The proposed framework consists of 2 interrelated components: structural foundations and a 3-layer coupled mechanism. The structural foundations include 3 parts. First, behavioral subjects are no longer defined solely by conventional socio-economic attributes, but are increasingly differentiated by digital literacy, platform experience, access to smart devices, and the ability to understand and respond to algorithmic logic. Second, digital-intelligent intermediaries, including platform interfaces, navigation tools, and recommendation algorithms and etc., actively reshape information visibility, behavioral opportunities, and decision structures. Third, the spatiotemporal context of behavior has shifted to a hybrid and computable environment in which virtual and physical spaces are intertwined and continuously sensed, recorded, and modeled. Together, these elements define the contextual foundation within which individual behaviors are generated in the digital-intelligent era. Building upon this foundation, the framework further identifies a 3-layer coupled mechanism: the real-world behavioral generation layer, the digital twin simulation and expansion layer, and the behavioral regulation and feedback layer. The first layer focuses on how individual behaviors are generated in real-world hybrid environments through cognition, decision-making, and action across pre-behavior, in-behavior, and post-behavior stages. The second layer emphasizes how real-world behaviors are transformed into computable and modelable representations through multi-source behavioral data and digital twin technologies, making scenario simulation and behavioral prediction possible. The third layer highlights how simulation results feed back into urban planning, governance, and behavioral guidance, while also continuously updating the twin system through new data inputs. In this sense, individual space-time behavior is no longer understood merely as a response to pre-existing spatial constraints, but as part of a dynamic loop linking behavioral generation, datafication, simulation, intervention, and renewed behavioral adaptation. The proposed framework thus contributes not only to advancing the theoretical development of behavioral geography under the digital-intelligent turn, but also to providing a conceptual basis for linking behavioral analysis with planning and governance practices in increasingly data-rich and algorithmically mediated urban environments.

Grounded in the social-technological-ecological transition of the digital intelligence era, this paper systematically reviews the theoretical evolution trajectory of Time Geography along the path of “classical foundation—theoretical innovation—digital intelligence leap”. It traces the theoretical development of key concepts from diorama to project and pockets of local orders, and articulates three ontological shifts of Time Geography. The new Time Geography witnesses an ontological shift: the view of space-time transforms from absolute space-time to relational space-time; the view of behavior shifts from space-time path description to project-driven activity system, connecting intention with action and introducing the social dimension; and the view of human-ecological interaction expands from the fabric of existence woven by individual paths to the all-ecological analysis of the project-activity system. Focusing on theoretical and methodological innovations in the digital intelligence era, this paper synthesizes recent conceptual developments—including physical-virtual hybrid paths, physical-digital dual constraints, and ICT-mediated pockets of local orders—and identifies emerging methodological approaches such as dual-activity diary surveys, hybrid physical-virtual measurement, and foreground-background activity visualization. The paper identifies emerging features of human activities in the digital intelligence era: persistent frontstage-background intertwining, space-time decoupling and recoupling, normalization of remote intervention, and diversification of pacesetters. It further argues that algorithms are becoming an endogenous force reshaping projects and their pockets of local orders. The innovative development of Time Geography in the digital intelligence era provides a robust scientific foundation for deeply analyzing the mechanism of digital transformation of daily life, spatial and temporal optimizing the people-oriented city, and improving residents’ well-being.

Since the rise of the behavioral revolution in the 1970s, the research paradigm of human geography has shifted from macro-scale pattern descriptions toward the exploration of micro-scale process mechanisms in human-land relationships. With the increasing availability of long-term spatiotemporal behavioral big data, empirical behavioral geography is transitioning from traditional “space-behavior” interaction analysis to refined process tracking and causal inference design. Despite the accumulation of significant research outcomes, the field still faces two core challenges: first, a lack of long-term dynamic tracking covering the entire interaction chain of “physical space-resident behavior-living space”; and second, a tendency to remain at the level of verifying the significance of correlations rather than establishing robust causal links. Traditional studies often rely on cross-sectional data and assume a static equilibrium between space and behavior, which overlooks the dynamic formation process of stable states and fails to address the fundamental question of how spatial adjustments trigger specific behavioral shifts. To address these limitations, this paper systematically reviews recent research progress and proposes a novel research framework that integrates process tracking with causal inference. This framework focuses on the dynamic interaction between physical space, individual behavior, and life space, aiming to move beyond “black-box” correlations toward the identification of causal effects. Methodologically, the paper emphasizes the use of “big yet thick” data, such as long-term mobile signaling data, to construct natural experiment scenarios, including passive spatial interventions like infrastructure development and active interventions like residential relocation. By introducing quasi-experimental designs, such as Difference-in-Differences, Regression Discontinuity, and Causal Forests, researchers can mitigate common biases like confounding, over-control, and residential self-selection. The research content is structured into 3 progressive components: first, tracking the impact of physical space on behavioral responses, distinguishing between immediate, lagged, gradual, and sudden response patterns; second, uncovering the evolutionary mechanisms of life space formation using advanced techniques such as Topological Data Analysis (TDA) and Inverse Reinforcement Learning to decode individual decision-making preferences; and third, integrating these perspectives to optimize urban planning and emergency response. This integrated approach allows for the scientific evaluation of spatial intervention policies under both normal conditions, such as the assessment of 15-minute community life circles, and emergency states, such as the resilience of urban mobility during public health crises like the COVID-19 pandemic. Looking ahead, the paper advocates for the fusion of multi-source heterogeneous data and the adoption of cutting-edge artificial intelligence, including Large Language Models and AI agents, to enhance the reliability of causal discovery. By clarifying the mechanisms by which space shapes behavior and behavior constructs life space, this paradigm shift provides evidence-based theoretical and methodological support for precision urban planning, sustainable built environment design, and the cultivation of urban resilience, ultimately advancing the study of micro-scale human-land relationships.

