ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15110508

Authors: Danny Ebanks Devika Jain

Research data repositories are essential infrastructure for scientific inquiry and for ensuring research datasets follow FAIR (Findable, Accessible, Interoperable, and Reusable) principles. At the same time, the effectiveness of repositories at promoting reuse depends on the quality and completeness of geospatial and thematic metadata, which is usually supplied voluntarily by a data depositor or researcher. Yet, due to the voluntary nature of such user-derived metadata and the curation and resource constraints facing research data repositories, it is no surprise that even on the Harvard Dataverse Repository, the world’s largest general-purpose research repository, most metadata fields are missing. A missing metadata field represents a loss of information, so an incomplete metadata record is also a less interoperable record. In fact, we find that datasets with more missing metadata are associated with fewer downstream citations, and, more to the point of this paper, with fewer resolvable connections to other datasets. The implications for geospatial datasets are even more stark, as only 0.3 percent of research datasets outside a single archaeological deposit have a bounding box and nearly all such bounding boxes are archival data points rather than full geographic shapes. We find that the value in geospatial metadata is that location helps connect datasets’ concepts across disciplines. After embedding all of the datasets on the Harvard Dataverse into a metadata knowledge graph, we find that datasets are twice as likely to be connected across scientific disciplines through shared geospatial metadata as through keyword metadata, evidence that geographic metadata is the most reliable attribute we have for making datasets interoperable across the disciplinary boundaries. We train a small language model and fine-tune it on datasets that are stored on the Harvard Dataverse repository. We find that through geospatial metadata enrichment, we are able to increase the share of datasets from different disciplines that are connected via metadata elements from 58.5 to 63.2 percent.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15110507

Authors: Zhenbao Wang Qun Yu Yuchun Wang Li Sun Fangzhou Liu

Improving the survival rates of critically ill and injured patients, particularly vulnerable groups such as children and older people, necessitates timely and effective emergency medical services (EMS). However, current research on EMS accessibility typically overlooks the additional travel time required to take patients to hospitals. This study proposes a two-stage nearest-facility supply–demand accessibility index for evaluating potential emergency medical services accessibility (PEMSA) (PEMSA represents only the morning-peak scenario), accounting for both supply–demand disparities and the time sensitivity of emergency response, with the area within Beijing’s Sixth Ring Road serving as a case study. The two-stage EMS framework involves transportation from the closest first-aid station to the residential community, and transportation from the residential community to the nearest emergency department. The Gini coefficient is employed to examine the equity of PEMSA across the research area’s administrative districts. A multiscale geographically weighted regression (MGWR) model is employed to evaluate the relationship between PEMSA and demographic characteristics, namely the percentages of children and older people. The results indicate that PEMSA follows a core–periphery spatial pattern. Based on accessibility and equitable outcomes, the identified spatial mismatch areas are classified as critical priority, priority, or sub-priority intervention zones. The primary reasons for this accessibility discrepancy are road traffic congestion and a geographic mismatch between emergency facility density and demographic distribution; accessibility is significantly correlated with the proportions of children and older people. These findings, as well as the proposed analytical framework, are valuable for evaluating urban EMS accessibility and optimizing the spatial layout of EMS.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15110506

Authors: Pengxin Zhang Taiyang Yang Dong Yang Nan Jiang Kaixuan Du Min Yang

Accurate vessel trajectory prediction in complex nearshore waters remains challenging because local traffic patterns are highly heterogeneous and similar observed trajectories may lead to different future route continuations. Although historical Automatic Identification System (AIS) archives contain valuable evidence of previous vessel movements, such information is typically absorbed into model parameters rather than explicitly accessed during prediction. To address this limitation, this study proposes a spatially constrained memory-augmented framework for vessel trajectory prediction. An extensible key–value memory bank is constructed from paired representations of historical past and future trajectory segments. During prediction, geographically relevant candidates are first identified through spatial filtering, and the most similar historical trajectories are retrieved in the learned representation space. The corresponding future representations are then selectively aggregated with the current observed trajectory representation through an attention mechanism conditioned to generate the future trajectory. Experiments on AIS trajectories from the Pearl River Delta and adjacent coastal waters show that the proposed model consistently outperforms baselines over 60, 120, and 180 min horizons, with particularly clear improvements over the long short-term memory (LSTM) prediction backbone without historical trajectory fusion. Ablation results further confirm the effectiveness of spatially constrained retrieval and attention-based memory fusion, particularly for longer-horizon prediction.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15110505

Authors: Yongbin Tan Rongfeng Cai Yan Ding Hong Wang Jun Wu Hao Lyu

Geospatial data contains abundant temporal, spatial, and attribute information related to geographic entities, serving as a critical foundation for the development of new quality productive forces. Achieving effective semantic representation of geographic features is a key prerequisite for enabling intelligent data analysis and application. Spatial relationships are an essential semantic characteristic of geographic features and are among the most commonly introduced characteristics for representation learning in recent studies. However, current research primarily focuses on point-based features (such as POIs) or on converting line and polygon features into point features, leading to inaccuracies in representing spatial proximity between features, which in turn affects the representation outcomes. In this paper, a feature unit delineation method based on geometric shapes is proposed to better construct the proximity relationship graph between features. It enriches the semantic features of feature units using a large language model and trains an edge optimization model to refine the graph structure. Building upon iGCL, a semantic representation learning model for geographic features is constructed. Finally, using Xiamen Island and Nanchang City as study areas, experimental datasets are built by integrating OSM and Amap (Gaode) POI data. The performance of the proposed model is validated and compared with other models through three tasks: semantic overlap calculation, node classification, and node clustering. The results show that the feature representations obtained by the proposed model achieve the best performance across all tasks, fully demonstrating its effectiveness and superiority in geographic feature representation learning, thereby providing a foundation for subsequent intelligent geographic feature tasks.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15110504

Authors: Xinmeng Li Jianhua Xu Qihao Huang Zhanwei Wu

Location-based services (LBS) have traditionally helped users answer questions such as where to go or what is nearby. With the emergence of large language model (LLM)-based agents, LBS are beginning to move toward understanding open-ended goals, coordinating spatial tasks, and supporting decisions about what to do and how to do it. This review refers to LBS that integrate such agentic capabilities into location-dependent service processes as Agentic LBS. To systematically characterize the current state of Agentic LBS and identify directions for future research, we conducted a scoping review and proposed an Autonomy–Innovation–Experience (A-I-E) framework as its analytical lens. Following the PRISMA Extension for Scoping Reviews (PRISMA-ScR), we searched the ACM Digital Library, Web of Science, and Scopus for peer-reviewed, English-language journal articles and conference papers published between 2021 and 2026. The search identified 228 records. After removing 21 duplicates, 207 records underwent screening, and 27 studies were included in the final review. The included studies were systematically charted and analyzed through descriptive mapping and qualitative coding to characterize their application directions, autonomy levels, innovation patterns, and user-experience implications. The results show that: (1) Agentic LBS research remains concentrated at lower autonomy levels, with 23 of the 27 studies (85.18%) classified as L1–L2 and four as L3; nevertheless, the emergence of L3 systems indicates a gradual movement toward more complex task planning and coordination. (2) LLM-based agents are expanding LBS beyond location-based information retrieval toward more personalized, context-aware, and task-oriented service delivery. (3) Reported user-experience benefits are concentrated in usability, accessibility, personalization, and task efficiency, whereas evidence concerning trust, user control, privacy, safety, and long-term use remains limited. (4) By synthesizing these findings across autonomy, innovation, and experience, this review proposes a future-oriented conceptual framework for understanding the emerging trajectory of Agentic LBS.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15110503

Authors: Eun-Joo Suk Yo-Han Han Gwang-Jae We Ju-Hwan Moon

Driving simulation needs labeled 3D scenes, but mobile mapping system (MMS) light detection and ranging (LiDAR) is often delivered unclassified, without point-wise labels for training. We present a training-free pipeline that transfers the semantics of a vector high-definition (HD) map, the Korean precision road map, onto MMS LiDAR and outputs a labeled 0.5 m voxel scene, and we analyze what limits its accuracy. Each voxel takes the label of the nearest map feature whose category-specific radius and elevation window it satisfies, with elevations measured from a ground model built from the map’s 3D vertices. The pipeline needs no training data but requires a 3D vector map. On an expressway interchange, the ground model reached a median error of 6.6 cm, on par with inverse-distance weighting (IDW), triangulated irregular network (TIN), and local-plane interpolation. A blind, stratified random sample of 695 cells and a blind review of all 91 output objects showed that 98.0% [95.8, 99.5] of the output voxels are static solid structure and 73% [63, 81] of the objects are the mapped facility; the cell-level macro-F1 over six categories was 0.44 [0.39, 0.52] (guardrail 0.73, tunnel 0.94, sign 0.14). Errors were concentrated: four tunnel-mounted signals labeled the tunnel ceiling, and the radii of signs and posts captured roadside vegetation. The results show where map-based labels are reliable for simulation and which map content needs treatment.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15110502

Authors: Yulong Peng Hongtao Sun Xiqian Peng Chenqi Li

Block-scale evaluation of sustainable urban form is a critical foundation for fine-grained urban renewal and governance in high-density central urban areas. However, most existing frameworks link sustainable development directly to outcome indicators without clarifying the mediating levels and transformation pathways between them, which leads to conceptual ambiguity and arbitrary indicator selection. To address this gap, this study proposes a four-layer mapping framework comprising the sustainable development dimensions, morphological attributes, spatial carriers, and quantitative indicators, in which the morphological-attribute layer acts as a mediating mechanism that translates normative goals into measurable spatial variables. The framework is operationalized through indicator screening based on expert questionnaires and a combined subjective–objective weighting scheme that integrates The order relation analysis method with entropy weighting, and it is applied to the Nanshi Area of Shenyang, China. The results reveal that block-scale sustainability arises not from individual morphological elements but from the coordinated development of multiple spatial components, with proximity, distribution, and complexity contributing most strongly. The proposed framework provides a transparent approach for urban-form diagnosis and fine-grained spatial governance and offers a reference analytical pathway for other high-density urban districts.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15110501

Authors: Rui Xu Yu Zheng Yanxi Kuang Zixuan Ren Meiyi Qi Jingqi Hong

County-level hurricane power outage risk prediction faces a mismatch in spatial support among hazard rasters, infrastructure exposure, and outage labels. Conventional county-mean aggregation can weaken the spatial association between localized high hazard values and transmission corridors. To address this problem, this study proposes the Raster-to-Grid Spatiotemporal GeoAI Network (R2G-STGeoNet). The method first integrates Sentinel-1 SAR, precipitation, wind fields, and geographical information to encode multisource hazard states. Raster-to-Grid Exposure Projection (R2GEP) then transforms continuous hazard fields into features describing county exposure, along-line exposure, local high values, and above-threshold exposure length. A dual-relation spatiotemporal graph combining geographic adjacency and transmission corridor relationships jointly predicts outage occurrence probability and outage rate. Using hurricanes as case studies, we conduct common-input baseline comparisons, structural ablations, cross-event transfer, and spatial holdout validation while considering missing labels and data latency. The full model achieves an AUROC of 0.913 in retrospective reconstruction, while the availability-constrained one-hour-ahead evaluation yields an AUROC of 0.884, decreasing to 0.862 when dynamic inputs unavailable at issuance are masked. Replacing R2GEP with county means reduces AUROC to 0.867, and the proposed model outperforms R2G-ST-GNN (AUROC = 0.895) under identical R2G inputs. The results indicate that spatial support transformation and infrastructure-relation modeling can improve county-level outage risk prediction under extreme weather.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15110500

Authors: Yuxuan Chen Dongzhou Li Jingxiu Wu Dandan Li Yutian Chao

Multi-scenario flood simulation supports urban flood-risk assessment, but scenario-specific outputs are difficult to use to identify facility-level risk evolution and critical risk conditions. This study proposes a multi-scenario flood risk reasoning method for critical urban facilities based on facility–scenario risk chains (FSRCs). The method constructs internally comparable rainfall–inundation scenario subsets, anchors each facility to a fixed triangular mesh cell, and organizes inundation depths and depth-based risk states into ordered chains. Boundary-crossing reasoning is then used to derive scenario-context-dependent critical rainfall threshold (CRT) estimates and ordered risk-state transition sequences. The method was evaluated in the Hexi drainage area of Nanjing, China, using 184 rainfall–inundation scenarios and 241 critical facilities. The results show that accumulated rainfall alone is insufficient for consistent risk ordering across different rainfall durations, while facility risk states vary across comparable rainfall-ordered scenario sequences. Representative facilities exhibited distinct CRT positions and transition intervals. Sensitivity analysis showed that CRT estimates varied with scenario spacing, while a computational consistency assessment confirmed the reproducibility of boundary-crossing results. The proposed method transforms discrete flood simulation outputs into structured object-level risk knowledge, supporting facility-level risk screening, risk-state comparison, and subsequent knowledge-driven applications.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15110499

Authors: Ming Zhang Yaoxin Han Jingnan Liu Yan Du

Classical fracture mechanics assumes a homogeneous medium, whereas a perilous rock body (a cliff block cut by dominant fissures, elsewhere called an unstable rock block) is a heterogeneous jointed rock mass; this contradiction, and the lack of an established route to rock mass scale fracture toughness, is the central obstacle to applying fracture mechanics to geological disasters. This paper discusses the fracture mechanisms of perilous rock collapse and their application to whole-process early warning. Models for the three types of perilous rock, sliding, toppling, and falling, are established and solved to give the mode I and mode II stress intensity factors, a mixed mode failure criterion, and a real-time stability coefficient Fs(t) = KIC/Ke. This paper proposes an approach for determining the fracture toughness of perilous rock using equivalent rock characterized by the rock mass rating (RMR) and the Hoek–Brown parameters; for the homogeneous model material, the route returns 1.88 MPa·m0.5 against a measured KIC of 2.11 MPa·m0.5, agreeing within 11%, and the characteristic length it yields bounds the rock mass quality for which the linear elastic criterion applies. In demonstration tests on four identical specimens, whose individual records deviate from their mean by 2.31–4.74%, the mean record traversed the four warning states in order; at the loading rate of the tests, the stable stage lasted 192 min against 2 min and 1 min for the quasi-stable and limited stability stages, and failure came with almost no visible prior deformation, so that in these tests, the fracture information remained observable through 98.5% of the loading history in which deformation gave no visible signal.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15110498

Authors: Zhiyuan Zheng Xinwang Sun Ziyang Yan

Urban green space accessibility is a key indicator for evaluating the rationality of green space distribution, service equity, and residents’ well-being. To examine the thematic structure and recent developments of urban green space accessibility research, this study analyzed 373 publications indexed in the Web of Science Core Collection between 2016 and 2025. Bibliometric analyses using VOSviewer and CiteSpace were integrated with BERTopic-based topic modeling to systematically examine the research landscape and thematic evolution of the field. The results indicate the following: (1) Within the selected corpus, annual publication counts increased markedly after 2020 and peaked in 2024. China and the United States were the leading contributors, while international collaboration varied across countries and regions. (2) Spatial accessibility measurement using GIS-based network analysis and the two-step floating catchment area method, together with equity assessments of green space provision, continued to constitute the methodological foundation of the field. (3) Inequalities in green space accessibility, assessments of green space provision, and green space equity and spatial justice were the three largest research themes. The research focus has gradually shifted from basic measurements of distance and service coverage toward supply–demand matching, socioeconomic disparities, health and well-being, perceived accessibility, vulnerable populations, 15-min living-circle governance, and climate adaptation. (4) Future studies should place greater emphasis on actual green space use, subjective perceptions and experiences, the integration of multi-source data, and accessibility to green space services. The proposed literature review framework, which integrates bibliometric analysis, topic modeling, and trend identification, provides a methodological reference for future reviews in related research fields.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15110497

