-to-pay survey were addressed to the users. The small-scale pilot showed that microtransit could improve public transportation in suburbs, being more accessible and reducing waiting times at stops. Particularly, the result of the design process led to a semi-flexible service accessible by 90 % of the resident population and with waiting times of less than 15 min in 76 % of the rides, lower than those currently experienced by bus users (20 min).

郊区的特点往往是缺乏获得服务的流动选择。引入微型交通是改善郊区公共交通的一种有希望的方式，可以确保更大的社会包容性，并将偏远地区与主要交通枢纽连接起来。本文旨在开发一种基于gis的多步骤方法，用于设计半灵活的基于站点的微交通，具有固定路线和灵活路线（绕路），并运行实时乘车预订（零提前期）。我们把意大利巴勒莫的一个郊区作为研究区域。确定固定和灵活的路线是基于预测的客流量，通过估计和分配微交通的每日始发-目的地矩阵，同时考虑安全、空间和技术限制。2022年11月至12月期间进行了小规模试点，以测试微交通路线和运营该服务的移动应用程序的可靠性。对用户进行了客户满意度和付费意愿调查。小规模试点表明，微交通可以改善郊区的公共交通，更方便，减少在车站等待的时间。特别是，设计过程的结果导致了一种半灵活的服务，90%的常住人口可以使用，76%的乘车等待时间不到15分钟，低于目前公共汽车用户的等待时间（20分钟）。

{"title":"Designing and testing microtransit routes to improve social inclusion: A pilot study in a suburban area","authors":"Alessandro Emilio Capodici , Martina Citrano , Gabriele D'Orso , Marco Migliore , Leonardo Minaudo , Riccardo D'Angelo","doi":"10.1016/j.compenvurbsys.2025.102394","DOIUrl":"10.1016/j.compenvurbsys.2025.102394","url":null,"abstract":"<div><div>Suburbs are often characterized by a scarcity of mobility options to access services. Introducing microtransit is a promising way to improve public transport in suburbs, ensuring greater social inclusion and connecting isolated areas to main transit hubs. The paper aims to develop a multi-step GIS-based methodology for designing semi-flexible stop-based microtransit, having fixed routes and flexible routes (detours) and operating with real-time ride bookings (zero lead time). We considered a suburban area in Palermo, Italy, as study area. The identification of fixed and flexible routes was based on the forecasted passenger flows, through the estimate and the assignment of the daily origin-destination matrix for microtransit, also considering safety, spatial, and technical constraints. A small-scale pilot was carried out between November and December 2022 to test microtransit routes and the reliability of a mobile application to operate the service. A customer satisfaction and a willingness-to-pay survey were addressed to the users. The small-scale pilot showed that microtransit could improve public transportation in suburbs, being more accessible and reducing waiting times at stops. Particularly, the result of the design process led to a semi-flexible service accessible by 90 % of the resident population and with waiting times of less than 15 min in 76 % of the rides, lower than those currently experienced by bus users (20 min).</div></div>","PeriodicalId":48241,"journal":{"name":"Computers Environment and Urban Systems","volume":"125 ","pages":"Article 102394"},"PeriodicalIF":8.3,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145841518","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Geodemographics and residential differentiation: A methodological review and future directions for learned representations of the social landscape 地理人口统计学与居住差异：社会景观表征的方法论回顾与未来方向

IF 8.3 1区地球科学 Q1 ENVIRONMENTAL STUDIES

Computers Environment and Urban Systems

Pub Date : 2026-04-01 Epub Date: 2026-01-07 DOI: 10.1016/j.compenvurbsys.2025.102396

Alex Singleton , Seth E. Spielman

Residential differentiation reflects the complex patterns by which social groups distribute themselves across urban spaces, fundamentally shaping social, economic, and spatial structures. This paper reviews the methodological development of geodemographic classification, tracing its evolution from early social area analysis and factorial ecology through to contemporary approaches. We critically evaluate this lineage of methods for quantifying residential patterns, and identifying persistent limitations in capturing the non-linear complexities of contemporary urban environments. Building on this review, we explore potential future directions involving learned representations of the social landscape, which may offer alternatives to traditional linear dimensionality reduction techniques. Drawing on recent empirical work applying deep learning to geodemographic classification, we consider how such approaches might address identified limitations while acknowledging that their advantages over established methods remain context-dependent and require further empirical validation. We emphasise that any adoption of these techniques must prioritise transparency and interpretability. The paper concludes by outlining potential directions for future research, including how learned representations might be integrated within existing geodemographic workflows.

居住差异反映了社会群体在城市空间中分布的复杂模式，从根本上塑造了社会、经济和空间结构。本文回顾了地理人口分类的方法论发展，追溯了其从早期的社会区域分析和因子生态学到当代方法的演变。我们批判性地评估了这一系列量化居住模式的方法，并确定了在捕捉当代城市环境的非线性复杂性方面的持续局限性。在此综述的基础上，我们探索了涉及社会景观学习表征的潜在未来方向，这可能为传统的线性降维技术提供替代方案。根据最近将深度学习应用于地理人口分类的实证研究，我们考虑了这些方法如何解决已确定的局限性，同时承认它们相对于现有方法的优势仍然依赖于上下文，需要进一步的实证验证。我们强调，采用这些技术必须优先考虑透明度和可解释性。论文最后概述了未来研究的潜在方向，包括如何将学习到的表征整合到现有的地理人口工作流程中。