With the continuous evolution of digital intelligence technologies, individual behavioral patterns are undergoing profound restructuring. Physical and virtual activities have coalesced into a physical-virtual activity system characterized by mutual permeation and real-time interactive feedback, while the technologically mediated hybrid physical-virtual environment has fundamentally altered the interaction between environment and behavior. This process exerts complex and far-reaching impacts on residents’ mental health by rendering the acquisition of mental health resources and social support more accessible, yet simultaneously engendering novel mental health risks. Against the backdrop of the deep integration of physical-virtual environments and the growing interweaving of physical and virtual activities, it is necessary to re-examine the dynamic evolution of environment, behavior, and mental health, as well as the complex interrelationships among the three. This paper systematically reviews mental health research from the perspective of behavioral geography and summarizes the reconfiguration of behavioral patterns and the corresponding changes in mental health in the age of digital intelligence. It further explores the mechanisms through which individual digital literacy, hybrid physical-virtual environments, and reconfigured behavioral patterns affect mental health under the influence of digital intelligence technologies. On this basis, the paper constructs a new “Environment-Behavior-Mental Health” research framework for the age of digital intelligence. This framework incorporates hybrid physical-virtual environments and physical-virtual activity systems, introduces individual digital literacy and its interaction with the environment to reveal population heterogeneity, characterizes the multi-dimensional features of physical-virtual activity systems from temporal, spatial, and interaction perspectives, and systematically reveals mental health outcomes from risk to well-being and from immediate responses to long-term scales. It also proposes mental-health-oriented strategies for the synergistic optimization of physical and virtual environments, together with behavioral guidance aimed at promoting healthier patterns of physical-virtual activity. Overall, this study provides new insights for environment health research in the context of digital intelligence and deepens the understanding of health mechanisms in the digital intelligence era from a behavioral geography perspective.

The increasingly pervasive application of digital technologies and artificial intelligence has accelerated the arrival of the digital-intelligent era, profoundly reshaping geographical space and behavioral geography. Building upon a comprehensive review of Behavioral Geography research in the digital-intelligent era, this paper conceptualizes how digital intelligence reconfigures spatial attributes, human subjectivity, spatial behavior, and spatial governance. Furthermore, by introducing well-being, this study constructs an analytical framework to decipher the interactive relations among geographical space, affective experience, activity systems, and spatial governance. Based on this framework, it systematically reflects on the shifting relationship among subjective well-being, spatial behavior, and spatial governance in the digital-intelligent age. The research highlights the threefold impact of digital intelligence. First, digitalization can negatively impact human subjective well-being by altering reference points for judgment, modifying sensitivity to environmental experiences, and creating misalignment between algorithmic optimization objectives and individual value systems. Second, it reshapes the “information gathering–value judgment–spatial decision-making” chain by restructuring the set of available spatial choices, reducing the role of value judgment in behavioral decisions, and redefining the very meaning of “spatial” behavior. Third, digitalization alters the drivers of political behavior, diminishes the perceived importance of transforming physical space, and highlights the growing significance of digital spaces design and temporal planning. The paper further proposes recommendations such as strengthening interpretable analytical frameworks, enriching governance approaches, promoting cross-scale integration, and reconstructing fundamental principles in the digital-intelligent era. Furthermore, the paper discusses how digitalization and intelligence reconstruct the connotations and impacts of core concepts in Behavioral Geography—namely “space” and “behavior”. This study expands theoretical innovation in Behavioral Geography for the digital-intelligent era and provides a crucial reference for deepening our understanding of the interrelationships among geographical environments, subjective well-being, and spatial behaviors.