Authors: Jinyun Yang Mei Liu Mengxiao Tian Yuxiang Zhang Wenxi Li

Agricultural heritage landscapes (AHLs) are multidimensional cultural landscapes shaped by interconnected productive, ecological, hydraulic, infrastructural, and sociocultural relationships. However, these dimensions are often assessed separately, limiting understanding of their spatial organization and internal differentiation. This study develops a knowledge-guided GeoAI approach for the multidimensional spatial characterization and assessment of AHLs. The approach integrates multisource geospatial data, high-resolution remote sensing, GIS-based spatial measurement, graph representation learning, and spatially constrained regionalization. It was applied to the Sangyuanwei Dike–Pond landscape in the Pearl River Delta, China. Four dimensions were examined through spatial indicators: Production and Livelihood, Water Management, Transportation Connectivity, and Ecological Networks. The results reveal strong associations between productive and ecological characteristics, while hydraulic organization and transportation connectivity show distinct relationships with traditional spatial patterns of land and water. Graph-learned configurations describe recurrent multidimensional combinations and their neighbourhood organization. Spatially constrained regionalization synthesizes these patterns into five geographically coherent and interpretable landscape types. By moving beyond separate element mapping, the approach provides integrated spatial evidence for landscape characterization and preliminary discussion of differentiated management. The integrated workflow offers a methodological reference for agricultural heritage landscape research, with potential applications in future environmental monitoring and vulnerability assessment.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15110496

Authors: Yinhua Tao Hao Guo Yongyi Zhang Binghao Li Jun Wen Linfeng Zhang Junyi Zhang

Ride-hailing provides convenient door-to-door mobility but may also reduce public transport use and transport-related physical activity. This study develops a spatially explicit utility-based counterfactual framework to compare observed ride-hailing trips with metro-based multimodal alternatives and assess potential health implications. Using over twenty million georeferenced ride-hailing orders recorded in Nanjing, China, in 2025, we generated complete multimodal alternatives by integrating origin–destination locations with metro and street-network data. The two travel options were compared in terms of generalized utility, travel impedance, spatiotemporal patterns, and physical activity. The results show that, under the baseline utility specification, 13.31% of ride-hailing orders were identified as having a comparatively competitive metro-based multimodal alternative. These orders were concentrated during weekday commuting peaks, around metro stations and corridors, and along inward flows toward central urban areas. Under a counterfactual scenario in which these identified ride-hailing orders were replaced by multimodal travel, transport-related physical activity was estimated to increase by 116.7 kcal per trip, resulting in remarkable gross health benefits when extrapolated to the citywide population. Overall, the framework provides a geospatial screening approach for identifying where and when multimodal alternatives may be competitive with ride-hailing, and highlights the importance of reducing access, egress, and transfer disutilities in improving multimodal competitiveness and associated health co-benefits.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15110494

Authors: Seonga Cho Mahad Imran

Facility location problems (FLPs) underpin decisions in emergency response, healthcare access, and infrastructure siting, yet applying them requires both optimization modeling skills and substantial spatial data preparation. This paper presents SOCA (Spatial Optimization Conversational Agent), a system that confines a large language model (LLM) to classifying the formulation and extracting scalar parameters through schema-validated function calls and delegates study area resolution, open-data acquisition, distance computation, and exact solution of four classic FLP formulations (P-Median, P-Center, MCLP, and LSCP) to deterministic tools. Every specification is confirmed by the user before solving, and solutions can be reported with substitutability and spatial equity diagnostics. We present SOCA as a technical feasibility study. Worked examples show how requests become staged, user-confirmed specifications; the MILP encodings reproduce the optima of a reference implementation exactly; and the data pipeline returned the targeted input layers across seven study areas, including cities in Africa and South Asia, using primary or fallback sources; retrieval does not establish data quality. A Manhattan fire station case illustrates coverage trade-offs and post-solve diagnostics on road network distances. Measuring interpretation accuracy, comparison against other LLM tool workflows, and a usability study with non-technical users remain future work.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15110493

Authors: Jiayong Yu Xiaoxue Du Maolun Zhou Deliang Lv Jizhe Li Jing Zhang Peng Lin

Accurate and timely extraction of land–water boundaries is essential for water resources management, flood monitoring, and wetland conservation. This study proposes a method for extracting land–water boundaries from unmanned aerial vehicle (UAV)-borne bathymetric light detection and ranging (LiDAR) point clouds. First, land point clouds are projected onto a reference plane constructed by fitting water surface points. Next, an alpha-shape algorithm extracts the boundary of these projected points. A combined algorithm of periodic curvature adaptive exponential tension spline interpolation and Savitzky–Golay (SG) filtering is then applied for secondary smoothing to generate the land boundary curve. Finally, curve segments adjacent to water surface points are identified as the land–water boundary. Experimental results show that, compared with manual extraction, the method yielded a root mean square error (RMSE) of 0.118 m at Danjiangkou Reservoir (main test area) and 0.184 m against independent real-time kinematic (RTK) measurements at Suzhou Reservoir. This demonstrates that UAV-borne bathymetric LiDAR meets the requirements for refined boundary extraction, and the proposed secondary smoothing algorithm successfully balances accuracy and smoothness to accurately reflect natural boundary morphology.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15110495

Authors: Lamya Albraheem

The integration of artificial intelligence (AI) with geospatial analysis, commonly referred to as GeoAI, provides new opportunities for renewable-energy planning and site-suitability assessment. However, limited attention has been given to integrating Multi-Criteria Decision Analysis (MCDA) with machine learning (ML) for offshore wind farm suitability assessment. This study proposes a GeoAI framework to predict offshore wind farm suitability using ensemble machine-learning techniques. A pooled technology-aware ML model was developed using nine predictors, including a turbine type to account for fixed-bottom and floating technologies. Gradient Boosting, XGBoost, LightGBM, Random Forest, Decision Tree, and Logistic Regression were evaluated, with a majority-class classifier included as a baseline. Model development and hyperparameter optimization were conducted using five-fold stratified group cross-validation based on geographic blocks, with weighted F1-score used as the primary model-selection metric. Gradient Boosting achieved the best model, with a weighted F1-score of 0.8077, followed closely by XGBoost (0.8031), and was therefore selected for final evaluation. SHapley Additive exPlanations (SHAP) analysis further revealed that wind power density, wind speed, and water depth were the most influential factors governing offshore wind farm suitability predictions.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15110492

Authors: Yuichi Yaguchi Yuki Nagasawa

This study extends hierarchical spatiotemporal identifiers beyond cell identification by using the Binary Spatiotemporal Identifier (BST-ID) as a computational representation for multi-cell regions. BST-ID supports independent resolution along x, y, altitude, and time; normalized finite prefix sets represent regions. We define subsumption, Union, anisotropic Intersection, selective Difference, refinement/coarsening, and neighborhood operations. Chained operations remain in the same BST-ID prefix domain. Full working-grid expansion provides reference semantics; selective execution depends on generated-prefix count K. All 43,444 bounded exhaustive, randomized, property-based, and targeted checks passed. In a synthetic 3D Difference case, selective traversal visited 4.97% of the full grid and reduced serialized size by 93.94%, preserving finest-grid occupancy. The real-world-derived Tsuruga Castle case normalized 31,334 fine XYH cells to 3884 prefixes (87.6% reduction), exactly reconstructing the declared fine-cell set. A dynamic UAS scenario with 34 simulated flights evaluated time-varying reservations, overlap, and conformance. Whole-scenario XYHT normalization reduced fixed-spatial dyadic-time records from 932,090 to 240,360; 75.72% of final prefixes had unequal X/Y zooms. The arbitrary exact-interval baseline remained more compact. Together, these results show that BST-ID provides representational continuity: independently resolved spatial and temporal prefixes support multi-cell region computation in the same prefix domain.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15110491

Authors: Md Jubier Ahammed Chao Xu Yaping Xu Hanlin Zhou

Daily nighttime light (NTL) images are widely used to map power outages and recovery, but clouds and quality screening leave gaps. We examined what NASA Black Marble data show about the February 2021 Winter Storm Uri outage in Harris County, Texas. When only high-quality retrievals were used, 16 February was the only usable night during the outage, and the next were 19 to 21 February. On 16 February, 5112 of 13,846 developed pixels (37%) were below half their pre-storm brightness, and 92% of those with an observed return were lit by 21 February. Return dates between these nights are unobservable. Recovery dates from a detector-based workflow depend on smoothing, persistence and aggregation choices; in a controlled experiment with known recovery dates and flag-0 observations only, an unsmoothed rule requiring three consecutive calendar nights has a median detection delay of 12 to 15 days. Compared with 20 ordinary clear nights, 16 February dimming was far more extensive and clustered. However, the estimated average total impact of the Social Vulnerability Index was 10.6 percentage points of reference radiance (95% interval −32.0 to 49.8), within the range estimated on ten ordinary pre-storm nights (−31.9 to 26.8). Single-night community comparisons need an ordinary-night baseline.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15110490

Authors: Lu Yu Qi Wang Peiran Kang Jingyuan Cheng

Map-platform visibility in historic urban districts does not necessarily increase in proportion to facility supply. This study develops a scale-adjusted spatial screening framework to identify locations where platform visibility is higher or lower than expected from POI quantity. The Gulou area in Kaifeng, China, is examined using 1461 points of interest (POIs) and API-derived rating and image fields. Thirteen POI categories are grouped into four functional types, and an entropy-weighted platform visibility index is constructed for 141 POI-containing grids at an approximately 200 m resolution. A cross-validated linear benchmark estimates the scale-based expectation of platform visibility, while a random forest examines nonlinear associations among functional characteristics. Consumption-stay facilities account for 58.38% of all POIs, whereas public-support facilities show substantially lower platform-field coverage. After scale adjustment, 20 high-supply–below-expected grids and 21 low-supply–above-expected grids are identified as candidate locations for field verification. Repeated cross-validation showed class-specific stability, with stronger persistence of the high-supply–below-expected stratum than of the low-supply–above-expected stratum, while threshold and grid-origin tests indicated uncertainty in exact spatial boundaries. The framework therefore treats platform records as a selective observation layer and uses scale-adjusted deviations to formulate spatial hypotheses for future field verification. Its practical usefulness for field auditing remains to be independently validated.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15110489

Authors: Patrick Davenport Levente Juhász

This study examines the spatial dynamics of climate-driven gentrification across 646 census tracts in Miami-Dade County, Florida, from 2010 to 2023, testing whether sea level rise exposure and elevation are associated with income growth, home value appreciation, and Black non-Hispanic demographic displacement, using American Community Survey data and NOAA 2.1 m inundation. County-wide Ordinary Least Squares (OLS) models indicate that elevation is not associated with income change, while elevation is significantly associated with Black non-Hispanic population decline in historically Black tracts (6.6 percentage points per meter where the 2010 share was at least 30%). Geographically Weighted Regression (GWR) indicates that this elevation–displacement relationship is spatially heterogeneous, concentrated in northern Miami-Dade and improves on OLS (mean local R2=0.57 vs. 0.35). Among high-elevation tracts (mean elevation > 2.1 m), GWR (R2=0.49 vs. 0.33) shows that baseline socioeconomic vulnerability is associated with gentrification intensity, while elevation retains a positive association. These findings extend climate gentrification theory, which centers on property value appreciation, by suggesting that demographic displacement was a more sensitive indicator than income or home value change in Miami-Dade. Nesting global county-wide and focused analyses suggests that climate gentrification may not be detectable at a single analytical scale, and income-based metrics may underestimate the phenomenon.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15110488

Authors: Han Wang Benkui Zhang Ying Chang Linlin Liang Yuanfei Chang Chengbo Wang

Geospatial target localization refers to determining the geographic location of targets from natural language queries, which is essential for applications such as urban management, emergency response, and location-based services. As user queries evolve from address matching toward complex forms incorporating multiple spatial semantic factors, the localization process increasingly demands the coordination of heterogeneous tools and data sources. Automation remains hindered by the lack of a unified structured representation for semantic factors, the difficulty of translating complex spatial constraints into executable operations, and the unreliability of single-round planning and execution. This study proposes SGTAgent, an agent framework for geospatial target localization from complex spatial semantic queries. The parser transforms queries into structured spatial semantic factors, the planner organizes dependent spatial workflows, and the feedback mechanism distinguishes local parameter refinement from workflow restructuring. Experiments show that SGTAgent achieves higher localization accuracy than the evaluated generic baselines across multiple LLM backbones. Using DeepSeek-V4.1-Flash, SGTAgent achieves localization accuracies of 39.86%, 68.76%, and 81.88% at distance thresholds of 100, 300, and 500 m, respectively, outperforming the evaluated agents under the corresponding experimental configurations. On complex queries using GPT-4o, it achieves an LA@500m of 80.47%, compared with 70.12% for ReAct. Ablation experiments show that removing the complete Restate–Replan feedback mechanism reduces the 500 m localization accuracy from 81.88% to 74.83%, demonstrating the effectiveness of the execution feedback mechanism under the evaluated configuration. An independent evaluation of the parser achieves macro-F1 scores of 89.76% on the original benchmark and 89.62% on independently formulated queries, indicating consistent semantic factor extraction capability across different query formulations.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15110487

Authors: Bowen Yang Canyu Tang Zhihao Zhang Feifan Yang

Timely congestion detection supports traffic monitoring and management, while spatial and temporal variations in road network conditions pose challenges to reliable detection. Most of the existing research on congestion detection focuses on the current traffic conditions of the road network and do not pay much attention to the factors with spatio-temporal characteristics, which usually change according to the situation of road network congestion. To this end, this paper proposes a novel spatio-temporal progressive convolution graph model for traffic congestion detection. The model introduces congestion factors and a spatio-temporal attention mechanism module to enhance the ability to model spatio-temporal factors and improve the effectiveness of congestion detection, thereby enabling the rapid identification of congestion areas in the road network. The experimental results on METR-LA and PeMS-Bay show five-level congestion classification accuracies of 75.3% and 78.9%, respectively, under the fixed speed-based labeling rule used in this study. The results support comparative evaluation under the fixed speed-based labeling rule used in this study; the congestion thresholds require local calibration before operational use on U.S. highways.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15110486

Authors: Runze Wang Yu Gu

Script-based linguistic-landscape monitoring records which languages appear on shop signs but not the design register in which they are rendered, so a business that keeps minority-language text while adopting mainstream or visitor-oriented visual conventions looks unchanged. We present a methodological proof-of-concept that treats signage design style as a geospatial monitoring object, using annual Kakao Road View cross-sections (2010–2024) of Wongok-dong, Ansan, Korea’s only Multicultural Village Special Zone. The pipeline combines YOLOv8 sign-panel detection, zero-shot CLIP ViT-L/14 classification into five design-style categories, post-stratified spatial weighting, crisis adjustment for closure signage, propagation of sampling and misclassification uncertainty, and spatial analysis (local spatial association and zone-boundary sensitivity). Raw style shares overstate pandemic-period contraction: the 2019–2021 decline in the weighted ethnic vernacular share is 14.4 percentage points on the raw metric but 10.8 points among operating premises. In the periphery, vernacular design gave way mainly to mainstream design; a shift toward visitor-oriented staged design in the commercial core is consistent in direction but rests on few panels and lies within sampling uncertainty. Design and script series diverge after 2019, showing what text-only monitoring misses. All associations are descriptive, and indicator thresholds for planning require validation across multiple enclaves.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15110485