{"title":"Geodemographics and residential differentiation: A methodological review and future directions for learned representations of the social landscape","authors":"Alex Singleton , Seth E. Spielman","doi":"10.1016/j.compenvurbsys.2025.102396","DOIUrl":"10.1016/j.compenvurbsys.2025.102396","url":null,"abstract":"<div><div>Residential differentiation reflects the complex patterns by which social groups distribute themselves across urban spaces, fundamentally shaping social, economic, and spatial structures. This paper reviews the methodological development of geodemographic classification, tracing its evolution from early social area analysis and factorial ecology through to contemporary approaches. We critically evaluate this lineage of methods for quantifying residential patterns, and identifying persistent limitations in capturing the non-linear complexities of contemporary urban environments. Building on this review, we explore potential future directions involving learned representations of the social landscape, which may offer alternatives to traditional linear dimensionality reduction techniques. Drawing on recent empirical work applying deep learning to geodemographic classification, we consider how such approaches might address identified limitations while acknowledging that their advantages over established methods remain context-dependent and require further empirical validation. We emphasise that any adoption of these techniques must prioritise transparency and interpretability. The paper concludes by outlining potential directions for future research, including how learned representations might be integrated within existing geodemographic workflows.</div></div>","PeriodicalId":48241,"journal":{"name":"Computers Environment and Urban Systems","volume":"125 ","pages":"Article 102396"},"PeriodicalIF":8.3,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145926313","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Estimating road speed classes: Integrating OpenStreetMap and Street View imagery for missing data imputation 估计道路速度等级：整合OpenStreetMap和街景图像用于缺失数据的输入

IF 8.3 1区地球科学 Q1 ENVIRONMENTAL STUDIES

Computers Environment and Urban Systems

Pub Date : 2026-04-01 Epub Date: 2025-12-17 DOI: 10.1016/j.compenvurbsys.2025.102392

Shiyu Tang , Sukanya Randhawa , Jin Rui , Christina Ludwig , Steffen Knoblauch , Charles Hatfield , Alexander Zipf

Traffic speed is a significant indicator for evaluating road network performance and supporting intelligent transportation systems, as it informs congestion management, routing, and operational decisions. Although traffic information is available from commercial platforms and sensor-based monitoring systems, such data are often costly, proprietary, or spatially limited, which restricts their broader usability. To overcome these limitations, we designed a spatial prediction model based on the Graph Sample and Aggregation (GraphSAGE) to infer traffic speeds in unobserved areas. Instead of predicting continuous speed values, we classified traffic into speed classes, which enhanced model robustness in the absence of historical observations and better reflected long-term typical traffic patterns relevant to downstream applications such as routing, emission assessment, and traffic management. Taking Berlin as a case study, the model incorporated multi-source features, including topological features, OpenStreetMap-based road features, and semantic Street View imagery indicators. Uber Movement average speed data were used as supervised learning labels. Results showed that the multi-source feature fusion improved the prediction performance, with the F1 score increasing from 0.6228 to 0.6917. Feature analysis revealed that OSM contextual features contributed the most under limited label coverage, while Street View imagery added complementary information to facilitate model discrimination. Despite only 28 % of road segments being covered by Uber observations, similar feature patterns between labeled and unlabeled areas enabled the model to generalize and infer missing speed data citywide. The framework makes scalable and low-cost speed class inference available for urban traffic monitoring and modeling.

交通速度是评估道路网络性能和支持智能交通系统的重要指标，因为它为拥堵管理、路线和运营决策提供了信息。虽然交通信息可以从商业平台和基于传感器的监测系统中获得，但这些数据通常价格昂贵、专有或空间有限，这限制了它们的广泛可用性。为了克服这些限制，我们设计了一个基于图样本和聚合（GraphSAGE）的空间预测模型来推断未观测区域的交通速度。我们没有预测连续的速度值，而是将交通划分为速度等级，这增强了模型在缺乏历史观测的情况下的鲁棒性，并更好地反映了与下游应用（如路由、排放评估和交通管理）相关的长期典型交通模式。以柏林为例，该模型结合了多源特征，包括拓扑特征、基于openstreetmap的道路特征和语义街景图像指标。优步运动平均速度数据作为监督学习标签。结果表明，多源特征融合提高了预测性能，F1分数从0.6228提高到0.6917。特征分析表明，在有限的标签覆盖范围下，OSM上下文特征贡献最大，而街景图像添加了补充信息，以促进模型识别。尽管Uber的观测数据只覆盖了28%的路段，但标记区域和未标记区域之间的相似特征模式使该模型能够在全市范围内推广和推断缺失的速度数据。该框架为城市交通监控和建模提供了可扩展和低成本的速度类推断。

{"title":"Estimating road speed classes: Integrating OpenStreetMap and Street View imagery for missing data imputation","authors":"Shiyu Tang , Sukanya Randhawa , Jin Rui , Christina Ludwig , Steffen Knoblauch , Charles Hatfield , Alexander Zipf","doi":"10.1016/j.compenvurbsys.2025.102392","DOIUrl":"10.1016/j.compenvurbsys.2025.102392","url":null,"abstract":"<div><div>Traffic speed is a significant indicator for evaluating road network performance and supporting intelligent transportation systems, as it informs congestion management, routing, and operational decisions. Although traffic information is available from commercial platforms and sensor-based monitoring systems, such data are often costly, proprietary, or spatially limited, which restricts their broader usability. To overcome these limitations, we designed a spatial prediction model based on the Graph Sample and Aggregation (GraphSAGE) to infer traffic speeds in unobserved areas. Instead of predicting continuous speed values, we classified traffic into speed classes, which enhanced model robustness in the absence of historical observations and better reflected long-term typical traffic patterns relevant to downstream applications such as routing, emission assessment, and traffic management. Taking Berlin as a case study, the model incorporated multi-source features, including topological features, OpenStreetMap-based road features, and semantic Street View imagery indicators. Uber Movement average speed data were used as supervised learning labels. Results showed that the multi-source feature fusion improved the prediction performance, with the F1 score increasing from 0.6228 to 0.6917. Feature analysis revealed that OSM contextual features contributed the most under limited label coverage, while Street View imagery added complementary information to facilitate model discrimination. Despite only 28 % of road segments being covered by Uber observations, similar feature patterns between labeled and unlabeled areas enabled the model to generalize and infer missing speed data citywide. The framework makes scalable and low-cost speed class inference available for urban traffic monitoring and modeling.</div></div>","PeriodicalId":48241,"journal":{"name":"Computers Environment and Urban Systems","volume":"125 ","pages":"Article 102392"},"PeriodicalIF":8.3,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145791515","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Impacts of spatial resolution on agent-based transportation simulations with shared autonomous vehicles 空间分辨率对基于智能体的共享自动驾驶交通模拟的影响