This paper reviews recent advances and extends the “space-time-human” coupling framework to the study of digitally mediated urban environments. Rather than treating digital media as an independent fourth dimension, the framework conceptualizes them as a cross-cutting force that reconfigures how time, space, and individual characteristics interact to shape health and well-being. Theoretically, digital media introduce instantaneous and cumulative exposure as well as temporal compression into the time dimension; extend the space dimension from physical settings to hybrid digital-physical environments; and make digital literacy and individual agency within algorithmic systems central to the human dimension. The paper further proposes 3 interrelated concepts. Digital landscape refers to algorithmically mediated information environments and individuals’ exposure to and perception of them; virtual behavior encompasses information seeking, online social interaction, navigation, and digital health management; and digital health support includes wearable devices, AI-assisted services, telemedicine, and context-aware interventions. Together, these elements form dynamic feedback loops in which digital exposure shapes virtual and physical behaviors, user activities modify subsequent information environments, and health status further influences patterns of engagement. Although these processes can expand access to health resources, they may also reproduce or intensify inequalities through digital divides, algorithmic bias, information overload, and unequal capacities to convert available resources into health benefits. Methodological advances follow 2 interconnected tracks. At the data level, the integration of GPS trajectories, platform and application logs, wearable sensor data, ecological momentary assessment, social media content, and physiological indicators enables more dynamic and context-sensitive measurement of individual exposure, behavior, and health outcomes. At the analytical level, explainable geographic artificial intelligence, causal inference, virtual experiments, digital twins, and agent-based simulations provide tools for identifying mechanisms, inferring causality, and evaluating interventions. In practice, these developments translate into two major application domains. At the individual level, geofenced and personalized health services, early risk detection, remote healthcare, and digital behavior interventions support proactive and precision health management. At the spatial level, mobility-responsive allocation of public health resources, evidence-based environmental design, and adaptive urban governance contribute to building healthy and resilient cities. Future research should move beyond the physical-virtual dichotomy, conduct longitudinal and cross-regional comparative studies across the life course, and integrate explainable artificial intelligence with causal research designs. Greater attention should be paid to privacy, algorithmic transparency, digital inclusion, public accountability, and environmental health justice.

Population mobility constitutes a critical factor in urban-rural integration. Clarifying the mechanisms underpinning urban-rural integration across diverse spatial patterns of population flow is essential for advancing theoretical explanations of urban-rural interaction dynamics in metropolitan-fringe counties. This study establishes a theoretical framework analyzing the directionality of urban-rural population flows and the processuality of county-level integration. Taking Sihui City, Guangdong Province—A significant spatial unit for urban-rural integration on the periphery of a major metropolis—as a case study, we employ the Entropy Weight Method and Grey Relational Analysis, supplemented by qualitative fieldwork interviews, to dissect spatial patterns of population flows and their causal mechanisms driving integration. Results indicate: 1) From 2010 to 2023, urban-rural population flow in Sihui City showed an overall upward trend, transitioning from rapid to stable mobility, with spatial patterns shifting from dominance by agglomerative flows to diffusive flows, while pendulum flows remained minimal; 2) Urban-rural integration exhibited significant stepwise growth during 2010—2023, where the Urban-Rural Interaction Index contributed most substantially and grew fastest among three sub-indices, followed by the Mutual Promotion Index, while the Complementarity Index increased least; 3) Population flows strongly correlate with integration development, yet association strengths vary markedly across indices, revealing distinct linkages: agglomerative flows with interaction, diffusive flows with mutual promotion, and pendulum flows with complementarity; 4) Integration in Sihui stems from bidirectional population flows and metropolitan-edge industrial diffusion: the Urban-to-Rural Agglomeration Pattern reflects safeguard-type interaction driven by urban support, the Urban-to-Rural Diffusion Pattern enables specialization-driven mutual promotion via industrial division, and the Urban-Rural Pendulum Pattern fosters equilibrium-oriented complementarity through workplace-residence separation. These findings demonstrate that the relationship between population mobility and urban-rural integration is conditioned jointly by flow direction and the socioeconomic processes through which resources, industries, and everyday activities are reorganized. The proposed direction-process framework extends the analysis of urban-rural relations beyond aggregate mobility effects and reveals differentiated pathways of integration in peri-metropolitan counties.