Authors: Zengli Wang Yi Hu Ke Lei Xiang Li

Unstructured geographic descriptions containing spatial relations, such as landmark references, directions, and relative distances, remain a major challenge for accurate geocoding and spatial analysis. Existing approaches based on string matching or sequence modelling often fail to interpret complex spatial semantics embedded in human-generated addresses. This study proposes a knowledge-guided large language model (LLM) approach for geocoding-oriented address standardisation. The proposed approach represents addresses using a dual-layer structure consisting of a standard address that can be geocoded and ancillary spatial information. By incorporating explicit address structural knowledge into LLM prompting, five representative address patterns are constructed to guide semantic classification and structured extraction of address components, landmarks, directions, and distances. A deterministic spatial module is further developed to transform the ancillary information extracted by the LLM into coordinate constraints, thereby linking semantic interpretation with geometric computation. Experiments on 7931 real-world crime records from Nanjing, China, demonstrate that the proposed approach significantly improves geocoding accuracy across multiple mapping platforms. Compared with raw addresses, the method reduces both mean positional errors and error variability, with particularly notable improvements for orientation-based descriptions. The results indicate that combining LLM-based semantic parsing with spatial computation provides an effective solution for transforming unstructured geographic descriptions into machine-readable spatial information.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15110484

Authors: Jaewook Lee Yonghwan Lee Junhee Youn

Real-time sensing platforms such as closed-circuit television (CCTV) and unmanned aerial vehicles (UAVs) continuously produce large volumes of urban observation data, yet the path from such streams to map-based visualization is usually built case by case. This study proposes a standards-aligned framework for real-time urban geovisualization that connects heterogeneous sensing data to 2D and 3D map representations. The framework comprises (i) a generalized dynamic urban information model that structures pedestrians, personal-mobility devices, vehicles, and fixed objects through a common core schema with object-specific extensions; (ii) a multidimensional classification of dynamic information by spatial support, temporal behavior, representation dimension, and encoding class; and (iii) transparent, rule-based procedures that select the representation dimension (2D/3D) and the encoding profile—vector data for 2D networks and areas, batched and instanced 3D content for quasi-static backgrounds and repeated state-updating objects, and a time-dynamic format for trajectories—under the constraints of the implemented web-based prototype. The framework was instantiated across five urban-monitoring scenario categories comprising six visualization cases: moving-object trajectories, side-road and arterial-road congestion, pedestrian environment, street-tree vitality, and event-based crowd density. The analysis is a scenario-based instantiation and representation-coverage analysis; controlled performance benchmarks and user experiments were not conducted and are identified as future work. The results indicate that a unified data-model framework and selection procedure can serve heterogeneous urban sensing applications, providing a documented, reusable basis for real-time urban geovisualization services.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15110483

Authors: Jiyoung Kim

Converting building information models in the Industry Foundation Classes (IFC) into CityGML 3.0 brings building interiors into urban digital twins and knowledge graphs, yet converted models are often assessed only for schema validity. Converters either translate IFC elements one by one (direct translation) or rebuild thematic surfaces, rooms and levels of detail through interpretive decisions (semantic reinterpretation). This study makes those decisions explicit and measurable. We implemented a reinterpretation converter with switchable decisions, an area-preserving polygon reconstruction, and a converter-agnostic GML-to-RDF transformation based on the unmodified LIRIS CityGML 3.0 ontology, GeoSPARQL and PROV-O. On a railway-station model (IFC2x3, 23,861 products), we compared our converter with an open-source direct-translation converter against reference quantities from IFC geometry. For the 3649 objects not simplified in the default run, both converters preserved surface area equally well (95.4% and 95.2% within tolerance); prism simplification multiplied the summed volume of the simplified objects by 7.2. Every IFC product was either converted one-to-one or excluded by a recorded decision, whereas the direct translation’s output lacked 728 products. With curtain walls excluded and simplification applied, the reinterpreted GML was 26 times smaller and answered queries on room area in the standard vocabulary and on wall surfaces by standard class with auxiliary storey and area properties, which the direct translation could not. Controlled runs varying the two decisions independently attributed the volume loss and most of the size reduction to simplification; with it off, reinterpretation matched direct translation on all 23,133 common objects (98.7% versus 98.6% within tolerance).

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15110482

Authors: Xiaoqian Qiu Manchun Li Samuel Zhu A-Xing Zhu Zhenjie Chen

Urban functional area identification is fundamental to understanding urban spatial structure. Existing approaches typically assign a single category to each basic spatial unit, but are unable to fully represent variations in functional intensity. In this study, we introduced a refined urban functional area identification method that explicitly accounts for the heterogeneous spatial influence of points of interest (POIs) using Changzhou as a case study. The functional intensity of individual POIs was evaluated by integrating the entity scale, public recognition, and facility hierarchy, and an adaptive influence radius was constructed based on the functional intensity of individual POIs and their spatial clustering characteristics. A Gaussian distance-decay model was introduced to characterize the attenuation of the POI functional influence with increasing distance. The dominant functional areas, mixed-use functional areas, and functional transition areas were identified through functional structure analysis, function determination, and spatial aggregation. Comparison with land-use parcels showed that the method effectively identifies dominant functions. The identified functional categories, functional composition, and spatial variations in functional intensity within representative areas were highly consistent with the actual distribution of urban features. This study provides a new methodological framework for differentiated urban functional analysis and refined identification of urban functional areas.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15110481

Authors: Nhat-Duc Hoang

The surface urban heat island (SUHI) effect, driven by rapid urbanization and global climate change, has become a critical concern for urban environments, particularly in tropical coastal cities. This growing concern necessitates robust assessment and forecasting approaches to support effective mitigation and urban planning. This study proposes an integrated remote sensing–machine learning framework to model and predict SUHI intensity. An extreme gradient boosting (XGBoost) model is employed to quantify SUHI intensity and its relationships with various conditioning factors, including topographic variables, land use/land cover (LULC), land surface characteristics (bare land, built-up, tree canopy, and low vegetation densities), and proximity variables (distance to roads, rivers, lakes, and coastline). The coastal urban core of Hue City, Vietnam, is selected as the case study, with SUHI intensity modeled for the 2025 dry season. The projected LULC pattern for 2035 is generated using an integrated cellular automata–Markov chain–artificial neural network framework, which enables prediction of SUHI intensity under the anticipated urban growth scenario. Prediction results indicate that the maximum SUHI intensity is expected to increase from 12.74 °C in 2025 to 13.55 °C in 2035. The SUHI intensity map helps identify intra-urban hot spots experiencing severe surface heat stress. Notably, the area exposed to strong SUHI intensity (greater than approximately 6 °C) in 2035 is projected to increase by 10% relative to its extent in 2025. These findings emphasize the urgency of targeted SUHI mitigation and provide insights for climate-resilient urban planning in the study area.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15110480

Authors: Ahmed Abdelrhman Ahmed Mohamed A. Damos Muna Elsadig Abdolraheem Khader Ali Ahmed Ammar Ahmed E. Elhadi

Routing a medical-delivery UAV to multiple hospitals must jointly account for distance, flight time, payload, and clinical priority. We present a reproducible GIS-based framework for multi-stop UAV emergency medical routing in Khartoum State, Sudan, using a common obstacle-aware 3-D leg model with grid-based A* routing around no-fly zones and payload-dependent climb, cruise, and descent. We show that a linear priority term in a composite tour cost is order-invariant and cannot distinguish visiting sequences. We therefore use a priority-weighted completion-time objective with hard deadlines for Critical hospitals, with criterion weights derived from health professionals using the Analytic Hierarchy Process. On a 14-hospital network, exact Held–Karp and label-setting dynamic programming provide proven optima for benchmarking classical, metaheuristic, and Advantage Actor–Critic sequencing methods (A2C-SP). The revised objective reduces optimal priority-weighted completion time by 14% with only a 0.2% increase in distance. A2C-SP is 14.7% above the optimum across 47 unseen instances while generating routes in approximately 2 ms. Without retraining, it also plans capacity- and range-feasible multi-sortie missions on a 59-hospital city network, achieving a priority-weighted completion time of 110 min versus 137 min for the published toolbox route. At city scale, a single UAV cannot serve more than 5 of 16 Critical hospitals within 15 min, highlighting fleet sizing as a key constraint for meeting such deadlines.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100479

Authors: Li Zhang Wei Zhao Tengfei Han Qi Ouyang

Urban seismic resilience represents a central concern in disaster risk management, reflecting the profoundly destructive nature of earthquakes. Conventional engineering-based assessment methods are often inadequate to meet the demands of risk response in densely interconnected urban systems. In this context, complex network theory offers a valuable analytical framework, as it can reveal the interdependencies among system components and the mechanisms of cascading failure, thereby providing reliable support for risk management. Following the PRISMA guidelines, this study systematically reviews the literature published up to June 2026 and synthesizes core concepts, modeling approaches, assessment techniques, and practical applications. The field is evolving from early linear indicators toward coupled network modeling across urban systems and dynamic resilience simulation, a development enabled by artificial intelligence and digital twin technologies. Nevertheless, considerable challenges remain. The limited coupling between social and physical systems constrains quantitative assessment, while data fragmentation and oversimplified models undermine the reliability of the resulting evaluations. This study underscores the systematic role of complex network theory and identifies key directions for future research on the seismic resilience of urban systems. As technological capabilities advance, the integration of system coupling mechanisms, digital twin benchmarks, and AI-assisted decision support is expected to facilitate the development of implementable, next-generation tools suited to the era of artificial intelligence.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100478

Authors: Jiyoung Kim Jiwon Lee

Spatial question answering requires modeling relations among entities, yet metric georeferencing faces structural constraints in routinely available operational data: two-dimensional coordinates cannot distinguish floors within a building, rule-defined address relations require geometric approximation after coordinate conversion, and unregistered indoor spaces lack official geometry. We propose question answering over a knowledge graph derived deterministically from the assignment rules of the Korean road-name address system, in which number parity, order, and difference encode the opposite-side, same-side successor, and address-order interval relations. Mapping these rules onto graph edges yields a three-layer graph—reference frame, physical entities, and facilities—that resolves relational queries by edge traversal without coordinate computation. On a graph of 191,808 entities and 380,743 relation edges for Jongno-gu, Seoul, 93.9% of the rule-derived opposite edges were geometrically confirmed, and on an independently constructed 400-question benchmark the proposed method reached an F1 score of 0.381—under a structural ceiling imposed by administrative–physical binding coverage—versus 0.018 for relation-ablated retrieval and 0.103–0.119 for relation-verbalized retrieval. Queries on unregistered floors, which lack official geometry, reached an F1 score of 0.647; the advantage persisted through answer generation with three open 7–8B models (0.316–0.359 vs. 0.004–0.017); and applications to Sydney, Seattle, Paris, and Amsterdam yielded 0.521–0.746 across address-assignment schemes.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100477

Authors: Jianhua Jiang Hui Hu Ling Hua Zhihua Liang Cheng Zhong Hui Li

By providing a globally unified benchmark for urban land-use efficiency (LUE), SDG 11.3.1 establishes a soft constraint rooted in monitoring, comparison, and national commitment, thereby encouraging more sustainable land-use policies. Although extensive assessments have been reported, they often overlook errors originating from data sources, particularly their propagation—and potential amplification—during calculations within specific spatiotemporal contexts. This can introduce significant uncertainty, undermining both the reliability and equity of LUE evaluations and ultimately posing risks to the formulation of sound, sustainable policies. To address this gap, the present study conducts a meticulous examination of biases stemming from data sources, indicator formulation, and spatiotemporal scales, with Yangtze Delta, China, serving as the empirical case. The results indicate that when inappropriate indicators—such as the Land Consumption Rate to Population Growth Rate (LCRPGR)—are employed, even minor inaccuracies in the data sources can be drastically magnified. In contrast, indicators with simpler structures, like built-up area per capita, ensure minimal propagation of input errors. Furthermore, the choice of spatial units and the temporal trend analysis method can significantly influence the assessment results. Therefore, it is essential to move beyond the traditional approach of neglecting data errors and focusing solely on constructing sophisticated algorithms. Instead, the LUE assessment framework must be optimized across data sources, indicator formulas, and spatiotemporal conditions to mitigate error propagation, enhance the reliability of results, and thereby support the formulation of scientifically sound and equitable land-use policies.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100476

Authors: Wenpeng Lin Linxuan Zhao Yuanfan Zheng

Approaches for assessing the environmental outcomes of green development using quantitative indicators remain limited. This study integrates socioeconomic indicators with geospatial environmental information to assess green development level (GDL) and eco-environmental quality (EEQ) in a cross-provincial demonstration zone within China’s Yangtze River Delta (YRD) urban agglomeration covering three administrative districts and 28 townships from 2000 to 2020. GDL index was established using 38 indicators aligned with 13 Sustainable Development Goals (SDGs) across five sub-dimensions, while EEQ was measured using the Comprehensive Evaluation Index (CEI), a remotely sensed environmental index integrating PM2.5, land surface temperature, and vegetation coverage. The CEI represents environmental change relative to the year 2000 baseline rather than absolute environmental quality. Spatial and temporal analyses were conducted at the administrative district and township scales to examine the relationship between green development and environmental quality. The results showed that GDL increased consistently in all three districts from 2000 to 2020 in the study area, while EEQ followed a non-monotonic three-stage pattern, indicating a temporal mismatch between green development and environmental improvement. Moreover, this dual-index approach allows policymakers to distinguish regional development and environmental status within the administrative districts: (1) synergistic townships with high GDL and improved EEQ; (2) mismatched townships; and (3) underdeveloped townships with low GDL and degraded EEQ. Although uncertainties remain due to cross-scale data integration and dependence on baselines, this study offers a useful approach to support targeted green development and environmental management in urban agglomerations.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100475

Authors: Jilin Gu Diantao Fu Zhenhua Zhang Yanlin Zhang Xueming Li Yuan Gao

Accurate prediction of PM10 concentrations is essential for safeguarding public health and forms the foundation for effective collaborative control of regional air pollution. The generalizability of PM10 prediction models is often limited by their localized training and variable selection. Consequently, their predictive performance deteriorates in new geographical contexts and fails to achieve the generalizability required for nationwide deployment. A novel framework—sifting influencing factors, comparing basic models, optimizing parameters intelligently, and verifying cross-regional universality—is proposed because of the limitations. Based on correlation analysis and literature evidence, five atmospheric pollutant variables and four meteorological factors were identified as key predictors. Seven machine learning algorithms were compared using multiple metrics, and Support Vector Regression (SVR) performed best across 82 cities (R2 = 0.801, RMSE = 21.396 μg/m3, MAE = 12.365 μg/m3). The Sparrow Search Algorithm (SSA) then optimized SVR’s penalty and kernel parameters, raising R2 to 0.836 (RMSE = 19.104 μg/m3, MAE = 11.544 μg/m3). Independent validation across seven heterogeneous regions yielded R2 values from 0.778 to 0.962 (mean 0.854), with RMSE of 16.712 μg/m3 and MAE of 8.910 μg/m3. These results demonstrate strong generalization of the SSA-SVR model across multiple heterogeneous regions, supporting its potential nationwide applicability to PM10 prediction.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100473