IF 8.3 1区地球科学 Q1 ENVIRONMENTAL STUDIES

Computers Environment and Urban Systems

Pub Date : 2026-03-01 Epub Date: 2025-11-27 DOI: 10.1016/j.compenvurbsys.2025.102371

Kentaro Mori , Fatemeh Fakhrmoosavi , Krishna Murthy Gurumurthy , Pedro Camargo , Kara M. Kockelman

Agent-based transportation models have been used to simulate shared autonomous vehicle (SAV) fleet operations, enabling a growing understanding of SAVs' operations, impacts, and opportunities. This paper investigates the issue of spatial resolution, since most studies have been conducted on coarsened networks, with many missing links and with aggregated addresses for trip origins and destinations. This work presents simulation results for dynamic traffic assignment with SAV fleet operations in Austin, Texas, comparing outcomes across two networks and two sets of addresses for trip ends in the region's six counties. The comparison involves the Capital Area Metropolitan Planning Organization's (CAMPO's) planning network with addresses highly aggregated (census block centroids supplemented with business establishment information), versus OpenStreetMap's (OSM's) real network with actual addresses sourced from OpenAddresses. CAMPO's network contains 40.6 % of the OSM lane-miles, while the aggregated address points are highly concentrated in the urban core and represent 23 actual addresses on average. Agent-based simulation results using the POLARIS model suggest that omitting a large share of collector and residential links significantly affects network flows, increasing VMT and VHT along non-expressway arterials by 18.9 % and 10.4 %, respectively, for the case of Austin. By contrast, address aggregation (at least at the level implemented in this study) has little impact on traffic. SAVs benefit from increased network connectivity and alternative routes in the complete network to reduce passenger pickup distances and ridepooling detours, lowering VMT by 10 % per SAV—nearly five times the reduction seen in network-wide VMT—and empty VMT (%eVMT) by 2.5 to 3.5 percentage points.

基于智能体的交通模型已被用于模拟共享自动驾驶汽车（SAV）车队的运营，从而使人们对共享自动驾驶汽车的运营、影响和机遇有了更深入的了解。本文探讨了空间分辨率问题，因为大多数研究都是在粗糙的网络上进行的，这些网络有许多缺失的链接，并且旅行起点和目的地的地址都是聚合的。本研究展示了德克萨斯州奥斯汀市SAV车队运行的动态交通分配模拟结果，比较了该地区六个县的两个网络和两组行程终点地址的结果。比较包括首都地区都市规划组织（CAMPO）的规划网络，其地址高度汇总（人口普查块中心点补充商业机构信息），与OpenStreetMap （OSM）的真实网络，其实际地址来自OpenAddresses。CAMPO的网络包含40.6%的OSM车道里程，而聚合地址点高度集中在城市核心，平均代表23个实际地址。使用POLARIS模型的基于agent的仿真结果表明，省去大量集线器和住宅链路会显著影响网络流量，以奥斯汀为例，非高速公路主干道上的VMT和VHT分别增加了18.9%和10.4%。相比之下，地址聚合（至少在本研究中实现的级别上）对流量的影响很小。sav受益于网络连接的增加和整个网络中的替代路线，以减少乘客接送距离和搭车绕路，每个sav降低10%的行驶里程，几乎是整个网络行驶里程减少的五倍，空车行驶里程（%eVMT）降低2.5至3.5个百分点。

{"title":"Impacts of spatial resolution on agent-based transportation simulations with shared autonomous vehicles","authors":"Kentaro Mori , Fatemeh Fakhrmoosavi , Krishna Murthy Gurumurthy , Pedro Camargo , Kara M. Kockelman","doi":"10.1016/j.compenvurbsys.2025.102371","DOIUrl":"10.1016/j.compenvurbsys.2025.102371","url":null,"abstract":"<div><div>Agent-based transportation models have been used to simulate shared autonomous vehicle (SAV) fleet operations, enabling a growing understanding of SAVs' operations, impacts, and opportunities. This paper investigates the issue of spatial resolution, since most studies have been conducted on coarsened networks, with many missing links and with aggregated addresses for trip origins and destinations. This work presents simulation results for dynamic traffic assignment with SAV fleet operations in Austin, Texas, comparing outcomes across two networks and two sets of addresses for trip ends in the region's six counties. The comparison involves the Capital Area Metropolitan Planning Organization's (CAMPO's) planning network with addresses highly aggregated (census block centroids supplemented with business establishment information), versus OpenStreetMap's (OSM's) real network with actual addresses sourced from OpenAddresses. CAMPO's network contains 40.6 % of the OSM lane-miles, while the aggregated address points are highly concentrated in the urban core and represent 23 actual addresses on average. Agent-based simulation results using the POLARIS model suggest that omitting a large share of collector and residential links significantly affects network flows, increasing VMT and VHT along non-expressway arterials by 18.9 % and 10.4 %, respectively, for the case of Austin. By contrast, address aggregation (at least at the level implemented in this study) has little impact on traffic. SAVs benefit from increased network connectivity and alternative routes in the complete network to reduce passenger pickup distances and ridepooling detours, lowering VMT by 10 % per SAV—nearly five times the reduction seen in network-wide VMT—and empty VMT (%eVMT) by 2.5 to 3.5 percentage points.</div></div>","PeriodicalId":48241,"journal":{"name":"Computers Environment and Urban Systems","volume":"124 ","pages":"Article 102371"},"PeriodicalIF":8.3,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145624519","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Generative AI for spatial regeneration planning: Integrating urban planning theories and ethics 生成式人工智能在空间再生规划中的应用：城市规划理论与伦理的融合