Universities and specialized, refined, distinctive and innovative (SRDI) enterprises are key supply and demand actors in regional innovation systems. Their coordinated spatial distribution is crucial for optimizing the allocation of innovation resources and building efficient regional innovation networks. Focusing on Beijing-Tianjin-Hebei, the Yangtze River Delta and the Pearl River Delta urban agglomerations, this study develops a theoretical framework to analyze the spatial mismatch between universities and SRDI enterprises. Methods including the spatial mismatch index, the optimal parameter-based geographical detector, and partial least squares (PLS) regression are employed to examine their spatial distribution patterns, mismatch characteristics, and influencing factors. The results indicate the following key findings. 1) Both universities and SRDI enterprises in the three major urban agglomerations exhibit significant spatial agglomeration characteristics. These can be broadly categorized into three typical patterns: a ‘single-center agglomeration’ model, a ‘multi-core networked’ model, and a ‘deep specialization-broad higher education’ model. The latter is exemplified by the Pearl River Delta region, where SRDI enterprises are highly concentrated in Shenzhen, while higher education resources are clustered in Guangzhou. Bivariate spatial analysis further shows that the spatial distributions of universities and SRDI enterprises demonstrate a relatively weak positive spatial autocorrelation overall. Among the three urban agglomerations, the Beijing-Tianjin-Hebei region displays the strongest spatial association between the two, followed by the Yangtze River Delta, while the Pearl River Delta shows comparatively weaker spatial linkage. 2) A widespread spatial mismatch exists between universities and SRDI enterprises, with notable variations in both the degree and types of mismatch within each urban agglomeration. These mismatches exhibit multi-level structural differences across cities. Specifically, the Pearl River Delta demonstrates a relatively high overall level of matching between universities and enterprises. In contrast, the Yangtze River Delta exhibits more pronounced spatial mismatch phenomena, characterized by a higher prevalence of ‘dual-low positive mismatch’ types, where both universities and enterprises are relatively underrepresented but maintain a positive spatial relationship. Meanwhile, ‘enterprise-dominated negative mismatch’ and ‘dual-high negative mismatch’ types—indicating imbalances where either enterprises or both entities are disproportionately concentrated—are less common. 3) The spatial mismatch between universities and SRDI enterprises is jointly driven by multiple dimensions of factors. Among these, road network density and elevation provide fundamental geographical and infrastructural conditions. Economic development levels and the scale of talent cultivation in higher education institutions create structural tensions that shape spatial alignment. In addition, public cultural resources and the quality of the business environment play important moderating roles, influencing how effectively universities and enterprises can co-locate and interact within the urban innovation system. Overall, these findings contribute to a deeper understanding of the spatial organization and interaction mechanisms of innovation actors, offering valuable insights for policy-making aimed at enhancing regional innovation efficiency.

Amid accelerated digital technology penetration, factor reorganization and governance transformation, urban functions are undergoing systematic changes in terms of connotation, structure, and representation. Against this backdrop, this study aims to clarify the evolutionary logic and core characteristics of urban functions in the new era and to reveal the drivers of their transformation. To achieve this, a three-dimensional analytical framework of “mechanism-connotation-representation” is constructed, through which the theoretical evolution, structural reconstruction, and representational transformation of urban functions are systematically examined. The city is conceptualized as a complex adaptive system integrating physical and virtual spaces, and urban functions are analyzed across driving mechanisms, spatial restructuring, and perceptual representation. The results indicate three fundamental transformations of urban functions in the new era. First, the dominant driving force has shifted from traditional physical factors, such as resource endowment and spatial proximity, to information flows composed of data, capital, technology, and knowledge, which reshape urban functional organization across multiple spatial scales. Second, the evolutionary logic has transitioned from planning-oriented and top-down control to organic growth driven by multi-agent interactions, in which governments, markets, and social actors jointly contribute to dynamic and self-organizing functional systems. Third, the operational mode has evolved from a linear metabolic pattern based on resource consumption to a resilient and circular model emphasizing functional regeneration, adaptive reuse, and sustainable development. In addition, urban functions demonstrate significant characteristics of physical-virtual integration and multi-scalar networked organization. Functional structures become increasingly heterogeneous, flexible, and dynamically reconfigurable, while the representation and perception of urban functions shift from static, place-based measurements to dynamic, data-driven, and multi-source sensing systems supported by emerging technologies. These forces jointly promote the reorganization of production, living, and ecological functions, and facilitate the transition toward a more integrated, adaptive, and sustainable urban system. The study concludes that urban functions have entered a new stage characterized by digital logic, physical-virtual integration, and multi-agent collaboration. This research provides a systematic perspective for understanding urban spatial restructuring and offers theoretical support for spatial planning optimization and sustainable development.

Artificial intelligence (AI) technology is experiencing an era of unprecedented innovation and intensifying competition. To effectively capitalize on emerging opportunities, a comprehensive understanding of the mechanisms driving its evolution is essential. This study draws on empirical data from AI invention patents across 285 cities at the prefecture level and above in China, covering the period from 2000 to 2022. It explores regional heterogeneity in AI technology development and investigates the underlying mechanisms that shape this evolution. The findings reveal the following: 1) AI technology in Chinese cities has evolved through a trajectory from “polarization dominance” to “neighboring diffusion”, ultimately reaching a stage of “multi-center diffusion”. 2) The development of AI technology in these cities is influenced not only by the local AI knowledge base but also by the rapid advancement of cross-industry technological integration. 3) Technological spillovers from neighboring regions, external collaborative innovation, and government policy support all play a significant role in reducing cities’ reliance on technological path dependency, thereby fostering breakthrough developments in AI technology. 4) The evolutionary mechanisms vary depending on regional AI development levels and technological categories: In technology hubs such as Beijing and Shenzhen, cities leverage both local technological foundations for self-reinforcement and technological spillovers from neighboring regions, alongside policy support, to achieve path-breaking advancements. In regions with moderate development, technological progress is primarily driven by collaborative innovation with core regions. In less developed areas, effective exogenous breakthrough paths have yet to emerge. From the perspective of technological categories, the development of hardware platform technologies heavily relies on local specialized technological foundations and external relations. In contrast, the development of general-purpose technologies is most responsive to cross-industry technological integration. The advancement of key technologies necessitates a combination of local specialized technological foundations, cross-industry technological integration, and external relations.