Authors: Tong Chen Bingxue Wei Jieheng Cui Yan Yu Chenghe Zhang

Urban heat island networks characterize the spatial organization and potential connectivity of relatively high-temperature patches, but multi-temporal analyses are constrained by the limited availability of fine-resolution and temporally dense land surface temperature observations. This study develops a dynamic analytical framework integrating LST reconstruction, morphological spatial pattern analysis, circuit theory, and multidimensional network indicators. Landsat 8/9 LST and daily 1 km TRIMS LST were integrated using STARFM to reconstruct daily 30 m LST for May–October during 2019–2023, which was subsequently aggregated to monthly mean LST for repeated UHIN identification. The reconstructed LST showed close agreement with Landsat observations, with an overall RMSE of 2.4675 °C and an SSIM of 0.9883. At the interannual scale, warm-season mean PLAND ranged from 16.38% to 18.16%, while PD, ACL, and the β index showed different temporal responses. At the intra-annual scale, increases in relative source-area extent toward August were generally accompanied by decreasing patch density, followed by renewed fragmentation later in the warm season. These asynchronous variations indicate that relative source-area extent alone cannot fully characterize UHIN changes. The proposed framework provides a consistent basis for multi-temporal UHIN analysis and network-informed urban heat-mitigation planning.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100474

Authors: Shuo Zhu Ying Li Jun Chen Xuan-Meng Cao

Advances in digital technology have improved virtual museum display interaction and user experience. Nevertheless, significant challenges remain in designing information interaction within existing virtual museums. These include inefficient browsing paths, low information reception efficiency (indicated by a total gaze duration of 0.726 in eye-tracking tests), and a limited range of interaction methods. To address these issues, this study applies spatial intelligence algorithms to optimize navigation paths and visual interaction in the “Five Kilns of the Song Dynasty” virtual museum. Shortest path planning is implemented using Dijkstra’s algorithm. Additionally, the A* algorithm and the Threshold Filtering algorithm, in conjunction with the actual exhibition layout and user interests, are employed to optimize paths for maximum information density. Visual interaction optimization is achieved through a display detail adjustment algorithm and a gaze interaction matching algorithm, which adjust exhibit display details and system interaction based on user position and gaze direction. Eye-tracking experiments examine the effects of virtual museum navigation path and visual interaction optimization on user information perception, with particular attention to improvements in visual information reception efficiency, information search efficiency, and cognitive effort and interest. These findings offer scientific, practical references for designing information exchange and user experiences in virtual museums focused on cultural heritage or handicrafts.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100472

Authors: Zhengduo Xiao Chiu-Wing Sham Ka-Shing Cheung

Accurate and timely Point-of-Interest (POI) information is important for location-based services, tourism management, and digital map maintenance. However, newly opened or weakly documented POIs may be updated slowly because conventional map maintenance often relies on user reports, field surveys, and manual verification. This problem is especially relevant to tourism-centric towns, where POI update demand exists but large map-service providers may prioritize dense metropolitan areas with higher user activity and stronger commercial returns. This study addresses mobility-driven early warning for candidate POI updates under cold-start conditions, where candidate locations lack stable category labels and long-term visit histories. We propose MobiPOI, a framework that combines anonymized mobile location observations, map-derived spatial boundaries, and shared-visitor priors. Candidate visits are first constructed within geographic boundaries to reduce roadside contamination and spatial spillover. Behaviourally related mature POIs are then identified through shared visitors and activity-rhythm similarity. These mature peers provide dwell-time references for visit filtering and temporal priors for cold-start verification. A prior-guided interval gated recurrent unit (GRU) further combines the cleaned candidate sequence, peer-derived temporal prior, and shared-visitor acquisition velocity to estimate dynamic visitor-volume intervals. Experiments are conducted in two New Zealand case-study regions, Auckland and Queenstown, representing metropolitan and tourism-centric mobility settings. Using delayed temporal OpenStreetMap (OSM) snapshots and manual map-based inspection for post hoc validation, the results show that MobiPOI improves prediction-interval reliability and candidate POI warning efficiency compared with spatial clustering, fixed-threshold, time-series, and prior-transfer baselines.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100471

Authors: Huayi Wu Lutong Xie Shuyang Hou Zhangxiao Shen Haoyue Jiao Shaowen Wu Guanyu Chen Ziqi Liu Donghai Liu Jianyuan Liang Yaxian Qing Rui Li Zhipeng Gui Xuefeng Guan

The rapid advancement of large language models (LLMs) offers new avenues for overcoming the limitations of current data-driven GIS analysis methods, giving rise to the emerging field of large language model-driven GIS analysis (LLM4GIS). This study presents a systematic review of 673 relevant research articles published between 2022 and July 2026. By tracing the five-stage evolutionary trajectory of GIS analysis methods and elaborating the framework for LLM4GIS, we conduct a quantitative bibliometric analysis that reveals the field’s rapid growth from sporadic exploration to explosive expansion, along with the shifting landscape of its dominant technical trajectories. Building on these analyses, we develop a taxonomy spanning four dimensions—technical approaches, datasets, applications, and evaluation—and systematically analyze the advances and evaluation methodologies across five core application areas: geospatial knowledge question answering, knowledge extraction, spatiotemporal reasoning, geospatial code generation, and analysis modeling. Key challenges and future research directions are further discussed along five dimensions: multimodal data integration, high-quality data development, domain-specific model development, integration of world models with real-world geospatial environments, and interpretability and trustworthiness. This review provides a systematic reference for advancing the bidirectional synergy between LLM4GIS and GIS4LLM and for fostering a new generation of spatiotemporal intelligence ecosystems.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100470

Authors: Xiaofang Zhai Fan Zhang Zhifeng Xiao Haoyang Xu Weiqi Yan

Against the backdrop of the continuing spread of cross-border telecommunications fraud and the scarcity of conventional data, this study examines how publicly accessible multi-source data can be used to monitor and characterize the spatiotemporal evolution of known or suspected fraud compounds under data-scarce conditions. Focusing on eight representative and manually delineated fraud compounds in Myanmar’s border regions, it develops a three-dimensional “space–time–representation” framework that integrates 2017–2025 VIIRS nighttime lights (NTL), Sentinel-1/2-derived built-up area (BUA) expansion, and Baidu Index time series as a proxy for Chinese-language public attention. Dynamic time warping (DTW) is used to examine temporal alignment between NTL dynamics and public attention, while spatial autocorrelation and exploratory grouping based on DTW dissimilarity are applied to characterize spatial clustering and heterogeneous temporal patterns. The results show that BUA expansion and nighttime brightness are not uniformly coupled across compounds and years. After excluding partial-year 2025 observations, the complete-year 2017–2024 park–year analysis indicates weak aggregate associations between BUA and annual mean NTL, highlighting the need to interpret the two indicators as complementary but non-equivalent proxies. NTL changes do not necessarily evolve synchronously with changes in public attention, suggesting potential value for monitoring and prioritization rather than direct early-warning or verification of illicit activities. The global Moran’s I of 0.2476 indicates significant spatial clustering of nighttime-light values within the defined compound–buffer analysis units. DTW alignment reveals stronger consistency between nighttime lights and Baidu Index attention in the Daxiang and Dingsheng compounds, whereas KK Park and Yatai City show weaker public-search responses relative to observed NTL dynamics. Exploratory grouping identifies three patterns of NTL–public attention correspondence, highlighting heterogeneous relationships between observed nighttime-light dynamics and external information visibility. The study demonstrates the complementarity of remote sensing and social sensing data under data scarcity and proposes a replicable approach for remote monitoring, typological characterization, and evidence-informed governance prioritization of known or suspected fraud compounds. The proposed approach should be interpreted as a screening and analytical framework that provides directional evidence rather than as direct verification of illicit activity or compound-specific operational status.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100469

Authors: Zhenyu Wang Yunfa Wu Fengliang Tang Weixing Xu

Urban vitality is often measured within predefined administrative units, blocks, or grids, limiting its translation into planning-oriented spatial diagnosis. This study examines 34 vitality clusters in the Nanjing Metropolitan Area and develops a multi-scale framework linking cluster identification, spatial interpretability, type stratification, morphological differentiation, micro-scale case examination, and planning translation. The identified clusters show substantial correspondence with built-up territories while extending across administrative boundaries in several cases. Five operational area–density groups are distinguished using study-specific thresholds. Their distribution reveals a gradient from high-density cores to medium-density and peripheral low-density clusters, together with an internal pattern of high-vitality cores and lower-vitality edges. After Benjamini–Hochberg correction, 30 of 60 non-redundant variables remained significant, mainly in functional, transport, and morphological dimensions. Micro-scale cases further show the co-occurrence of smaller blocks, continuous frontages, mixed functions, and favorable walking conditions. These findings suggest that vitality-derived clusters can serve as morphologically interpretable, activity-based diagnostic units that complement conventional planning units in stock-based urban regeneration.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100468

Authors: Yiming Li Xianpu Liang Guangming Guo Peng Su Qi An

Visual simultaneous localization and mapping (SLAM) often degrades in low-light environments because of insufficient brightness, reduced contrast, and loss of texture, which cause sparse features, accumulated mismatches, and unstable trajectories. This paper presents RetiGlue-SLAM, a visual-inertial SLAM framework that retains the third version of ORB-SLAM (ORB-SLAM3) as the system backbone and improves the front end through task-driven coupling of three components: Retinexformer for low-light image enhancement, SuperPoint for learned local feature extraction, and LightGlue for robust cross-frame matching. The enhanced images and learned correspondences are integrated into the tracking thread, whereas the geometric back-end and inertial optimization of ORB-SLAM3 remain unchanged. Experiments on low-light sequences from EuRoC and the night-time Old Town sequence of 4Seasons show improved feature matching, tracking completion, and localization accuracy relative to the baseline. A two-stage ablation separates the learned front end from the full RetiGlue-SLAM pipeline, and a module-level runtime analysis reports a front-end latency of 41 ms per stereo frame (24.39 FPS). The results support RetiGlue-SLAM as a benchmarked low-light visual-inertial localization framework; field deployment and end-to-end real-time operation remain outside the scope of this study.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100467

Authors: Jiyoung Kim Hyojoong Kim

Urban function labels in multi-dimensional geo-data cubes (GDCs) assign a single dominant function to each voxel, discarding within-voxel composition. This study extends multi-dimensional grid-point (MDGP) scaling from isotropic two-dimensional rasters to anisotropic three-dimensional urban voxels built from building-register floor-level use records. The extension separates planar and vertical aggregation factors, represents voxel composition with gross floor area (GFA), and computes GFA-weighted information retention. The procedure converts each floor-level record into a three-dimensional point, allocates it to an anisotropic voxel, sums GFA by functional category to obtain a continuous composition vector, assigns each voxel to the simplex grid point that maximizes the Czekanowski retention, and then applies a representativeness threshold and adaptive partitioning before the cell-level and global retention are computed. The method was applied to 2,737,362 floor-level use records from 552,560 buildings across Seoul. In a controlled comparison on uniform grids before vocabulary pruning, the three-part mixed label improved information retention over GFA-based majority assignment by 11.63, 16.47, and 20.17 percentage points at 125 m × 8 m, 250 m × 16 m, and 500 m × 32 m, respectively. At 250 m × 16 m, the final adaptive, thresholded configuration retained 89.37% of the GFA composition on the refined grid, against 88.85% for the same label on the uniform grid before partitioning and pruning, and 72.38% for the single-label baseline on that uniform grid. Gains were stable across 30 random grid origins. A cell-level analysis of officially assessed unit building prices showed higher complexity-adjusted association for mixed labels: the complexity-adjusted ω2 rose from 0.105 to 0.254 and the out-of-sample R2 from 0.443 to 0.501 under spatial-block cross-validation, with the root-mean-square error of the log price falling from 0.480 to 0.454. These results show that mixed labels preserve more within-voxel functional composition under the tested Seoul data and aggregation settings; validation in other cities remains necessary.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100466

Authors: Mingke Erin Li Mozhou Gao Steve H. L. Liang

Heterogeneous gridded environmental datasets are increasingly used in Earth system modeling, atmospheric studies, and spatial analytics, but differences in coordinate reference systems, spatial resolutions, sector classifications, reporting units, and data formats hinder interoperability and cross-dataset comparison. Conventional latitude–longitude grids further introduce unequal cell areas that bias aggregation and regional comparison. This study presents a framework for harmonizing heterogeneous gridded datasets within a Discrete Global Grid System (DGGS), using explicit grid-selection criteria, area-weighted mass-preserving conversion, cross-scheme sector mapping, and five measurable evaluation criteria. Using methane as a demanding case, we transformed 13 open-access inventories from global to subnational scales into the rearranged Hierarchical Equal-Area isoLatitude Pixelization (rHEALPix) DGGS, harmonizing 324 source variables from five classification schemes and standardizing values to megagrams per year per cell. Conversion preserved emission totals, with negligible differences for continental, national, and regional inventories and differences within 2% to 3% for global inventories. Cell area is constant by construction, aggregation along the hierarchy is exact, and inventories at different resolutions can be compared after aggregation. The framework is reproducible and extensible to other gridded spatial data.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100465

Authors: Norberto Carrillo-Garcia Marco Moreno-Ibarra Yanil Contreras-Jimenez

The digital transformation of urban environments has generated a hybrid space where physical distance and digital reputation concurrently shape consumer behavior. Traditional spatial interaction models, such as the classic Huff model, rely heavily on static socioeconomic data and geographic friction, omitting the growing influence of online user perceptions. This paper introduces a hybrid commercial gravitation framework that incorporates Volunteered Geographic Information (VGI) through natural language processing (NLP). Using a dataset of coffee shops in the Cuauhtémoc municipality, Mexico City, we computed a composite digital attraction indicator (Wj) by multiplying the text sentiment intensity extracted via pysentimiento (a RoBERTa-based architecture) by the logarithm of review volume. The model’s predictive performance was validated against real-world consumption proxies using Spearman’s rank correlation (ρ), competing against Nearest Facility (Voronoi), classic Huff, and XGBoost algorithms. The proposed hybrid framework achieved a remarkable correlation (ρ=0.8647,p<0.001), drastically outperforming the classic Huff model (ρ=0.2782) and the deterministic Voronoi approach (ρ=−0.1580). While XGBoost achieved a perfect empirical fit, it lacked spatial interpretability. These findings demonstrate that combining linguistic sentiment with physical spatial friction provides a highly predictive, explanatory, and theoretically robust framework for modern geomarketing and urban planning.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100464