IF 8.3 1区地球科学 Q1 ENVIRONMENTAL STUDIES

Computers Environment and Urban Systems

Pub Date : 2026-03-01 Epub Date: 2025-11-24 DOI: 10.1016/j.compenvurbsys.2025.102380

Yan Wang , Yanjie Fu , Ward Lyles

Generative Artificial Intelligence (GenAI) is rapidly emerging as a promising tool for urban planning, particularly spatial regeneration planning (SRP) aimed at ongoing redevelopment in urban areas. Driven by increasing availability of urban spatial data and a strong interest in technological innovation, GenAI models can generate diverse, context-sensitive planning scenarios beyond the limits of conventional approaches. However, the application of GenAI to SRP also raises pressing concerns: Are planning problems appropriately defined and rigorously modeled? How closely do model design and data processing align with established planning theories and ethics? What strategies can mitigate the risk of perpetuating spatial disadvantages embedded in historical data? How do we meaningfully evaluate GenAI's real-world impact? And how can GenAI be governed to ensure equity, transparency, and meaningful collaboration among all planning stakeholders? This Review critically assesses the intersection of SRP theories and GenAI, identifying key vulnerabilities along the modeling process, from problem formulation and data selection to model training, evaluation, and governance. We propose key technical pathways for developing GenAI-based SRP that are grounded in planning theories and ethics, aiming to advance both the rigor and societal relevance of spatial planning research for sustainable, smart, and resilient cities.

生成式人工智能（GenAI）正迅速成为城市规划的一个有前途的工具，特别是针对城市地区正在进行的重建的空间再生规划（SRP）。在城市空间数据可用性不断提高和对技术创新的强烈兴趣的推动下，GenAI模型可以产生超越传统方法限制的多样化、环境敏感的规划情景。然而，GenAI在SRP中的应用也引起了迫切的关注：规划问题是否得到了适当的定义和严格的建模？模型设计和数据处理与已建立的规划理论和伦理的一致性有多紧密？哪些策略可以减轻历史数据中存在的空间劣势的风险？我们如何有意义地评估GenAI对现实世界的影响？如何治理GenAI以确保所有规划利益相关者之间的公平、透明和有意义的合作？本综述批判性地评估了SRP理论和GenAI的交集，确定了建模过程中的关键漏洞，从问题制定和数据选择到模型训练、评估和治理。我们提出了发展基于genai的SRP的关键技术途径，这些路径以规划理论和伦理为基础，旨在提高可持续、智能和弹性城市空间规划研究的严谨性和社会相关性。

{"title":"Generative AI for spatial regeneration planning: Integrating urban planning theories and ethics","authors":"Yan Wang , Yanjie Fu , Ward Lyles","doi":"10.1016/j.compenvurbsys.2025.102380","DOIUrl":"10.1016/j.compenvurbsys.2025.102380","url":null,"abstract":"<div><div>Generative Artificial Intelligence (GenAI) is rapidly emerging as a promising tool for urban planning, particularly spatial regeneration planning (SRP) aimed at ongoing redevelopment in urban areas. Driven by increasing availability of urban spatial data and a strong interest in technological innovation, GenAI models can generate diverse, context-sensitive planning scenarios beyond the limits of conventional approaches. However, the application of GenAI to SRP also raises pressing concerns: Are planning problems appropriately defined and rigorously modeled? How closely do model design and data processing align with established planning theories and ethics? What strategies can mitigate the risk of perpetuating spatial disadvantages embedded in historical data? How do we meaningfully evaluate GenAI's real-world impact? And how can GenAI be governed to ensure equity, transparency, and meaningful collaboration among all planning stakeholders? This Review critically assesses the intersection of SRP theories and GenAI, identifying key vulnerabilities along the modeling process, from problem formulation and data selection to model training, evaluation, and governance. We propose key technical pathways for developing GenAI-based SRP that are grounded in planning theories and ethics, aiming to advance both the rigor and societal relevance of spatial planning research for sustainable, smart, and resilient cities.</div></div>","PeriodicalId":48241,"journal":{"name":"Computers Environment and Urban Systems","volume":"124 ","pages":"Article 102380"},"PeriodicalIF":8.3,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145624521","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Leveraging sidewalk robots for walkability-related analyses 利用人行道机器人进行与步行相关的分析

IF 8.3 1区地球科学 Q1 ENVIRONMENTAL STUDIES

Computers Environment and Urban Systems

Pub Date : 2026-03-01 Epub Date: 2025-11-24 DOI: 10.1016/j.compenvurbsys.2025.102381

Xing Tong , Michele D. Simoni , Kaj Munhoz Arfvidsson , Jonas Mårtensson

Walkability is a key component of sustainable urban development. In walkability studies, collecting detailed pedestrian infrastructure data remains challenging due to the high costs and limited scalability of traditional methods. Sidewalk delivery robots, increasingly deployed in urban environments, offer a promising solution to these limitations. This paper explores how these robots can serve as mobile data collection platforms, capturing sidewalk-level features related to walkability in a scalable, automated, and real-time manner. A sensor-equipped robot was deployed on a sidewalk network at KTH in Stockholm, completing 101 trips covering 900 segment records. From the collected data, different typologies of features are derived, including robot trip characteristics (e.g., speed, duration), sidewalk conditions (e.g., width, surface unevenness), and sidewalk utilization (e.g., pedestrian density). Their walkability-related implications were investigated with a series of analyses. The results demonstrate that pedestrian movement patterns are strongly influenced by sidewalk characteristics, with higher density, reduced width, and surface irregularity associated with slower and more variable trajectories. Notably, robot speed closely mirrors pedestrian behavior, highlighting its potential as a proxy for assessing pedestrian dynamics. The proposed framework enables continuous monitoring of sidewalk conditions and pedestrian behavior, contributing to the development of more walkable, inclusive, and responsive urban environments.