Against the backdrop of global supply chain restructuring and China’s “land-sea coordination” strategy, this study systematically analyzes the spatial evolution mechanisms of China’s five major coastal port clusters from 2005 to 2020 from an integrated land-sea perspective, leveraging maritime network data and regional big data on port-adjacent enterprises. The findings reveal that: 1) When the average weighted degree of ports in the Yangtze River Delta, the Pearl River Delta, and the Bohai Rim clusters exceeds 6, nearby high-end port-related industries is likely to grow by over 10%, reflecting a positive feedback between stronger shipping networks and industrial upgrading; 2) There is significant spatial heterogeneity in land-sea factor interaction—ports such as Shanghai, Ningbo-Zhoushan, Xiamen and Shenzhen exhibit deep land-sea coupling characterized by “shipping network reinforcement-seaward spatial expansion-cargo throughput upgrading”, whereas other ports lack sufficient land-sea linkage; 3) The evolution of port clusters has diverged. Maritime network expansion and inland industrial upgrading have jointly reshaped China’s coastal port system, transforming it from a monocentric structure centered on the Pearl River Delta into a tripolar configuration comprising the Yangtze River Delta, the Pearl River Delta, and the Bohai Rim. The Yangtze River Delta and the Pearl River Delta, leveraging deep-water channels and industrial clusters in their hinterlands, have achieved leapfrog development from single-core to multi-core hubs; the Bohai Rim cluster has fostered a balanced land-sea interaction through coordinated development across Tianjin, Hebei, Shandong, and Liaoning; southeast coastal regions, constrained by limited hinterland scale, maintain a monocentric structure centered on Xiamen Port; while southwest coastal ports show fluctuating development due to underdeveloped maritime networks and weak inland industrial foundations. The study underscores that the evolution of container port systems is shifting from simple land-sea interaction toward complex coordination, calling for the construction of resilient supply chain networks and the transformation of ports into global supply chain orchestrators.

Exploring the spatial heterogeneity and influencing factors of the well-being effects from ecological product value realization in tourism scenic byway under a human-centered orientation, holds significant practical importance for enriching the theoretical framework of ecological product value realization research and enhancing the quality of life for farmers. Based on the survey data of 1 374 rural households in the Sichuan-Xizang route in southern Anhui, a theoretical analysis framework was constructed, under the needs of farmers’ dominant position. The GIS spatial analysis, geographical detector, and multi-scale geographically weighted regression models were employed to investigate the spatial patterns and multiple influencing factors of the well-being effects resulting from the realization of ecological product values in tourism scenic byway. The results show that: 1) The spatial distribution of well-being effects from ecological product value realization along the Sichuan-Xizang route in southern Anhui showed distinct regional disparities. High well-being effects demonstrated spatial characteristics of “along roads”, “near scenic spots”, and “around towns”, whereas low well-being effects were primarily found in areas with significant topographic relief. 2) Both the comprehensive index and individual dimensional indices of well-being effects from ecological product value realization exhibited significant spatial auto-correlation in the Sichuan-Xizang route in southern Anhui. Cold spots and hot spots showed marked clustering, with notable differences in the distribution patterns of high-value and low-value clusters. 3) The spatial differentiation pattern of well-being effects was shaped by the synergistic effects of natural environment, scenic byway elements, socioeconomic factors, and policy, with notable differences in the contribution of individual factors and synergistic enhancement effects in factor interactions. 4) The spatial scale of influence of these factors showed certain differentiation. The degree of influence of each factor on the well-being effects exhibited clear spatial heterogeneity, and the strength of their impact varied across sample units. This research offered theoretical reference and empirical insights for understanding the practical implications of well-being effects from ecological product value realization and advancing high-quality development.