Authors: Yao Wu Haochen Feng Yichang Shao Junzhuo Hu

Autonomous vehicle collision reports provide valuable geospatial information for understanding how emerging mobility systems interact with urban built environments. However, existing studies have mostly examined autonomous vehicle crashes from a global statistical or traffic safety perspective, with less attention paid to the construction of spatial analysis units, multi-source geospatial data integration, and local spatial non-stationarity. This study develops a Traffic Analysis Zone (TAZ)-based geospatial framework to analyze the spatial distribution of autonomous vehicle collision reports in San Francisco. A total of 427 collision reports from 2015 to 2024 were geocoded and integrated with road infrastructure, land-use, and geographic information data within a unified Geographic Information System (GIS) environment. Several global and spatial regression models, including negative binomial regression, generalized linear models, spatial autoregressive modeling, and geographically weighted regression, were compared to examine the association between built-environment variables and collision report counts. The results show that autonomous vehicle collision reports are spatially clustered and that several road infrastructure variables, including bicycle lanes, traffic signs, and intersections, are associated with their spatial distribution. The geographically weighted regression results further indicate that these associations vary across San Francisco, suggesting clear local spatial non-stationarity. This study contributes a GIS-based workflow for integrating autonomous vehicle collision reports with urban spatial data and provides exploratory evidence for spatially targeted safety diagnosis in smart city and urban mobility contexts.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100463

Authors: Limin Guo Yukang Guo Xing Su Zhi Cai Ying Liu Wenjie Tian

Next point-of-interest (POI) recommendation seeks to predict the POI that a user will visit next from historical check-in records. Accurate recommendations require modeling POI relations and user trajectory patterns. Existing methods remain limited in representing continuous spatial influence decay, reliability-filtered temporal associations, semantic hierarchies, and multi-membership attributes. Furthermore, user trajectories may contain relatively persistent behavioral regularities and short-term variations within a check-in sequence, while standard time-domain models do not explicitly decompose trajectory representations by variation scale. Accordingly, this study introduces the Dynamic Hierarchical Attention Fusion Spatio-Temporal Network (DHAFST). Our Adaptive Dual-Graph Convolutional Network (ADGCN) employs Gaussian kernels with dataset-specific bandwidths to model continuous spatial influence and uses activity- and co-occurrence-based reliability filtering to construct temporal associations. The Bidirectional Hierarchical Graph Attention Network (BiHGAN) estimates soft memberships through a nested Gaussian mixture hierarchy and constructs cross-level affiliation relations for bidirectional hierarchical propagation. The Attention-Frequency Fusion Transformer (AFFT) combines time-domain self-attention with FFT-based low-frequency reconstruction and high-frequency residual modeling through an adaptive gate. Experiments on three real-world check-in datasets show that DHAFST achieves the highest Acc@1 among the evaluated methods on all three datasets.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100462

Authors: Longlong Xiang Lan Wu Ming Yang Qian Chen Lintong Han

City-scale UAV planning requires a clear distinction between areas screened out by operational constraints and areas retained for graded suitability assessment. To address this need, this study develops a grid-based framework for near-ground airspace up to 120 m above ground level in Yangzhou, China, and examines the resulting spatial patterns. Multi-source geospatial and regulatory data are integrated into regular hexagonal cells with a side length of 100 m. The framework first applies no-fly zone-related constraint screening and then evaluates the remaining cells using nine suitability indicators weighted by the Criteria Importance Through Intercriteria Correlation method. Global and local Moran’s I are used to identify spatial dependence, local clusters, and spatial outliers, while ordinary least squares and geographically weighted regression characterize global and spatially varying associations. The suitability index exhibits significant positive spatial autocorrelation (Moran’s I = 0.738, p = 0.005), with continuous high–high and low–low clusters and localized transition zones. Compared with OLS, GWR increases R2 from 0.876 to 0.911 and reduces residual Moran’s I from 0.275 to 0.127, providing evidence of spatial non-stationarity across urban subregions. Overall, the framework integrates constraint screening, graded suitability assessment, and spatial heterogeneity analysis into a consistent geospatial decision-support structure for differentiated pre-operational UAV deployment and urban low-altitude airspace planning.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100461

Authors: Jungjoo Lee Nahye Cho Kirim Lee

Greenbelt monitoring requires both the detection of physical change and its interpretation against parcel and administrative records. This study develops a multistage GeoAI–GIS workflow for that purpose and examines its application in Siheung, Republic of Korea. Satellite, aerial, and unmanned aerial vehicle imagery support regional change screening and detailed mapping, while cadastral attributes and municipal registers provide the context for selecting parcels for review. A metric-learning change module trained with triplet loss, a YOLOv8-based structure segmentation module, and a SegFormer land-cover module address separate analytical tasks. Their internal validation F1-scores were 87.3%, 88.4%, and 84.8%, respectively. Class-specific results showed stronger performance for forests, buildings, and greenhouses than for installations, stockpiles, farmland, and bare land. Spatial integration converted the predictions into parcel-linked geometries, attributes, and reasons for administrative examination. For the building- and greenhouse-related comparison, the workflow produced 2486 candidate parcels, whereas the manual aerial-photo interpretation dataset contained 1836 cases. The difference of 650 records, or 35.4%, describes candidate volume rather than additional confirmed violations or an increase in measured recall. The study demonstrates a practical connection between image analysis and parcel-based review, and identifies class reliability, spatial dependence, and the distinction between physical change and legal status as central to its interpretation. Automated screening supports the preparation of evidence; permit assessment and decisions about enforcement remain administrative responsibilities.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100460

Authors: Zhiwei Zhu Yuxuan Li Xinxin Zhang

Understanding the complex spatiotemporal heterogeneity between ride-hailing demand and the urban built environment is crucial for optimizing urban space and allocating transportation resources. Previous analyses using fixed spatial scales (e.g., regular grids or administrative boundaries) often suffer from the modifiable areal unit problem (MAUP) and fail to capture the actual distribution of ride-hailing demand. Furthermore, traditional geographically weighted regression (GWR) models struggle with nonlinear spatial relationships. To address these challenges, this study uses Xiamen, China, as a case study. We employ the K-means++ clustering method on ride-hailing origin–destination (OD) data to generate dynamic spatial partitioning units, integrating them with points of interest (POIs), land use, and population data. A geographically neural network weighted regression (GNNWR) model is then introduced to quantify how built environment factors influence mobility, explicitly capturing nonlinear variations. Results show that dynamic spatial partitioning effectively mitigates the MAUP: the traditional GWR R2 improved from 0.414 (administrative units) and 0.577 (regular grids) to 0.649. Using these dynamic clustering units, the GNNWR model achieved an average R2 of 0.750, outperforming multiscale GWR (MGWR, 0.614) and traditional GWR (0.596) by approximately 22.1% and 25.8%, respectively. Empirically, the built environment’s impact exhibits pronounced spatiotemporal nonstationarity. Residential and office areas reveal distinct tidal commuting patterns driven by job-housing separation, while public transit density shows a dual substitution–complementarity relationship with ride-hailing. Ultimately, this research validates GNNWR’s applicability for microlevel geographic analysis, providing a robust scientific basis for differentiated traffic management and precise urban spatial planning.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100459

Authors: Liming Gao Lanqing Wang Qian Wang Liming Zhang

Zero-watermarking technology constructs copyright identifiers by extracting stable features from the original data without modifying the data itself, making it highly suitable for lossless copyright protection of high-precision vector geographic data. However, existing methods rely on topology construction, triangulation, or complex handcrafted features, resulting in high computational overhead and insufficient generalization capability for multiple types of vector data. This paper proposes a zero-watermarking algorithm that combines a geometric encoder with multi-channel geometric representation. The method encodes vector data into a four-channel geometric tensor consisting of an occupancy field, distance field, orientation field, and density field. Then it uses a geometric encoder to extract stable and distinguishable features. Joint training with identity classification, feature consistency, and triplet uniqueness constraints ensures feature stability before and after attacks as well as strong discriminability across different data. Experimental results show that the proposed method remains stable under rotation, translation, and uniform scaling attacks, and still retains a certain level of authentication capability under feature deletion. Meanwhile, the method can effectively distinguish different vector data identities and construct zero-watermarks with relatively low computational overhead, demonstrating good uniqueness and computational efficiency.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100458

Authors: Runhao Zhang Qin Li Chongbo Ma Yijun Liu Lixin Jia

To address the shortage of officially designated shelters and the uncertain usability of candidate open spaces in high-density historic urban districts, this study develops a GIS-based framework integrating open-space screening, road network analysis, network-based travel time accessibility, the enhanced two-step floating catchment area method (E2SFCA), and multidimensional validation in Xicheng District, Beijing. Ten official shelters and 173 candidate shelters were identified. At 72 m/min, 15 min population coverage increased from 23.98% for official shelters to 97.72% for the combined configuration. Independent coefficient perturbation produced a median Spearman correlation of 0.9945, but random failure of 20% and 40% of candidates reduced the median correlation to 0.9172 and 0.8126, respectively. Capacity-constrained assignment served 77.11% and 54.65% of the population at 1 and 2 m2/person. A hutong-speed penalty, multiscale catchments and decay, symmetric boundary buffers, exact network closeness, and a 30-site image check further define the conditions under which the candidate system remains effective. The framework separates statutory provision from planning-stage candidates and identifies where shelter verification, spatial supplementation, and route or entrance improvements are most needed.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100457

Authors: Bader Alshehri Glenn Campbell Zhenyu Zhang Xiaoye Liu

Urban expansion in arid environments remains insufficiently understood, particularly where growth occurs with limited competition for agricultural and vegetated land. This study examined long-term Land Use and Land Cover (LULC) change, land transformation, and urban expansion in Riyadh, Saudi Arabia, from 1985 to 2025 using multi-temporal Landsat imagery, Random Forest (RF) classification, and Google Earth Engine (GEE). Five decadal benchmark years were analyzed to quantify urban growth dynamics, land transition pathways, and spatial patterns of expansion. Built-up land increased from 250.06 km2 in 1985 to 1713.98 km2 in 2025, representing a net increase of 1463.92 km2 (585.42%). Expansion occurred predominantly through the conversion of barren desert land into built-up surfaces, while vegetation and water contributed comparatively little to urban land conversion. Temporal analysis additionally showed increasing absolute territorial expansion despite declining proportional growth rates in later decades, providing complementary evidence of changing growth dynamics as the metropolitan footprint enlarged. These findings characterize Riyadh as a context-specific expression of metropolitan growth in a hyper-arid setting, where extensive barren-land availability conditions the dominant land-transformation pathway. The study provides a reproducible framework for long-term urban monitoring and supports evidence-based urban management and planning in rapidly expanding arid cities.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100456

Authors: Ramazan Yoldaş Satılmış Recep Çakır

Wildfires are increasingly shaped by climate change, as rising temperatures alter fire regimes. Accurate prediction of wildfire susceptibility under warming scenarios is essential for effective risk management and land-use planning. This study develops a Geographic Information Systems (GIS)-integrated Artificial Neural Network (ANN) framework, optimized via Bayesian hyperparameter tuning, to predict wildfire susceptibility and assess its spatial response to incremental temperature increases. The proposed methodology integrates multi-source geospatial data (Sentinel-2 NDVI, SRTM terrain models, Copernicus ERA5 reanalysis, and VIIRS active fire observations) within a GIS environment, coupling event-based feature extraction with the Bayesian-optimized ANN, which was trained on climatic, topographic, vegetation, and environmental predictor variables. Model evaluation showed that smaller, balanced training datasets achieved stronger generalization than larger ones. Scenario-based analyses simulated temperature increases of +0.5 °C, +1.0 °C, +1.5 °C, and +2.0 °C, reflecting global climate model projections, with all other variables held constant. Results show that warming intensifies susceptibility in existing fire-prone regions and drives the emergence of new wildfire-prone areas with no historical fire occurrence. Relative to current conditions, newly predicted fire-prone locations increased by approximately 36%, 51%, 85%, and 118% across the respective scenarios, indicating a nonlinear susceptibility response to warming. These findings underscore the value of incorporating climate scenarios into GIS-integrated machine learning-based wildfire susceptibility modeling and offer a spatially explicit framework for proactive mitigation and climate-resilient land-use planning.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100455

Authors: Rabia Bovkir Tonbul

Rapid urbanization and the replacement of vegetated surfaces by impervious materials can substantially alter urban thermal environments, while water bodies, topography, and coastal setting may further modify these patterns. Understanding how these factors contribute to land surface temperature (LST), particularly across large and geographically heterogeneous cities, is important for spatially targeted heat mitigation. This study investigates the spatial variation in summer LST across Istanbul, Türkiye, using an explainable geospatial machine learning framework that integrates satellite observations with blue–green, built-environment, topographic, and coastal variables. Landsat 8/9 observations from June to August 2020–2025 were integrated with multi-source land cover, imperviousness, topographic, and coastline data within a common 500 m grid. Following predictor screening and comparison of five tree-based algorithms, XGBoost was applied to the full study area and evaluated using nested spatially blocked cross-validation. The model achieved a mean R2 of 0.9531 ± 0.0083, RMSE of 1.0616 ± 0.0293 °C, and MAE of 0.7734 ± 0.0201 °C across the five outer folds of the nested spatial cross-validation. Predictive performance decreased under the larger 20 km block configuration, indicating sensitivity to the spatial scale of validation. SHapley Additive exPlanations (SHAP) identified the Normalized Difference Vegetation Index (NDVI), used here as a spectral proxy for vegetation greenness rather than as a direct measure of vegetated area, as the largest contributor to the fitted model, followed by imperviousness, water body cover, and grassland. The importance of water body cover was sensitive to coastal grid geometry and decreased substantially when low-land-fraction coastal cells were excluded. Because vegetation-related information is also used in the emissivity adjustment of the Landsat Level-2 surface temperature product, the magnitude of the NDVI attribution is interpreted as model-based rather than as an independent estimate of vegetation cooling. A sensitivity model excluding the NDVI retained substantial predictive performance (R2 = 0.8063), with imperviousness becoming the leading predictor. A sensitivity analysis replacing the NDVI with independently derived tree cover percentage yielded lower model performance (R2 = 0.9160). Dependence analyses revealed nonlinear relationships between the main predictors and LST. Spatial SHAP mapping further showed that although the NDVI had the largest local SHAP attribution across most of Istanbul, water-related, coastal, and topographic factors became locally important under specific geographic conditions. These findings show the value of spatially explicit explainable machine learning for identifying both city-wide and local factors associated with urban surface temperature.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100454

Authors: Anette Ganske Angelina Kraft Alexander Wolodkin

Advancing Earth System Sciences (ESS) depends on our ability to integrate highly diverse data from a range of disciplines, including geographical information sciences, palaeontology, marine science, biodiversity research, atmospheric sciences and biology. A major scientific obstacle to this integration is semantic heterogeneity: different disciplines use different methods and terms to describe data, resulting in fragmented vocabularies. We address this challenge by providing the ESS terminology collection, which is a source of state-of-the-art terminologies for this scientific community. This collection already contains 50 curated terminologies. New terminologies can be suggested by scientists at any time. The ESS collection enables the consistent annotation and semantic alignment of heterogeneous ESS research data. It enhances the FAIR data principles—particularly findability, accessibility, interoperability, and reusability—thereby facilitating interdisciplinary research and data reuse. This collection enables the ESS community to assess the quality and completeness of ontology metadata and the extent to which their terms comprehensively represent the scientific content of a field.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100453