可步行性是可持续城市发展的关键组成部分。在可步行性研究中，由于传统方法的高成本和有限的可扩展性，收集详细的行人基础设施数据仍然具有挑战性。人行道送货机器人越来越多地部署在城市环境中，为这些限制提供了一个有希望的解决方案。本文探讨了这些机器人如何作为移动数据收集平台，以可扩展、自动化和实时的方式捕获与步行性相关的人行道级特征。配备传感器的机器人被部署在斯德哥尔摩KTH的人行道网络上，完成了101次行程，覆盖了900个路段记录。从收集到的数据中，可以得出不同类型的特征，包括机器人的行程特征（如速度、持续时间）、人行道条件（如宽度、表面不平度）和人行道利用率（如行人密度）。通过一系列分析调查了它们与步行相关的影响。结果表明，行人的运动模式受到人行道特征的强烈影响，人行道的密度增大、宽度减小、表面不平整与更慢、更可变的轨迹相关。值得注意的是，机器人的速度密切反映了行人的行为，突出了它作为评估行人动态的代理的潜力。拟议的框架能够持续监测人行道状况和行人行为，有助于发展更适合步行、包容和响应的城市环境。

{"title":"Leveraging sidewalk robots for walkability-related analyses","authors":"Xing Tong , Michele D. Simoni , Kaj Munhoz Arfvidsson , Jonas Mårtensson","doi":"10.1016/j.compenvurbsys.2025.102381","DOIUrl":"10.1016/j.compenvurbsys.2025.102381","url":null,"abstract":"<div><div>Walkability is a key component of sustainable urban development. In walkability studies, collecting detailed pedestrian infrastructure data remains challenging due to the high costs and limited scalability of traditional methods. Sidewalk delivery robots, increasingly deployed in urban environments, offer a promising solution to these limitations. This paper explores how these robots can serve as mobile data collection platforms, capturing sidewalk-level features related to walkability in a scalable, automated, and real-time manner. A sensor-equipped robot was deployed on a sidewalk network at KTH in Stockholm, completing 101 trips covering 900 segment records. From the collected data, different typologies of features are derived, including robot trip characteristics (e.g., speed, duration), sidewalk conditions (e.g., width, surface unevenness), and sidewalk utilization (e.g., pedestrian density). Their walkability-related implications were investigated with a series of analyses. The results demonstrate that pedestrian movement patterns are strongly influenced by sidewalk characteristics, with higher density, reduced width, and surface irregularity associated with slower and more variable trajectories. Notably, robot speed closely mirrors pedestrian behavior, highlighting its potential as a proxy for assessing pedestrian dynamics. The proposed framework enables continuous monitoring of sidewalk conditions and pedestrian behavior, contributing to the development of more walkable, inclusive, and responsive urban environments.</div></div>","PeriodicalId":48241,"journal":{"name":"Computers Environment and Urban Systems","volume":"124 ","pages":"Article 102381"},"PeriodicalIF":8.3,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145624520","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Detecting uneven impacts of rental housing financialisation and platformisation on neighbourhoods in China: A multi-source data mining approach 中国租赁住房金融化和平台化对社区的不均衡影响：一种多源数据挖掘方法

IF 8.3 1区地球科学 Q1 ENVIRONMENTAL STUDIES

Computers Environment and Urban Systems

Pub Date : 2026-03-01 Epub Date: 2025-11-15 DOI: 10.1016/j.compenvurbsys.2025.102369

Zixin Luo , Mengzhu Zhang , Dongwei Liu , Juan Li , Anthony G.O. Yeh

The rapid financialisation and platformisation of rental housing (FPR) have been examined as a major factor for neighbourhood changes worldwide after the 2010s. Studies have revealed the appreciation and disamenity effects with which the financialisation and platformisation reshape neighbourhoods, whereas the uneven impacts of the variegated FPR on neighbourhoods remain insufficiently understood. Therefore, this study fills this gap by the evidence from 6645 neighbourhoods in five Chinese cities to detect the uneven impacts of FPR on neighbourhoods and excavate those vulnerable communities. Leveraging geographically weighted regression, computer vision analysis on street view images and SOFM-based clustering analysis, this study reveals that (1) the appreciation effects widely spread across the cities, particularly in the new urban areas in Suzhou, the suburban areas in Shenzhen and the central city in Shanghai. Conversely, the disamenity effects are more severe in suburban areas in Shenzhen and Beijing than other cities due to the large number of urban villages and dilapidated buildings there; (2) type-C neighbourhoods featuring old low-rise buildings with local elderly populations is least affected by FPR (both appreciation and disamenity). By contrast, type-D neighbourhoods encompassing a large number of migrant workers in the suburbs are most susceptible to FPR. This study introduces the housing market segmentation theory to understand FPR'S effect on neighbourhood segregation and provides policy implications for niche and means to mitigate the negative externalities of FPR

租赁住房（FPR）的快速金融化和平台化已被视为2010年代后全球社区变化的主要因素。研究揭示了金融化和平台化重塑社区的升值和破坏效应，而多样化的FPR对社区的不平衡影响仍未得到充分理解。因此，本研究通过中国5个城市6645个社区的证据来填补这一空白，发现FPR对社区的不均衡影响，挖掘弱势社区。利用地理加权回归、街景图像的计算机视觉分析和基于sofm的聚类分析，研究发现(1)升值效应在城市间广泛分布，特别是在苏州新城区、深圳近郊和上海中心城区。相反，深圳和北京的城郊由于城中村和残破建筑较多，其残废效应较其他城市更为严重；(2)以老旧低层建筑为主的c类社区受FPR的影响最小（无论是欣赏还是厌恶）。相比之下，包含大量郊区外来务工人员的d型社区最容易发生FPR。本研究引入住房市场分割理论来理解住房市场分割对邻里隔离的影响，并提供政策启示和减轻住房市场分割的负外部性的方法