To explore the perceptual characteristics and spatial heterogeneity of 6 ecosystem service (ES) types in Guangzhou Haizhu National Wetland Park, the article employed a mixed-methods approach combining discourse analysis with behavioral analysis. The findings reveal that: 1) The analysis of text data showed a high perception of biodiversity services (44.17%), particularly triggered by the “flower sea” landscapes, while concurrent attention to cultural and educational services highlighted the wetland’s role as a nature education venue; 2) The analysis of footprint data demonstrated that recreational services constituted the dominant function (67.94%), evidenced by the formation of high-intensity activity clusters around Haizhu Lake, which substantiates the wetland’s role in fulfilling urban leisure demands; 3) The perceived disparities between text analysis and footprint analysis reveal a “discourse-behavior divergence” in ES perception. Biodiversity and cultural-educational services were prioritized in discourse, while recreation dominated spatial practice. Carbon sequestration and plant provision services demonstrated balanced perception, whereas significant perceptual gaps existed for water purification services. Spatially, the differential attractiveness between “flower seas” and “lakeside areas” substantiates the interplay of service perceptibility (explicit/implicit), spatial facility guidance, and public value preferences. The research constructs a “social-spatial” dual-dimensional framework that innovatively elucidates the dialectical relationship between discursive representation and spatial practice in the perception of ecosystem services within urban wetland parks. This framework deepens the understanding of the social construction process underlying the service values of urban natural spaces. The findings offer spatial governance pathways for ecological recreation zoning, visualisation of implicit services, and guidance on public behaviour, thereby facilitating coordinated human-nature development in high-density urban environments.

As an important micro-spatial unit of residents’ daily activities, the environmental inclusiveness of urban branch roads reflects the basic conditions for safe, continuous and accessible traffic of different groups. Taking urban branch roads in Xuanwu District, Nanjing as the research subject, this article develops an environmental inclusiveness evaluation framework using street view images, road network, and POI data. This framework specifies four evaluation dimensions (completeness, safety, accessibility, and comfort), and simultaneously classifies the roads into residential, commercial, cultural and educational, landscape, and historical and cultural types according to POI characteristics and streetscape recognition. On this basis, a human-machine collaborative evaluation method based on street view images is constructed to quantitatively measure the environmental inclusiveness of branch roads. Furthermore, the Moran’s I is applied to investigate the spatial distribution characteristics, and the SHAP method is adopted to rank the influence of various environmental factors on inclusiveness by their relative strength. The results show that: 1) The inclusiveness of urban branch road environments shows significant spatial differentiation and local agglomeration characteristics at the microscale, and the high-value branch roads are mostly concentrated in areas with high renewal intensity and complete slow travel facilities, while the low-value branch roads are mainly distributed in historical streets and alleys and areas with weak infrastructure. 2) There are stable differences in the inclusiveness level and internal structure of different types of branch roads, and the overall performance of landscape and cultural branch roads is better, while the overall performance of historical and cultural branch roads is low in multiple evaluation dimensions. 3) Environmental factors such as green viewing rate, enclosure and sky openness have obvious nonlinear effects on the inclusiveness of branch roads, among which the green viewing rate generally shows a positive effect, while the enclosure degree and sky openness show obvious phased change characteristics. This study overcomes the difficulties of traditional road inclusiveness evaluation, which is hard to unify due to subjective differences and difficult to quantify owing to road diversity. From the micro-scale of branch roads, it reveals the spatial heterogeneity of urban spatial equity at the level of daily mobility, thereby providing a quantitative analytical approach and empirical basis for inclusiveness-oriented branch road renewal and refined governance.

The Huaihe River Basin suffers from frequent extreme rainstorms and flood disasters. Conducting research on Holocene palaeoflood hydrology and uncovering the Holocene paleoflood data will provide scientific support for the construction of water conservancy projects, as well as the prediction and response to flood risks under future climate change in the Huaihe River Basin. Through extensive and detailed field investigations in the basin, the Beiweiwan profile containing overbank flood deposits (OFD) was discovered in the first terrace on the north bank of the Luoshan-Xixian reaches in the upper Huaihe River. Based on stratigraphic division of the profile, combinging with indicators such as macro-sedimentological characteristics, grain-size, magnetic susceptibility, total organic carbon, and quartz grain micromorphology, the typical macro-sedimentological and micro-physicochemical characteristics of OFD were revealed. Using stratigraphic correlation and optically stimulated luminescence (OSL) dating, the overbank flood events recorded in the profile occurred in the mid-Holocene from 4.2 ka B.P. to 4.0 ka B.P., which is a high-frequency period of palaeoflood in the Huaihe River Basin. Meanwhile, the climate-driven mechanism of Holocene palaeoflood events in the upper reaches of the Huaihe River was explored. During 4.2—4.0 ka B.P., the North Atlantic Oscillation (NAO) was in a persistent negative phase, the number and variability of El Niño-Southern Oscillation (ENSO) events increased, the East Asian Summer Monsoon was unstable, and the Western Pacific Subtropical High strengthened. In addition to the impacts of shear lines, low vortices, low-level jets, and typhoons, resulting in abnormal climate, increased frequency and intensity of extreme precipitation, and frequent occurrence of extreme flood events in the Huaihe River Basin. Therefore, the synergistic effects of the negative phase of the NAO and increased ENSO variabilities led to the frequent occurrence of the 4.2 ka B.P. palaeoflood events. The 4.2 ka B.P. palaeoflood events in the Huaihe River Basin were the response of the fluvial system to climate change in a regional scale, which is an evidence for understanding how fluvial hydrological systems respond to global climate change.