Authors: Siming Zheng Wentao Yan

Urban waterfronts serve as important recreational spaces for both residents and tourists. However, differences between these groups in recreational behavior and environmental influences remain insufficiently understood. Using mobile phone signaling data from 16–22 July 2021, this study compares resident and tourist waterfront visitation in central Hangzhou during a summer peak-season week. The “behavior–structure–correlates” framework combines spatiotemporal analysis, social network analysis, K-means clustering, and Geodetector. Residents accounted for 81.10% of all recorded waterfront visits, but the resident–tourist composition varied across waterfront systems: tourists contributed 37.95% of visits to West Lake, compared with 17.80% to the Grand Canal and 9.53% to the Qiantang River. Residents showed proximity-oriented and relatively dispersed visitation patterns, whereas tourists concentrated around iconic attractions. The resident recreation flow network was denser than the tourist network, with densities of 0.43 and 0.20, respectively. Core waterfront units function as both major attractions and intermediary nodes, while key paths demonstrate the guiding role of major transport corridors. Metro station availability showed the highest explanatory power for resident visitation (q = 0.226), while surrounding accommodation density ranked highest for tourist visitation (q = 0.307). Interaction analysis further highlighted combinations of waterfront resources and public transport for residents, and urban functional diversity and metro availability for tourists. By linking group-specific behaviors, spatial structures, and environmental correlates, the framework provides empirical support for matching differentiated recreational demands, implementing service-type-based waterfront management, and improving waterfront service efficiency in tourist cities.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100452

Authors: Yang Jiao Suhaibah Azri Uznir Ujang

Learned indexes are presented as a new direction for spatial data access, yet their transfer to native 3D geospatial and point-cloud tasks remains unclear. Building on existing reviews and benchmarks, this review connects index structures with geospatial application scenarios, query requirements, and benchmark dimensions. Following PRISMA 2020 reporting guidance, it combines searches of Scopus, Web of Science, IEEE Xplore and ACM Digital Library with revision-stage reference checking. After eligibility assessment and consolidation of overlapping reports, all 111 retained studies were included in the qualitative synthesis. The review develops a unified taxonomy of traditional and learned spatial indexes, including hybrid designs, along two analytical dimensions, rule-driven and learning-driven designs, and 2D and 3D data support. The synthesis reveals an imbalance: learned indexes are widely evaluated in 2D and general multidimensional settings, whereas native 3D geospatial indexing remains dominated by traditional and hybrid structures. The review reframes the learned–3D gap as a workload-adaptation and benchmark-standardization problem. Finally, it identifies design requirements and benchmark dimensions for geospatial-aware learned 3D indexes, including density-aware partitioning, geometry-preserving learning, exact–approximate refinement, update-aware maintenance, out-of-core scalability, and system integration.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100449

Authors: Yinglian Gao Wei Hua Beibei Zhao Binhao Du Zhenhua Gao Miaole Hou

Ancient sites are vulnerable cultural heritage assets that are increasingly threatened by climate change, natural erosion, and anthropogenic pressures, posing significant challenges to the stability and integrity of their landscapes. Therefore, the identification and systematic assessment of landscape risks at ancient sites are essential for preventive conservation and sustainable landscape management. To address the limitations of risk assessment and long-term dynamic risk analysis of landscape risks at ancient sites, this study develops a landscape risk assessment indicator system based on the Pressure-State-Response (PSR) framework, incorporating three dimensions: natural environment, human activities, and site protection. The entropy weight method was employed to determine indicator weights from the relative dispersion of the standardized indicators. From a landscape-scale perspective, this study systematically analyses the landscape risk of ancient sites in Yuncheng City from 1990 to 2024. The results indicate the following: (1) The landscape risk of ancient sites in Yuncheng exhibits significant spatial heterogeneity. Very high, high, and medium risk levels are primarily distributed in the foothill transition zone, while low and very low risk levels are mainly found in the inland plain areas. This spatial differentiation remained generally stable throughout the study period. (2) The landscape risks associated with ancient sites in Yuncheng City exhibit significant positive spatial autocorrelation and clustering characteristics, and their clustering structure remained relatively stable throughout the study period. (3) The overall risk structure is dominated by low to medium risk levels, with risk-level transitions occurring primarily between adjacent levels in both directions. During the later stage of the study period, the risk structure showed a shift toward medium and high risk levels. The findings provide a scientific basis for risk management and the sustainable development of regional ancient site landscapes.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100451

Authors: Xiangqiang Min Junhao Li Zirou Wang Shengxin Guo Shihang Wang Lei Wang Yi Huang

To address the bottlenecks of weak cross-domain generalization, insufficient representation of complex spatial behaviors, and the scarcity of real anomaly labels in the detection of multi-semantic pedestrian trajectory anomalies (e.g., running, loitering, and wrong-way) in public security, this paper proposes an unsupervised detection framework combining kinematic constraints and a Bidirectional Long Short-Term Memory Autoencoder (Bi-LSTM-AE). While deep graph neural networks and attention-based models capture multi-agent interactions, they are frequently prone to spatial weight drift and severe computational overhead when transferred across disparate surveillance layouts. In contrast, our framework focuses on the intrinsic kinematic manifold of individual trajectories, striking an effective trade-off between cross-domain generalizability and linear computational complexity O(L). To this end, we construct a rigorous cross-domain evaluation paradigm, utilizing normal trajectories from a massive public dataset (SDD) to drive the unsupervised training of the model, aiming to fit a generalized human walking manifold unconstrained by specific physical scenes. To tackle the lack of multi-dimensional evaluation benchmarks, we design a multi-semantic anomaly generation strategy based on underlying kinematic constraints, programmatically synthesizing highly realistic challenging test samples on the ground-truth manifold of the target domain (UCY). Cross-domain experimental results demonstrate that the proposed framework can overcome the Gaussian smoothing trap of generative models and the cross-domain generalization degradation of complex graph networks. In highly challenging unseen scenes, the overall detection F1 score is maintained above 0.90; notably, in the detection of “spatial loop anomalies (loitering)” that relies on a global spatio-temporal receptive field, a precise recall rate of 93.76% is achieved. With linear computational complexity, our method provides a novel solution for highly robust intelligent video surveillance early-warning in the open world.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100450

Authors: Peng-Tao Guo Mao-Fen Li

Spatial prediction requires determining how sampled observations support estimates at unsampled locations. Existing methods commonly represent these relationships in geographic space, environmental feature space, or through a global statistical relationship between predictors and the target property. We propose a dual-space comprehensive similarity (CS) framework that combines target-oriented environmental similarity (ES) with geographic spatial proximity (SP). ES uses partial least squares screening and principal component transformation, whereas SP derives from scaled geographic distance. Their product gives high influence only to observations supported in both spaces. We embedded CS in individual predictive soil mapping (iPSM), creating iPSM-CS, and evaluated it using 547 soil organic carbon density observations from the Tibetan Plateau. Repeated experiments were conducted using training datasets of 100, 200, and 400 samples, with an independent 27-sample validation set and a 20-sample test set. The 500-sample scenario was evaluated separately using fixed neighbourhood sizes determined before final testing. Across the three repeated sample-size scenarios, iPSM-CS generally produced higher correlation coefficients (r) and lower root mean square error (RMSE) and mean absolute error (MAE) values than iPSM-ES and the benchmark models. In the 500-sample evaluation, iPSM-CS achieved r = 0.79, RMSE = 2.44 kg·m−2, and MAE = 1.93 kg·m−2, compared with r = 0.55, RMSE = 3.34 kg·m−2, and MAE = 2.52 kg·m−2 for iPSM-ES. Top-1 CS and Top-1 ES were significantly negatively associated with relative prediction error in the independent test set, while CS provided a clear stratification of errors across the predefined similarity intervals. The results support the use of dual-space similarity for spatial prediction, although its transferability requires evaluation across additional regions, attributes, and spatial validation designs.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100448

Authors: Eric Vaz

Markov chains remain the default engine for land-use allocation. This paper argues for integrating territorial memory into the process. That is, an explicit representation of the persistence of land-use states. We subject that proposal to an unfavourable test. Using the five-epoch Agriculture and Agri-Food Canada (AAFC) Land Use series (2000–2020), harmonised to an epoch-stable legend, we compare five nested models across 260 Ontario population centres under a strict temporal hold-out in which every model receives identical class totals, isolating allocation skill. Three results follow. First, conventional validation is uninformative: because these settlements change only ~0.2% of cells per five-year interval, a null model predicting no change attains 99.8% accuracy and outscores every fitted model, so reported accuracies above 0.95 measure landscape inertia. Second, a register built from previous states is analytically and empirically near-inert: the previous state equals the current state for 98.9% of cells, so it merely reinforces an already-dominant transition diagonal and changes essentially no predictions. Third, allocation skill comes from spatial neighbourhood structure (0.00 to 0.03); re-specifying memory over conversions rather than states adds a small forward-fold gain that does not survive fold reversal, so we present it as suggestive rather than established; a sensitivity analysis ties the reversal to over-strong calibration of the momentum weight rather than to the change-memory field itself. We also show that the AAFC v5 legend, when used raw, inflates the 2010–2015 change twenty-fold. We conclude that territorial memory should be re-specified as memory of change, that accuracy and kappa cannot validate models on persistent landscapes, and that, once a neighbourhood term is present, errors in the projected quantity of change cost more skill than further refinement of the allocation rule in urban and peri-urban environments.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100447

Authors: Hongyun Zhang Jin Liu Fang Wang Yahong Zhao Ye Xuan

Flood assessments require reproducible links between spatial products and the measurements ultimately reported to analysts, yet a large language model (LLM) agent may complete a geographic information system (GIS) workflow without generating the required product or traceable claim. We present GeoFloodAgent, an LLM-orchestrated flood-GIS workflow that combines typed planning and deterministic GIS tools with a separate evaluation of runtime disposition and post-run alignment with task-specific product, lineage, and action contracts. We evaluated 80 Chinese-language tasks from the Zhengzhou (2021) and Forlì (2023) flood cases under seven configurations and three repetitions, yielding a registry-remediated composite of 1680 assignments. In the Full configuration, 212 assignments (88.3%) were accepted and contract-aligned, whereas 16 accepted assignments did not align with their registered contracts; all 16 concerned clarification or refusal actions. Three runs were rejected despite retained contract-aligned evidence. The cluster-weighted Full-Tool Agent differences were −1.15 percentage points for accepted contract mismatch and +1.07 percentage points for contract-aligned acceptance, with empirical intervals spanning benefit and harm. Under frozen Zhengzhou inputs, deterministic outputs were sensitive to the minimum SAR object-size setting: relative to 80 pixels, candidate flood extent changed by +105.6% at 40 pixels and −30.9% at 120 pixels, whereas a 20 m simplification tolerance reproducibly failed during geometric union. These configuration-specific results do not validate hydrological accuracy or identify universal optimal defaults. These findings show that completion rate alone is inadequate for evaluating LLM-orchestrated flood GIS: runtime disposition and evidence alignment should be reported jointly. The study evaluates a finite, study-authored benchmark and does not establish hydrological accuracy, operational decision validity, or open-world error detection.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100446

Authors: Cephas Bisimwa Mwimangire Odera Zephania Eddy Opiyo Eliel Chirongozi Mutabesha Paul Rawiel Dietrich Schröder

Africa remains the most under-addressed continent on the planet, with an estimated 60–70% of streets and dwellings lacking formal, standardised addresses. This deficit has profound consequences for governance, emergency response, e-commerce, logistics, and the daily lives of over 1.4 billion people. We present a hybrid graph-theoretic and machine learning framework for automated address generation, trained on high-quality open data from three European cities (Stuttgart, Paris, and Bern) and progressively adapted to seven African cities (Kigali, Dakar, and Kampala as training cities; Dar es Salaam, Nairobi, Kinshasa, and Lagos as unseen test cities). The framework models street networks as primal and dual graphs, applies the Intersection Continuity Negotiation algorithm for stroke detection, and deploys four supervised classifiers for connectivity error detection, road type classification, building-to-street assignment, and address quality scoring. Building footprints are fused from three complementary sources (OpenStreetMap, Google Open Buildings, Microsoft Building Footprints) and augmented with SRTM elevation data and the World Settlement Footprint 3D (WSF 3D) dataset. Across ten study cities, the pipeline produces 3.2 million merged building footprints, 467,000 street edges, and 880,464 quality-scored addresses, and a five-version iterative improvement cycle reduces the EU–Africa address quality gap by 54.2%. To validate African address outputs independently, we construct a cross-validation database of 1000 verifiable addresses of known organisations across the seven African cities; cross-validation achieves a geometric match rate of 79.6% within 100 m and 83.5% within 150 m. The system is implemented as an open-source QGIS plugin and ArcGIS Toolbox.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100445

Authors: Lewen Zhan Zhenguo Wang Lifei Wang Shengchen Yin Zhen Xu

In high-density cities, limited land availability constrains the expansion of new public sports facilities, increasing the importance of activating existing spatial resources. School sports facilities, which are widely distributed across urban residential areas, may provide supplementary opportunities for community physical activity during non-teaching periods. Using Nanjing, China, as a case study, this study adopts a four-phase research design comprising spatial assessment, stakeholder salience analysis, empirical investigation, and governance strategy development. First, GIS-based analysis was used to examine the spatial distribution of non-school sports facilities and school campuses and to compare 15-min walking accessibility under a baseline scenario and a hypothetical school-sharing scenario. Modeled population coverage increased from 62.93% to 71.67%, while approximately 147.37 km2 and 605,700 residents were additionally covered, indicating substantial potential for spatial supplementation. Second, six stakeholder groups were assessed using the Mitchell Salience Model based on power, legitimacy, and urgency. Third, 32 semi-structured interviews and a resident questionnaire survey (N = 216) were used to examine stakeholder concerns, residents’ perceived needs and attitudes, and major implementation conflicts. The findings indicate that the principal barriers to joint use are concentrated around safety responsibility, cost sharing, and use priorities. Based on these findings, four governance strategies are proposed: spatio-temporal decoupling, risk hedging, embedded negotiation, and digital supervision. Overall, school sports facilities show considerable potential to supplement existing urban sports provision, but their effective joint use depends on appropriate arrangements for stakeholder coordination, risk management, and operational implementation.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100444

Authors: Lujin Hu Hao Liu Jianing Ma Xinyu Zhang

Commercial and residential spaces constitute two fundamental components of urban spatial structure, and their coupling relationship critically shapes urban functional organization and planning interventions. However, integrated analyses of static facility layouts and observed human mobility remain limited. This study develops a static–dynamic framework that integrates facility co-distribution with observed travel connectivity. It further distinguishes catering, shopping, living, financial and insurance, accommodation, and vehicle-related services for category-specific comparison. Kernel density estimation (KDE), average nearest neighbor (ANN) analysis, spatial autocorrelation, Spearman’s rank correlations, median-based classification, and social network analysis characterize static–dynamic relationships. Data comprise Amap commercial points of interest (POIs), Anjuke residential POIs, and a single weekday of Didi origin–destination (OD) records. After removing records with invalid positioning or missing fields, 10:00–16:00 trips within the study area were matched to POI buffers. Both commercial and residential POIs exhibit clustering, with commercial POIs more strongly clustered. Residential POIs follow a core–periphery gradient with a stronger presence in western Beijing. Residential intensity is positively associated with overall commercial intensity in neighboring subdistricts (bivariate Moran’s I = 0.645). Category-specific associations range from 0.282 for vehicle-related services to 0.642 for living services, remaining supported after false-discovery-rate correction (q ≤ 0.0002). Residential–commercial facility intensity and commercial facility–inbound mobility intensity show strong rank correspondence (Spearman’s ρ = 0.869 and 0.883, respectively). Median-based classification identifies 170 subdistricts with concordant high–high or low–low facility–mobility patterns and 20 with contrasting high–low or low–high patterns. Catering, shopping, and living services form more continuous networks, whereas financial and insurance, accommodation, and vehicle-related services exhibit sparser or more spatially peripheral network patterns. Distinguishing facility co-distribution from observed travel connectivity provides service-specific evidence for commercial facility allocation, community life-circle planning, and peripheral cluster development.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100443