{"title":"Detecting uneven impacts of rental housing financialisation and platformisation on neighbourhoods in China: A multi-source data mining approach","authors":"Zixin Luo , Mengzhu Zhang , Dongwei Liu , Juan Li , Anthony G.O. Yeh","doi":"10.1016/j.compenvurbsys.2025.102369","DOIUrl":"10.1016/j.compenvurbsys.2025.102369","url":null,"abstract":"<div><div>The rapid financialisation and platformisation of rental housing (FPR) have been examined as a major factor for neighbourhood changes worldwide after the 2010s. Studies have revealed the appreciation and disamenity effects with which the financialisation and platformisation reshape neighbourhoods, whereas the uneven impacts of the variegated FPR on neighbourhoods remain insufficiently understood. Therefore, this study fills this gap by the evidence from 6645 neighbourhoods in five Chinese cities to detect the uneven impacts of FPR on neighbourhoods and excavate those vulnerable communities. Leveraging geographically weighted regression, computer vision analysis on street view images and SOFM-based clustering analysis, this study reveals that (1) the appreciation effects widely spread across the cities, particularly in the new urban areas in Suzhou, the suburban areas in Shenzhen and the central city in Shanghai. Conversely, the disamenity effects are more severe in suburban areas in Shenzhen and Beijing than other cities due to the large number of urban villages and dilapidated buildings there; (2) type-C neighbourhoods featuring old low-rise buildings with local elderly populations is least affected by FPR (both appreciation and disamenity). By contrast, type-D neighbourhoods encompassing a large number of migrant workers in the suburbs are most susceptible to FPR. This study introduces the housing market segmentation theory to understand FPR'S effect on neighbourhood segregation and provides policy implications for niche and means to mitigate the negative externalities of FPR</div></div>","PeriodicalId":48241,"journal":{"name":"Computers Environment and Urban Systems","volume":"124 ","pages":"Article 102369"},"PeriodicalIF":8.3,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145528120","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A critical assessment of selected urban resilience decision-support tools in the United States 对美国选定的城市弹性决策支持工具的关键评估

IF 8.3 1区地球科学 Q1 ENVIRONMENTAL STUDIES

Computers Environment and Urban Systems

Pub Date : 2026-01-01 Epub Date: 2025-10-22 DOI: 10.1016/j.compenvurbsys.2025.102352

Sarbeswar Praharaj

The concept of resilience has gained increasing prominence in research and policy discussions. Various web-based data visualization tools have been developed to communicate hazard risks and make complex climate models more accessible to the community. This article presents a comprehensive assessment of 30 selected resilience tools across the U.S. to examine their decision-support capabilities. We conducted an extensive search to shortlist tools from diverse sectors, including public, private, non-profit, and academic organizations, addressing different hazard contexts, such as floods, heatwaves, wildfires, storms, infrastructure failures, and land degradation. We designed and applied a broad-based analytical framework to assess the tools' performance, encompassing the comprehensiveness of indicator systems, robustness of design and visual communication, scenario planning to handle uncertainties, participatory and interactive approaches, and decision-support and action-enabling characteristics. Several notable findings emerge from this study: (a) resilience tools overwhelmingly include environmental datasets while overlooking social and institutional dimensions, necessary to capture the complexity of urban systems; (b) minimal integration of city and neighborhood level information poses barriers to localized assessments and adaptation planning; (c) while the tools effectively communicate vulnerability, future innovations in scenario-based planning utilizing AI, real-time data, and predictive modeling capabilities are needed to plan for a dynamic and uncertain future; and (d) these tools must foster an interoperable design for users to combine new data layers to adjust their resilience assessments to local contexts. This study outlines future research and development avenues, providing resilience planners and decision-makers with a strategic foresight to enhance computer-based applications for assessing, planning, and managing disaster risks to human and urban systems.

弹性的概念在研究和政策讨论中越来越突出。已经开发了各种基于网络的数据可视化工具，以传达灾害风险，并使社区更容易获得复杂的气候模型。本文对美国30个选定的弹性工具进行了全面评估，以检查其决策支持能力。我们进行了广泛的搜索，从不同部门（包括公共、私人、非营利和学术组织）获得了候选工具，以应对不同的灾害背景，如洪水、热浪、野火、风暴、基础设施故障和土地退化。我们设计并应用了一个基础广泛的分析框架来评估工具的性能，包括指标体系的全面性、设计和视觉传达的稳健性、处理不确定性的情景规划、参与性和互动性方法、决策支持和行动支持特征。本研究得出了几个值得注意的发现：(a)复原力工具绝大多数包括环境数据集，而忽略了社会和制度层面，这是捕捉城市系统复杂性所必需的；(b)城市和社区一级信息的最低限度整合对本地化评估和适应性规划构成障碍；(c)虽然这些工具有效地传达了脆弱性，但未来需要利用人工智能、实时数据和预测建模能力进行基于场景的规划创新，以规划动态和不确定的未来；(d)这些工具必须培养一种可互操作的设计，以便用户结合新的数据层，根据当地情况调整其弹性评估。本研究概述了未来的研究和发展途径，为弹性规划者和决策者提供战略远见，以加强基于计算机的应用，以评估、规划和管理人类和城市系统面临的灾害风险。