Based on monthly precipitation data with a spatial resolution of 1 km and sea surface temperature (SST) data with a spatial resolution of 1.0°×1.0° spanning the period from 1970 to 2020, this study employed empirical orthogonal function (EOF) analysis and Morlet wavelet analysis to investigate the dominant spatial modes and the interannual periodic variations of the flood season precipitation in the Poyang Lake Basin (PYLB). Furthermore, the Liang-Kleeman information flow method was applied to identify the key oceanic SST driving regions of the flood season precipitation in the PYLB. The results show that: 1) During the period from 1970 to 2020, the flood season precipitation in the PYLB is mainly characterized by two spatial modes, namely a basin-wide coherent mode and a north-south out-of-phase mode. The basin-wide coherent mode features in-phase precipitation variations across the entire basin, with stronger variability in the northeastern part of the PYLB, and exhibits an interannual periodicity of 2-4 years. The north-south out-of-phase mode is divided roughly near 28°N, showing opposite variations between the southern and northern parts of the basin, with periodicities of 2-4 years and 3-6 years. 2) The equatorial central-eastern Pacific and the Indian Ocean serve as the key SST driving source regions for basin-wide coherent precipitation variability. Specifically, warm SST anomalies in these regions favor increased flood season precipitation of the PYLB, whereas cold SST anomalies favor decreased flood season precipitation over the entire basin. 3) Under the combined influence of El Niño events and warm SST anomalies in the Indian Ocean, the “southern drought-northern flood” pattern during the flood season in the basin tends to intensify. Conversely, under the combined influence of La Niña events and cold SST anomalies in the Indian Ocean, the “southern flood-northern drought” pattern during the flood season in the basin tends to be enhanced. The research results can provide scientific support not only for the assessment of flood and drought risks, but also for the prediction of extreme hydrological events in the PYLB and the middle and lower reaches of the Yangtze River.

This study is based on socio-economic data, land use data, and high-precision water system data that have been repeatedly corrected during the same period. By constructing a comprehensive evaluation system for river networks and storage and flood control capacity under the background of urbanization, a spatiotemporal geographically weighted regression model (GTWR) is used to analyze the quantitative relationship between river networks system changes and storage and flood control capacity in the Changsha-Zhuzhou-Xiangtan urban agglomeration from 1995 to 2020, as well as the impact of urbanization factors and various river networks system factors on water system storage and flood control capacity. Research has shown that: 1) The 3-dimensional urban development level of the Changsha-Zhuzhou-Xiangtan area exhibits notable temporal and spatial disparities, with a discernible shift in the urban centre of gravity towards the northwest. 2) The hydrological system has undergone significant changes. The indicators of the hydrological systems exhibit considerable variation across different cities and time periods. In quantitative terms, the water surface ratio (W_p) and river network density (D_r) exhibited a 33% and 2% reduction. With regard to structural characteristics, the decay rates of the tributary development coefficient K and river network complexity C_r are 5.50% and 5.11% respectively. 3) During the study period, the water system’s storage and flood control capacity showed significant regional differences. The trench storage capacity (C) and the unit area trench storage capacity (SR) exhibited an increase, whereas the available storage capacity (AC) and the unit area available storage capacity (ASR) demonstrated a decline. 4) There is a significant correlation between the water system’s storage and flood control capacity and a number of factors, including per capita GDP, water surface ratio W_p, river network density D_r, area-length ratio R_al, and river network complexity C_r. In terms of temporal heterogeneity, the impacts of per capita GDP, R_al and C_r on trench storage capacity C have undergone a process of shifting from negative to positive and increasing year by year. With regard to spatial analysis, per capita GDP has been found to have the greatest impact on trench storage capacity in Zhuzhou, while in Changsha, the impact was found to be the least significant. This study provides support for the environmental protection of the Changsha-Zhuzhou-Xiangtan urban agglomeration in the lower reaches of the Xiangjiang River Basin.