Authors: Sungmo Ku Jinho Lee

Systematic scenario classification supports the characterization of test conditions in simulation-based autonomous driving validation. This paper proposes a multi-task framework that combines OpenSCENARIO text embeddings with Ego-vehicle motion representations learned by Autoencoders trained on the SinD real-world dataset. The framework jointly predicts trajectory type and binary perception and decision disturbances, defined using environmental conditions and collision occurrence, respectively. We evaluate 24 combinations of four text encoders and six trajectory Autoencoders under fixed and uncertainty-based task weighting. On generated test scenarios with modified temporal and environmental parameters, Qwen3-Embedding-0.6B combined with Mamba-AE achieves the highest Full Match Accuracy of 91.61%, requiring all three labels to be predicted correctly, and trajectory accuracy of 99.97% under fixed weighting. Perception classification limits joint performance, and high validation accuracy does not consistently extend to the modified test conditions. Uncertainty-based weighting provides no consistent improvement, while latent-vector ablations show task-dependent contributions from text and motion representations. These findings support offline classification within the evaluated scenario configurations while identifying limitations in robustness to parameter changes.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100442

Authors: Chenyang Li Wenjie Li Jianhua Gao Yongji Tang Jian Hu Maomao Zhang Chenming Zhang

Urban vitality varies systematically within the day, yet its time-specific associations with stable built-environment conditions remain insufficiently understood at fine spatial scales. This study examines two-hourly urban vitality in central Zhengzhou, China, using Baidu Heatmap-derived activity intensity and multi-source built-environment data at the parcel scale. Twelve two-hourly MGWR models form the primary analysis, identifying temporal changes in the direction, magnitude, and spatial distribution of local associations. A daily mean MGWR model with the same independent variables serves as an aggregate benchmark. Results show that: (1) vitality exhibited a clear daily pulse while retaining a persistent center–corridor–node structure, with Global Moran’s I ranging from 0.508 to 0.562; (2) the two-hourly MGWR models explained 69.1–81.8% of spatial variation, compared with 39.2–66.1% for OLS, while the daily mean MGWR achieved an R2 of 0.791; and (3) relative to the benchmark, the two-hourly coefficient distributions revealed variable-specific association windows, directional transitions, and changing local dispersion across transportation, land-use, and facility variables. These findings establish the added interpretive value of two-hourly analysis, support time-sensitive management, and provide diagnostic evidence on how observed daily activity patterns are distributed in relation to stable urban configurations.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100441

Authors: Chaorui Huo Yongji Wang Keke Li Xinhui Wang Donglei Si Qili Wang Zhihui Tian

Reconstructing long-term spatiotemporal processes from unstructured historical records presents a fundamental challenge in historical Geographic Information Systems (GIS). Local chronicles contain rich information regarding the timing, location, and context of historical events. However, their narrative form hinders systematic spatiotemporal interpretation and timeline mapping. To address this challenge, this paper proposes an integrated framework for timeline mapping using local chronicles, which includes (1) a dimension-importance evaluation method to identify the most relevant analytical dimension for visualization, (2) a large language model (LLM)-based workflow for extracting, matching, and clustering historical events from unstructured texts with respect to the identified analytical dimension, and (3) a multi-perspective timeline mapping approach to reconstruct and visualize spatiotemporal narratives. The framework is demonstrated through a case study in Dengfeng, China, where historical educational records from local chronicles are employed to reconstruct the long-term spatiotemporal evolution of educational equity. The case study demonstrates that (a) structural balance is identified as the most relevant analytical dimension, guiding subsequent event mining; (b) the LLM-based workflow extracts and structures 721 historical events, with 415 events matched to the identified dimension and clustered into three thematic groups of institutional plurality, school system reform, and spatial reorganization; and (c) the multi-perspective timeline maps generated from these clusters effectively reveal a long-term, coherent process of structural reconfiguration, illustrating the framework’s capacity to transform narrative historical records into interpretable spatiotemporal narratives. The proposed framework provides a transferable historical GIS workflow for timeline mapping using local chronicles and offers a practical approach for reconstructing spatiotemporal processes from unstructured historical texts in various geographical contexts.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100440

Authors: Aiman Batkalova Anton Ellenbroek Pengyu Hao Dimitar Taskov Duygu Uyar Zhongxin Chen Karl Morteo Kiran Viparthi Juan Alarcon Ramirez

Accurate mapping of aquaculture ponds is essential for monitoring production systems and supporting the sustainable management of coastal and inland environments. However, aquaculture pond detection from satellite imagery remains challenging because aquaculture ponds often share spectral and structural similarities with natural water bodies, wetlands, and seasonally flooded agricultural areas. This study introduces a method for identifying aquaculture ponds by characterizing management-driven hydrological cycles associated with aquaculture production using Sentinel-1 and Sentinel-2 time series data. Water indices and water surface extent dynamics were used to characterize key stages of aquaculture management, including pond drying and preparation, water filling, grow-out, and harvesting. A hybrid 1D-CNN–MLP model was then developed to classify ponds into aquaculture, non-aquaculture, and inactive pond categories. Independent validation showed that the proposed model achieved overall accuracies of 0.927, 0.936, and 0.954 for 2023, 2024, and 2025, respectively, with F1-scores for aquaculture ponds ranging from 0.842 to 0.980. Compared with existing land-cover products (including Dynamic World and ESRI Land Cover), the proposed method advances from water-surface detection to the identification of aquaculture use, demonstrating its potential to support operational aquaculture monitoring and sustainable aquaculture management.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100439

Authors: Meiyan Gao Zongmin Wang Fuxin Chai Kuang Li Xiaosong Shu Qiuhua Liang Junhua Li

Increasing urban extreme rainfall, driven by global climate change and rapid urbanization, highlights the deficiencies of existing studies on urban system responses to flooding. Current functional assessments of floods often neglect spatial variations and the temporal dynamics of urban functions. We first developed a GIS-based spatiotemporal assessment framework that integrates system functional curves and complex network analysis to evaluate urban functional conditions. Focusing on the 20 July 2021 Zhengzhou flood, the High-Performance Integrated Hydrodynamic Modeling System (HiPIMS) coupled with the Storm Water Management Model (SWMM) was used to simulate flood processes and spatiotemporal variations. The results are as follows: (1) Four rainfall events increased inundation areas, with the extreme event (Pr3) causing an 82.4% expansion relative to the initial water surface. (2) Flood stages (absorption, resistance, overload) were classified by maximum flood depth and inundation area, as well as the maximum growth rate of flood depth and inundation area. Grids with medium-level maximum inundation depth predominated, while those with the lowest-level maximum depth growth were least frequent. (3) Overall urban function declined sharply at 17:20, reaching a minimum of 18.2% at 21:00, resulting in a 32.8% total functional loss. (4) Complex network analysis reveals strong temporal co-movement among urban functions, with the temporal correlation network comprising 10 edges (density = 0.667, global efficiency = 0.865). Education and emergency shelter were consistently identified as the most central nodes across different correlation thresholds, indicating potential systemic vulnerability. The framework elucidates the spatiotemporal dynamics of urban function assessment during flooding and provides actionable insights for understanding urban system functioning mechanisms and emergency planning.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100438

Authors: Ioannis Pateras Polychronis Velentzas Michael Vassilakopoulos Antonio Corral

Algorithms for processing large-scale spatial datasets are of significant interest in both scientific research and industrial applications. The efficient implementation of such algorithms is crucial for modern data-intensive systems, and GPU-based parallel processing has emerged as a particularly effective approach for accelerating spatial queries. Among these, computing the k closest pairs (k-CP) between large spatial datasets is a fundamental problem with applications in spatial data analysis, geographic information systems, and data mining. This paper presents an exact GPU-based framework for processing k-CP queries in both two- and three-dimensional spaces, including variants capable of processing datasets that exceed the available GPU memory capacity. Starting from a baseline brute-force CUDA implementation, we develop memory-aware partitioning strategies and adapt DSPP-based spatial pruning to reduce unnecessary distance computations in the global k-CP problem. Furthermore, we incorporate several performance optimizations, including pinned memory and concurrent kernel execution, to overlap data transfers with computation and improve scalability. Experimental evaluation using both synthetic and real-world datasets demonstrates that the proposed methods substantially outperform baseline approaches. In particular, the DSPP + PEA variant employing a max-heap buffer generally achieves the best overall performance, especially for large datasets and higher values of k, highlighting its effectiveness and scalability for large-scale spatial query processing.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100437

Authors: Xuhai Fan Linglong Zhu

Traffic flow prediction is a core task in intelligent transportation systems, yet existing Transformer-based models suffer from high computational complexity, weak dynamic spatial modeling, and a lack of strict temporal causality constraints. We propose the dynamic sparse causal attention network (DSCFormer), which jointly models static and dynamic spatial information by combining Laplacian positional encoding with a sliding window-based dynamic adjacency matrix. A sparse causal attention mechanism reduces complexity through neighbor selection and enforces temporal causality via a lower-triangular causal mask, with a controlled accuracy–efficiency trade-off (12-step averaged mean absolute error (MAE) 13.18 at γ=0.20 versus 12.89 for dense attention on PeMS08). A dual-window hierarchical temporal self-attention separately captures short-term fluctuations and long-term periodicities, fused by a data-driven weight. Experiments on six public datasets, including PeMS highway and urban grid-based data, show that DSCFormer achieves 1.6%, 1.4%, and 1.9% lower MAE, root mean squared error (RMSE), and mean absolute percentage error (MAPE) than PDFormer (the strongest baseline), averaged over nine evaluation settings (three PeMS datasets and six grid inflow/outflow tasks), and 5.2%, 6.8%, and 4.7% lower than the mean of all 16 baselines under the same settings, where relative improvement is computed as Δ=(Scorebaseline−ScoreDSCFormer)/Scorebaseline×100%. DSCFormer also reduces inference time by 8.7%, parameter count by 6.7%, and total FLOPs by 8.3% relative to PDFormer on PeMS08.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100436

Authors: Jiqiu Deng Chaolin Zhang Hui Zhang Longbo Li Liji Sun Xiao Ma Zhiyong Guo

Natural-language interfaces make urban spatial databases accessible, but generated structured query language (SQL) can execute successfully while returning a semantically incorrect answer. We evaluate a post-generation correctness-ranking layer on 18,900 candidates from Nanjing, Wuhan, and Shenzhen. The original city split is a controlled, highly template-aligned benchmark rather than a structure-disjoint deployment test. Under strict source-side separation of fitting, isotonic calibration, threshold selection, and target evaluation, Spatial-safe improves executed non-empty area under the receiver operating characteristic curve (AUROC) in all nine transfer directions by 0.062 on average. Grouped-threshold area under the risk–coverage curve (AURC) is favorable in 8/9 directions, and calibration is mixed. Source-selected thresholds raise mean conditional target coverage within the executed non-empty operating subset from 0.457 to 0.669, while mean empirical risk rises from 0.110 to 0.136. Unseen-template and SQL-skeleton-disjoint effects are smaller and mixed. To determine whether the existing benchmark could support the adequately powered structure-disjoint comparison, we applied a predeclared power/data-sufficiency feasibility gate. With only 77 independent SQL-skeleton groups, 42/60 required cells fail the gate; therefore, the present benchmark cannot support an adequately powered structure-disjoint performance comparison without additional skeleton-diverse questions. We do not substitute an underpowered estimate for that missing evidence. The evidence therefore supports empirical correctness-ranking improvement in this controlled benchmark, not universal structure-disjoint generalization, formal SQL verification, benchmark-wide human validation of all correctness labels, or target-domain risk guarantees.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100435

Authors: Wantong Sun Xinyi Niu Mingqi Shu Yixin Huang

As regional integration and high-speed transport networks advance, intercity travel flows within mega-city regions (MCRs) provide a lens for understanding regional functional division of labor and spatial organization. Concentrated high-intensity flows between core cities and major peripheral nodes suggest that explanations should move beyond distance constraints and single-end city attributes to origin–destination functional linkages. Consumer and producer services help connect these functional linkages with actual travel flows, yet existing studies have focused mainly on service scale and paid insufficient attention to service level, which reflects functional non-ubiquity and diversity. This study incorporates service type, scale, and level, and origin–destination relationships into a unified framework to identify key factors associated with intercity travel flows within MCRs. Focusing on the Yangtze River Delta, this study examines subdistrict/town-level flows from 15 peripheral cities to Shanghai’s central urban area. Mobile phone signaling records measure flow intensity. Counts and complexity are used to measure service scale and level. Two XGBoost models are estimated: an origin–destination service model and a difference model. SHAP identifies contributions and directions, while two-dimensional ALE examines interactions between services at origins and destinations. Results show that both service scale and level are associated with intercity travel flows within MCRs, but their patterns differ by service type. Significant interaction effects exist between services at origins and destinations. These interactions mainly combine origin-scale foundations with destination-level-based attraction, with destination service level often showing threshold-like characteristics. This study extends the explanation of intercity travel flows to origin–destination functional linkages and service level.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15100434

Authors: Kamer Özgün Emre Demir

Urban rail investments can improve accessibility and support compact development, but their effectiveness depends on spatial and infrastructural conditions that are observable before operation. This study proposes an ex ante diagnostic framework for planned corridors based on three dimensions: coverage, directness, and connectivity. Coverage captures urban opportunity within graduated pedestrian catchments of 600–800 m, scaled to the urban context of each station group, through built-up intensity, activity concentration, and land-use diversity. Directness compares route and straight-line distances at corridor, segment, and individual link scales. Connectivity measures station integration with existing bus and rail services and road-based feeders within 300 m using a Network Connectivity Index (NCI). Applied to Antalya’s planned 4th Stage Rail System Line in Türkiye (20 stations; five segments), the framework reveals substantial spatial variation. Built-up ratios range from 0.10% to 29.47%, with no station leading all coverage indicators. The end-to-end directness ratio is 0.898, while the five contiguous segments range from 0.874 to 0.996; indirectness is not distributed evenly but concentrates at three turning points that, together, account for 47% of the corridor’s total angular deviation. NCI values range from 0.000 to 0.708 (mean 0.363), and rankings remain stable across nine weighting scenarios (Spearman’s ρ = 0.994–1.000). Rather than aggregating dimensions, the framework compares them to identify mismatches and guide targeted pedestrian, feeder-integration, and station-area interventions.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15090433