{"title":"A critical assessment of selected urban resilience decision-support tools in the United States","authors":"Sarbeswar Praharaj","doi":"10.1016/j.compenvurbsys.2025.102352","DOIUrl":"10.1016/j.compenvurbsys.2025.102352","url":null,"abstract":"<div><div>The concept of resilience has gained increasing prominence in research and policy discussions. Various web-based data visualization tools have been developed to communicate hazard risks and make complex climate models more accessible to the community. This article presents a comprehensive assessment of 30 selected resilience tools across the U.S. to examine their decision-support capabilities. We conducted an extensive search to shortlist tools from diverse sectors, including public, private, non-profit, and academic organizations, addressing different hazard contexts, such as floods, heatwaves, wildfires, storms, infrastructure failures, and land degradation. We designed and applied a broad-based analytical framework to assess the tools' performance, encompassing the comprehensiveness of indicator systems, robustness of design and visual communication, scenario planning to handle uncertainties, participatory and interactive approaches, and decision-support and action-enabling characteristics. Several notable findings emerge from this study: (a) resilience tools overwhelmingly include environmental datasets while overlooking social and institutional dimensions, necessary to capture the complexity of urban systems; (b) minimal integration of city and neighborhood level information poses barriers to localized assessments and adaptation planning; (c) while the tools effectively communicate vulnerability, future innovations in scenario-based planning utilizing AI, real-time data, and predictive modeling capabilities are needed to plan for a dynamic and uncertain future; and (d) these tools must foster an interoperable design for users to combine new data layers to adjust their resilience assessments to local contexts. This study outlines future research and development avenues, providing resilience planners and decision-makers with a strategic foresight to enhance computer-based applications for assessing, planning, and managing disaster risks to human and urban systems.</div></div>","PeriodicalId":48241,"journal":{"name":"Computers Environment and Urban Systems","volume":"123 ","pages":"Article 102352"},"PeriodicalIF":8.3,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145365250","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A hierarchical cross-departmental agent-based approach to explore the impacts of policy interplay on land use dynamics 以分层跨部门代理为基础的方法探讨政策相互作用对土地使用动态的影响

IF 8.3 1区地球科学 Q1 ENVIRONMENTAL STUDIES

Computers Environment and Urban Systems

Pub Date : 2026-01-01 Epub Date: 2025-10-14 DOI: 10.1016/j.compenvurbsys.2025.102355

Jing Gao , Jian Gong , Nina Schwarz , Richard Sliuzas

Understanding the interactions between competing land policies is crucial for identifying governance challenges and assisting urban planners and policy analysts in making informed decisions. However, a methodology for incorporating land use patterns and the policy implementation processes within the framework of hierarchical land management remains underexplored. Here, we employ an agent-based model (ABM) to investigate how land use change occurs as policies intersect across different hierarchical levels and branches of government in Wuhan, China. Changes in land use arise from the interplay between five agents—the central level, the local level that incorporates three departments, and the village collective level—in the decisions on land acquisition, conversion, and reclamation. Four parameters characterize the enforcement levels of relevant policies, and multi-objective optimization with genetic algorithms was applied to calibrate them. The results show that: (1) Our ABM exhibits a figure of merit value of 0.3 at the city level and 0.58 in the larger urban area, indicating its capability to simulate real land use dynamics. (2) Policy implementation gaps led to high land conversion and low farmland reclamation. (3) The dynamic enforcement scenarios provide a viable pathway for negotiated governance, enabling demand-responsive rate attenuation and conflict mitigation, which is distinct from the exacerbated land use conflicts observed under the other scenarios. (4) Policy should incorporate adaptive mechanisms to maintain a buffer between competing land demands rather than binary constraints. This ABM introduces a novel hierarchical framework to decode policy interplay and implementation tensions, advancing sustainable land governance and urban planning insights.

了解相互竞争的土地政策之间的相互作用对于确定治理挑战和协助城市规划者和政策分析师做出明智决策至关重要。但是，在分级土地管理框架内结合土地使用模式和政策执行过程的方法仍未得到充分探讨。在此，我们采用基于主体的模型（ABM）来研究中国武汉不同层级和政府部门之间政策交叉时土地利用变化的发生情况。土地利用的变化源于五个主体——中央层面、由三个部门组成的地方层面和村集体层面——在土地征用、转换和复垦决策方面的相互作用。采用遗传算法进行多目标优化，对相关政策的执行水平进行了标定。结果表明：(1)ABM模型在城市层面的优值为0.3，在较大城市区域的优值为0.58，表明ABM模型具有模拟真实土地利用动态的能力。(2)政策执行缺口导致土地流失率高，耕地复垦率低。(3)动态执行情景为协商治理提供了可行途径，使需求响应率衰减和冲突缓解成为可能，这与其他情景下观察到的土地利用冲突加剧有所不同。(4)政策应纳入适应性机制，在相互竞争的土地需求之间保持缓冲，而不是二元约束。该ABM引入了一个新的分层框架来解读政策相互作用和实施紧张关系，促进可持续土地治理和城市规划见解。

{"title":"A hierarchical cross-departmental agent-based approach to explore the impacts of policy interplay on land use dynamics","authors":"Jing Gao , Jian Gong , Nina Schwarz , Richard Sliuzas","doi":"10.1016/j.compenvurbsys.2025.102355","DOIUrl":"10.1016/j.compenvurbsys.2025.102355","url":null,"abstract":"<div><div>Understanding the interactions between competing land policies is crucial for identifying governance challenges and assisting urban planners and policy analysts in making informed decisions. However, a methodology for incorporating land use patterns and the policy implementation processes within the framework of hierarchical land management remains underexplored. Here, we employ an agent-based model (ABM) to investigate how land use change occurs as policies intersect across different hierarchical levels and branches of government in Wuhan, China. Changes in land use arise from the interplay between five agents—the central level, the local level that incorporates three departments, and the village collective level—in the decisions on land acquisition, conversion, and reclamation. Four parameters characterize the enforcement levels of relevant policies, and multi-objective optimization with genetic algorithms was applied to calibrate them. The results show that: (1) Our ABM exhibits a figure of merit value of 0.3 at the city level and 0.58 in the larger urban area, indicating its capability to simulate real land use dynamics. (2) Policy implementation gaps led to high land conversion and low farmland reclamation. (3) The dynamic enforcement scenarios provide a viable pathway for negotiated governance, enabling demand-responsive rate attenuation and conflict mitigation, which is distinct from the exacerbated land use conflicts observed under the other scenarios. (4) Policy should incorporate adaptive mechanisms to maintain a buffer between competing land demands rather than binary constraints. This ABM introduces a novel hierarchical framework to decode policy interplay and implementation tensions, advancing sustainable land governance and urban planning insights.</div></div>","PeriodicalId":48241,"journal":{"name":"Computers Environment and Urban Systems","volume":"123 ","pages":"Article 102355"},"PeriodicalIF":8.3,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145322867","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