Spatio-temporal changes of pond land in the hilly rice-growing areas of southern China are significant, yet there is relatively little exploration into its evolutionary trends, spatial patterns, with the aim to drive mechanisms. This study explores the spatiotemporal evolution characteristics and driving mechanisms of pond land in Xiangxiang City of Hunan Province, and analyzes the magnitude of area changes, sources of newly added ponds, and destinations of vanished ponds, with the aim to reveal the changing trends and relative influence of various driving factors, and provide scientific basis for spatial optimization and governance of pond land use. Research results show as follows: 1) Over the past decade, Xiangxiang City experienced a net reduction of 723.35 hm² in pond surface area, while the total number of ponds increased by 8 835. This corresponded to a 29.17% decrease in average pond size, indicating a clear trend toward pond miniaturization. The combined effects of existing pond abandonment and new pond excavation in alternative locations resulted in significant transformations in pond land use patterns. 2) Pond areas demonstrated progressively strengthened spatial autocorrelation, ultimately developing a clustered distribution pattern described as “one central cluster with four satellite groups”. A clear spatial differentiation appeared along G320 road, with the northwestern region primarily showing area reduction while the southeastern area was dominated by area expansion. An overwhelming majority (86.65%) of administrative villages fell into the “quantity-increase” category. 3) Cultivated land, forest land, and construction land served as both the primary sources of newly added pond areas and the main destinations of vanished ponds. The spatial optimization of pond land resulted in a net reduction of 252.80 hm² of cultivated land, while garden land, forest land, and construction land all experienced substantial increases. 4) The roles of natural conditions, land-use patterns, and socioeconomic factors in driving pond transformation progressively intensified. Ponds gradually relocated to gentle slopes and low-lying plains, while those in high-altitude steep terrain were progressively abandoned. Increased cropland ratio and settlement density promoted spatial expansion of ponds, whereas the designation and enhanced proportion of permanent basic farmland effectively restrained pond encroachment. The evolution of pond land in Xiangxiang City was driven by multidimensional factors, including physiographic conditions, locational advantages, socioeconomic development, and regional resource allocation. Future efforts should enhance the control of pond expansion, crack down on illegal activities, implement pond rehabilitation projects, and adopt strategies to maximize multifunctional benefits of ponds.

Elucidating the human-land coupling mechanisms of urban wetlands across different latitudinal gradients is crucial for ensuring regional ecological security. Focusing on Harbin (a typical cold-region city) and Haikou (a typical tropical city)—both designated as International Wetland Cities—this study utilizes multi-source remote sensing data, spatial statistics, and landscape metrics to conduct a comparative analysis of their spatiotemporal evolution and underlying driving mechanisms. The results indicate that: 1) In terms of spatiotemporal evolution, Harbin’s wetlands, influenced by the dual forces of the Farmland-to-Wetland Conversion Policy and urban expansion, exhibit a northwestward centroid shift, with enhanced landscape heterogeneity but intensified fragmentation and impaired hydrological connectivity. Conversely, Haikou’s wetlands are rigidly constrained by topography, maintaining a relatively stable spatial distribution, with high-density clusters concentrated in coastal zones and well-preserved connectivity in mangrove landscapes. 2) Regarding driving mechanisms, the evolution of wetlands in Harbin is primarily driven by hydro-ecological factors, specifically vegetation coverage and river network density, while being subject to significant interference from economic development. In contrast, spatial differentiation in Haikou is predominantly governed by geomorphic factors such as elevation and slope. 3) Policy implications: cold-region cities must prioritize balancing the trade-off between ecological restoration and agricultural/economic development by repairing severed ecological corridors. Tropical cities should focus on maintaining the stability of low-elevation “tidal-terrain” coupling systems and strictly regulating coastal development.

This study takes the Huaying mining ecological restoration area as the research object. Based on domestic ZY-1 02D hyperspectral data and contemporaneous field aboveground biomass sampling data, feature screening was conducted through correlation analysis and partial least squares (PLS) method. 6 machine learning methods, including Random Forest (RF), Support Vector Machine (SVM), Gaussian Process Regression (GPR), Extreme Gradient Boosting (XGBoost), MARS, KNN were employed to construct grassland aboveground biomass estimation models, and the optimal model was selected for aboveground biomass inversion. Given the simple vegetation structure but strong spatial heterogeneity of grasslands in mine ecological restoration areas, this study systematically examined the contributions of spectral transformations, vegetation indices, and textural features to estimation accuracy and compared differences in model performance before and after feature selection. The results show that: 1) The first-order derivative spectral features demonstrate significant potential in aboveground biomass estimation, and the SVM model constructed based on this feature set achieves the highest accuracy; 2) Feature optimization using partial least squares (PLS) can effectively improve model performance; 3) Incorporating the texture feature second-order moment (SEM) further enhances model performance. The SVM model built on the combined spectral-textural feature dataset (WLP8) performed the best, with mean absolute error (MAE), coefficient of determination (R²), and root mean square error (RMSE) values of 41.5645 g/m², 0.8478, and 45.0014 g/m², respectively, demonstrating high prediction accuracy and better stability; 4) The total grassland aboveground biomass in 2023 in the study area was 44.7863 t, which decreased slightly compared with that in 2022, but its stability was enhanced. The hyperspectral remote sensing estimation framework proposed in this study provides an accurate and practical method for estimating grassland aboveground biomass in mine ecological restoration areas. The study demonstrates that integrating hyperspectral spectral, textural, and spatial features with machine learning can effectively enable the quantitative inversion of aboveground grassland biomass in mine ecological restoration areas, which provides data support and technical methods for evaluating the effectiveness of mine ecological restoration and for subsequent ecological management and conservation.