City identity recognition: how representation bias influences model predictability and replicability? 城市身份识别：表征偏差如何影响模型的可预测性和可复制性？

IF 8.3 1区地球科学 Q1 ENVIRONMENTAL STUDIES

Computers Environment and Urban Systems

Pub Date : 2026-01-01 Epub Date: 2025-11-11 DOI: 10.1016/j.compenvurbsys.2025.102370

Xiang Zhang , Fan Yang , Zongze He , Weijia Li , Min Yang

Street-view images (SVIs) and social media photos have been widely used with deep learning in urban perception (e.g., predicting socioeconomic variables, recognizing city identities, and geolocalization). However, different data sources and sampling schemes are arbitrarily used in existing studies, and the induced representation bias that can drastically alter model predictions has long been overlooked. To bridge this gap, we systematically evaluate how different biases influence model predictability and replicability using city identity recognition (CIR), a task to understand the unique characters of cities and to recognize their identities from photos. The task was implemented by firstly extracting features with a model pretrained on a general scene recognition task, and secondly training city classifiers with the extracted features. We answer the research question by carefully designed experiments, where factors like intrinsic similarity between cities, data sources, camera perspectives, and spatial sampling are examined. In general, we show that recognizing cities from photos was not equally replicable across the world: intra-country CIR was more challenging than inter-country CIR. Contrast to common wisdom, the claimed advantages of social media photos in indoor and social scenarios were not effective on CIR and were largely surpassed by SVIs. For SVIs, perspectives along streets surprisingly better capture the uniqueness of a city than looking at building facades; biased spatial sampling can even overturn model predictions (e.g., rankings of city uniqueness). Finally, we explore the data diversity by projecting the learned representations into a low-dimensional semantic space, and critically discuss our results and implications.

街景图像（SVIs）和社交媒体照片已被广泛用于城市感知的深度学习（例如，预测社会经济变量，识别城市身份和地理定位）。然而，在现有的研究中，不同的数据来源和抽样方案被任意使用，而诱导的表征偏差可能会极大地改变模型预测，长期以来一直被忽视。为了弥补这一差距，我们使用城市身份识别（CIR）系统地评估了不同的偏见如何影响模型的可预测性和可复制性，CIR是一项了解城市独特特征并从照片中识别城市身份的任务。该任务首先通过在一般场景识别任务上预训练的模型提取特征，然后利用提取的特征训练城市分类器来实现。我们通过精心设计的实验来回答研究问题，其中包括城市之间的内在相似性、数据源、相机视角和空间采样等因素。总的来说，我们发现从照片中识别城市在世界范围内的可复制性并不相同：国家内部的CIR比国家之间的CIR更具挑战性。与普遍观点相反，社交媒体照片在室内和社交场景中的优势在CIR上并不有效，并且在很大程度上被svi所超越。对于svi来说，沿着街道的视角比看建筑立面更能捕捉城市的独特性；有偏差的空间抽样甚至可以推翻模型预测（例如，城市独特性排名）。最后，我们通过将学习到的表示投射到低维语义空间来探索数据多样性，并批判性地讨论我们的结果和意义。

{"title":"City identity recognition: how representation bias influences model predictability and replicability?","authors":"Xiang Zhang , Fan Yang , Zongze He , Weijia Li , Min Yang","doi":"10.1016/j.compenvurbsys.2025.102370","DOIUrl":"10.1016/j.compenvurbsys.2025.102370","url":null,"abstract":"<div><div>Street-view images (SVIs) and social media photos have been widely used with deep learning in urban perception (e.g., predicting socioeconomic variables, recognizing city identities, and geolocalization). However, different data sources and sampling schemes are arbitrarily used in existing studies, and the induced representation bias that can drastically alter model predictions has long been overlooked. To bridge this gap, we systematically evaluate how different biases influence model predictability and replicability using city identity recognition (CIR), a task to understand the unique characters of cities and to recognize their identities from photos. The task was implemented by firstly extracting features with a model pretrained on a general scene recognition task, and secondly training city classifiers with the extracted features. We answer the research question by carefully designed experiments, where factors like intrinsic similarity between cities, data sources, camera perspectives, and spatial sampling are examined. In general, we show that recognizing cities from photos was not equally replicable across the world: intra-country CIR was more challenging than inter-country CIR. Contrast to common wisdom, the claimed advantages of social media photos in indoor and social scenarios were not effective on CIR and were largely surpassed by SVIs. For SVIs, perspectives along streets surprisingly better capture the uniqueness of a city than looking at building facades; biased spatial sampling can even overturn model predictions (e.g., rankings of city uniqueness). Finally, we explore the data diversity by projecting the learned representations into a low-dimensional semantic space, and critically discuss our results and implications.</div></div>","PeriodicalId":48241,"journal":{"name":"Computers Environment and Urban Systems","volume":"123 ","pages":"Article 102370"},"PeriodicalIF":8.3,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145520364","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0