Authors: Meiyan Gao Zongmin Wang

Refined urban flood risk assessment requires a paradigm beyond traditional single-scale analyses. Existing approaches often overlook fine-scale functional differentiation and cascading risk connectivity across urban systems. This study proposes an integrated framework combining semantic–spatial representation of Point-of-Interest (POI) data, hierarchical feature integration, and a hydrological–functional–road tripartite potential flood risk connectivity network. A semantic–spatial dual-branch autoencoder extracts POI features for functional zone identification, while hazard, exposure, and coping capacity indicators are integrated to quantify block-scale flood risk. Hydrological connectivity, functional coupling, and road cascading failure models are further combined to characterize multi-path flood risk connectivity patterns. Results show that (1) functional zones were derived from dominant POI categories. The semantic–spatial clustering achieved a silhouette coefficient of 0.564 with 0.2% noise blocks. The spatial-only configuration achieved 0.554. This suggests that semantic information provided complementary but limited refinement. (2) A total of 302 high-risk blocks (20.5%) were identified, while the median-based classification showed that public service zones had the highest proportion of the high-risk group (61.3%). (3) The coupled flood risk connectivity network covers 99.2% of blocks with significant spatial dependence (Moran’s I = 0.703). Connectivity intensity varied among functional zones, but most pairwise differences were not significant after Bonferroni correction. Network centrality showed significant associations with connectivity intensity and range. (4) The mean standardized functional mixture degree was 0.629. Geographically weighted regression (GWR) revealed spatial variation in the local association between functional mixture and physical flood susceptibility, but the Monte Carlo test did not provide evidence of statistically significant global spatial non-stationarity (p = 0.960). This framework provides a function-oriented perspective for understanding urban flood resilience by integrating functional structure, multidimensional flood risk, and multi-path risk connectivity, and supports the characterization of potential flood-sensitive areas through function-based flood risk assessment.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15090432

Authors: Jiayue Zuo Yuheng Tu Linggui Liu Simin Tao

Against the background of urban renewal, urban governance has increasingly pursued the coordinated development of spatial provision and public needs. Taking the central urban area of Shanghai as a case study, this study integrates street-view imagery and geotagged Weibo data to examine the spatial matching relationship between Street Space Quality (SSQ) and public sentiment. The results show that although SSQ and public sentiment both exhibit significant spatial clustering, no stable synchronization is observed between their spatial patterns. Most mismatches remain localized, whereas some display clear neighborhood continuity; moreover, the stable matching types exhibit distinct SSQ profiles. Overall, spatial mismatch in central Shanghai is characterized by the coexistence of locality and neighborhood continuity and by pronounced differences in quality profiles across types. These findings provide a spatial analytical basis for coordinated diagnosis and fine-grained governance in urban renewal.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15090431

Authors: Tri Wibisono Trias Aditya Catur Aries Rokhmana

Large-scale monitoring of land rights remains constrained by substantial time requirements, limited spatial coverage, and considerable human resource demands associated with conventional image interpretation and field verification. This study develops an operational Geospatial Artificial Intelligence (GeoAI) workflow for indicative monitoring of land-use compliance by integrating semantic segmentation of land cover within registered land rights (HAT) boundaries with spatial planning information. U-Net and SF-Net architectures were evaluated for temporal monitoring using two models: M3 based on 10 m Sentinel-2 imagery from 2017–2023 and M4 based on 0.07 m orthophotos from 2022–2023. Data augmentation increased the number of M3 samples from 445 to 3560 and M4 samples from 1106 to 8848. The U-Net M3 model achieved an F1-score of 0.961 and an IoU of 0.927, whereas the SF-Net M4 model achieved an F1-score of 0.969 and an IoU of 0.940. At Keera Plantation, the 28–29% difference in estimated oil palm area between M3 and field verification reflects inter-method variation in a heterogeneous landscape rather than model error alone. M3 processed the entire 680,380.99 ha administrative area of East Luwu Regency in only 4 min 31 s to assess operational scalability and monitoring efficiency. The proposed GeoAI workflow is effective for large-scale initial screening and prioritization. However, GeoAI outputs are intended as decision-support evidence and do not replace field verification or legally accountable administrative decision-making.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15090429

Authors: Wen Cao Gang Chen Siqi Zhao Tianchi Yang

Aggregated epidemics, characterized by rapid transmission over short periods, pose severe threats to public health security, necessitating the development of precise source tracing and simulation methods to support efficient prevention and control. However, existing studies suffer from three major gaps: (1) predominant focus on national/regional scales with limited urban-scale analysis; (2) reliance on proprietary mobile data that are often inaccessible; and (3) NP-hard computational complexity in traditional source tracing methods. To bridge these gaps, this paper proposes an integrated spatiotemporal diffusion model comprising two core components: outbreak point estimation and spatial diffusion simulation. The model first uses the SEAIR infectious disease dynamics model to predict trends, combines the 3-Sigma criterion and viral incubation period to screen early epidemiological survey data, employs a road network-constrained DBSCAN algorithm for spatial clustering, and locates the outbreak point via an improved inverse distance weighting method incorporating time and POI density weights. Subsequently, using the estimated outbreak point as the initial transmission center, and based on the “cell-type” living structure hypothesis of populations, it fuses multi-source geographic data to quantify regional attractiveness and simulate viral diffusion in grid space. The model is validated using COVID-19 epidemic data from Xi’an, Shanghai, and the Hong Kong Special Administrative Region of China. Results show that the proposed outbreak point estimation method effectively estimates the initial transmission center, with distances between estimated points and officially announced points of 0.786 km, 1.676 km, and 5.441 km, respectively—shortened by 179 m, 1091 m, and 711 m compared to related studies. The spatiotemporal diffusion model effectively simulates daily incidence patterns, achieving average coverage rates of 67.32% and 72.9% and average precision rates of 60.67% and 78.84% in Shanghai and Hong Kong, respectively, with significantly better simulation accuracy than traditional methods in the early and middle stages of the epidemic. This study provides a scientifically robust, data-parsimonious framework for precise source tracing, early warning, and resource allocation in urban epidemics, with strong generalizability to other resource-limited settings.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15090430

Authors: Yong Zhang Yaqin Sun Xun Zhou Yifan Zhang Hui Yang Kefei Zhang

Urban pluvial flooding is a rapidly evolving disaster that requires timely and fine-grained water-depth information for emergency response and impact assessment. However, street-scale water-depth observations are often difficult to obtain during short-duration flood events due to limited coverage and insufficient spatial detail of conventional monitoring systems. To alleviate this problem, this study proposes an image-led framework for urban street-scale flood-depth estimation from crowdsourced image–text data. The proposed method integrates reference-object-based visual evidence construction, textual depth extraction, and conditional image–text fusion to estimate water depth from crowdsourced image–text records. Specifically, image-based depth evidence is constructed by identifying reference objects, determining submerged parts, and establishing object-depth relationships through knowledge mapping. Textual information is further analysed through rule-based extraction and semantic verification to identify quantitative depth cues, including explicit values, ranges, and relative water-level descriptions. Considering the potential inconsistency between image and text observations, a conditional fusion strategy is designed to incorporate textual information only when quantifiability, directional, and image–text discrepancy conditions are satisfied. Experiments demonstrate that the proposed method can effectively extract quantitative flood-depth information from crowdsourced multimodal data and improve the utilization of heterogeneous disaster observations. The resulting depth estimates are intended as auxiliary evidence for post-disaster situational assessment rather than as substitutes for in situ water-level measurements.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15090428

Authors: Rui Si Yaoyu Lin

Functional complementarity between residential and commercial functions explains inter-zonal shopping flows, yet commerce is typically treated as a single category and the intervening-opportunity pillar of spatial interaction theory is left empirically dormant. We develop a spatial interaction model that partitions commercial supply into non-daily and daily tiers, recasts the competing-destinations (CD) operator as a directional, kernel-weighted detour-cost term, and embeds origin-community socioeconomic status (SES) as a multiplicative moderator of complementarity, competing destinations, and distance impedance. Estimated by Poisson pseudo-maximum likelihood on 1668 Chengdu communities and shopping ride-hailing trips, the model yields three findings. First, SES moderates Spatial Function Complementarity in opposite directions across tiers: high-SES communities respond more strongly to non-daily complementarity, whereas the negative daily-tier effect is weaker among low-SES communities. Second, competing destinations act as an agglomeration signal rather than a flow-diverter, with high-SES communities responding more strongly. Third, network distance was significantly and negatively associated with shopping ride-hailing flows, whereas its interaction with origin-community SES was not statistically significant. These findings show that consumption frequency and directional commercial structure help explain socioeconomic heterogeneity in shopping ride-hailing flows.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15090427

Authors: Jonghwa Lee Baekjun Kim Kweonhyoung Lee

This study examined public library use intensity across seven metropolitan cities in South Korea, considering spatial dependence and the surrounding urban environment. From 485 public libraries, model-specific analytical samples were constructed for visits per m2, loans per m2, and borrowers per m2. Urban facility variables within 1- and 2-km catchments were used as predictors, and academy density was corrected using education office records geocoded using the Kakao Local API. Moran’s I and Anselin–Florax Lagrange Multiplier diagnostics were used to assess spatial dependence among libraries and compare ordinary least squares, spatial autoregressive (SAR), and spatial error model results. Moran’s I indicated significant spatial clustering across all three outcome variables. SAR results revealed that the indirect component accounted for 46.3% and 40.8% of the total academy effect on visits and loans per m2, respectively, representing model-based indirect effects rather than direct evidence of user movement. A supplementary negative binomial model using library floor area as an exposure did not corroborate the academy-density association, indicating that this finding is specific to the area-normalized model rather than scale-invariant. These findings highlight the importance of considering spatial dependence and urban facility contexts when interpreting public library use intensity at the regional scale.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15090426

Authors: Madhu Sudan Adhikari Subash Ghimire Dev Raj Paudyal

Built-up expansion is reshaping landscapes across Nepal; however, consistent multi-temporal mapping remains challenging due to rugged terrain, fragmented settlements, and heterogeneous land-cover conditions. This study develops and evaluates a multi-product and terrain-informed workflow in Google Earth Engine for mapping built-up expansion across three physiographically contrasting districts of Nepal: Arghakhanchi, Lalitpur, and Chitwan, from 2017 to 2025. Annual predictor stacks were generated by integrating Sentinel-2 spectral bands and derived indices, Dynamic World built-up probabilities, and SRTM-derived elevation and slope variables. ESRI Global Land Cover datasets were used separately for auxiliary cross-product comparison and assessment of the mapped outputs. Preliminary yearly built-up masks were generated using district- and year-specific Random Forest classifications, followed by the post-classification constraints, and were subsequently integrated through cumulative expansion mapping. Accuracy assessment for 2017, 2021, and 2025 yielded overall accuracy values of 86.4–92.4%, built-up F1-scores of 84.7–91.3%, and Kappa coefficients of 0.81–0.91. Between 2017 and 2025, cumulative built-up extent expanded by 8054.65 ha in Chitwan, 2406.20 ha in Arghakhanchi, and 2215.96 ha in Lalitpur; Arghakhanchi recorded the highest proportional increase (117.6%). The mapped expansion was comparatively dispersed in Arghakhanchi, concentrated within metropolitan and peri-urban areas in Lalitpur, and broader and corridor-oriented in Chitwan. Because previously detected built-up pixels were retained in subsequent cumulative outputs, the resulting extents were non-decreasing by construction and did not represent demolition or other land use reversals. Consequently, annual built-up expansion should not be interpreted as net annual land-cover change. The proposed framework provides a practical and transferable approach for comparative built-up expansion monitoring and urban growth assessment across contrasting physiographic settings.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15090425

Authors: Mervana Mograby Amal Iaaly

Seismic geospatial decision models can produce misleading outputs without rigorous input validation. Fusion conceals the defect: weight-sensitivity testing on an inadmissible layer reports stability rather than validity. This paper presents Evidence-Gated Screening (EGS), a four-gate framework. It tests computational provenance, applicability, decision-relevant signal, and input support before fusion. EGS audited six candidates for 1552 building footprints in Aabadiyeh, Lebanon: a construction-epoch proxy, two soil products, an exploratory Age-Soil Index, a Newmark displacement raster, and an Earthquake Risk Index. Only the proxy was conditionally admitted. Two interpreters validated it on a stratified probability sample of 216 buildings, giving binary accuracies of 0.833 (0.782–0.880) and 0.861 (0.815–0.907), with κ = 0.832 on the image evidence. The remaining candidates were rejected, including the risk index. Of 87,492 evaluable cells, 56.30% fell outside the stated critical-acceleration domain; raster-to-building transfer omitted 270 footprints. Gating returned a queue of 400 buildings and not the composite’s 303. Four products were assessed by two independent assessors, who agreed on 13 of 16 gates and on all four dispositions: two were conditionally admitted, two rejected. This indicates that EGS can screen out inadmissible inputs before fusion and support building inspection prioritization, but not vulnerability prediction without building-level validation.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15090424

Authors: Xiang Chen Liu Liu Yao Shen

Efficient and accurate flood risk assessment is essential for cultural heritage conservation in the context of increasingly frequent extreme weather events. Conventional flood risk assessment methods primarily aim to approximate inundation patterns across the entire study area, and their analytical units are not necessarily aligned with object-oriented decision-making. To address this limitation, this study develops a site-oriented surrogate modeling approach. The approach integrates site-specific characteristics with multi-scale neighborhood features to characterize local environmental constraints and investigates how neighborhood scale influences predictive performance. To this end, five models are constructed and compared, i.e., M1_Site, M2_50m, M3_200m, M4_500m, and M5_Multi-Scale. Experiments were conducted on 714 cultural heritage sites in Shanghai. Overall, the M2_50m, M3_200m, and M5_Multi-Scale models demonstrate strong predictive performance, with Test F1, GroupKFold F1, and Test Accuracy all above 0.85. According to the results, the site-oriented surrogate modeling approach is feasible, and neighborhood features are important. Model performances also exhibited clear scale sensitivity. This approach provides a practical framework for rapid, site-specific flood risk assessment and offers new insights for decision-making in cultural heritage conservation. Future research could focus on validating this approach in other contexts and applying it to multi-scenario risk analyses to further support heritage conservation planning.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15090423

Authors: Haowen Xu Sisi Zlatanova Ruiyu Liang Ismet Canbulat

The rapid growth of high-resolution 3D voxel datasets derived from LiDAR, BIM, and urban digital twin platforms has created new opportunities for volumetric environmental simulation. However, existing fire-spread models are often surface-based or computationally intensive for large-scale 3D applications, motivating the investigation of efficient voxel-native alternatives. This study presents a pilot investigation of a physics-based, GPU-accelerated framework for rapid 3D fire-spread simulation in wildland–urban interface (WUI) environments. Fire propagation is represented through simplified formulations of conduction, radiation, and wind-driven convection on a structured voxel grid, with combustion behavior parameterized using fuel and material properties. The framework is not intended to replace high-fidelity computational fluid dynamics (CFD) models, but rather to provide a computationally efficient approach for rapid evaluation of fire-spread scenarios in large 3D urban environments. A voxel-native parallel memory layout and stencil-based computational scheme enable efficient neighbor access and GPU-parallel updates. The framework is demonstrated using a voxelized model of Liverpool, NSW, Australia, and its computational performance is evaluated on both local GPU and high-performance computing (HPC) platforms. The results demonstrate predictable runtime scaling and practical performance for domains exceeding one million active burnable voxels. An initial cross-model comparison with the FDS CSIRO scenario further demonstrates substantial spatial agreement while identifying remaining differences in burned area. The results demonstrate the feasibility of the framework for rapid urban-scale 3D fire-spread evaluation, with potential future applications in emergency response and time-critical decision support following further calibration and validation.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15090422

Authors: Hanyu Xu Chaoyong Zhang Yuhang Xiong Nan Jiang Huige Xing

With climate change and urbanization accelerating, cities face increasing disaster risks. As critical facilities for safeguarding residents’ safety, emergency shelters have attracted growing attention due to spatial supply–demand mismatches. From the perspective of the 15-min community life circle, this study aims to develop a fine-scale analytical framework for assessing emergency shelter accessibility and inequality, using Pidu District, Chengdu, China, as a case study. The three-step floating catchment area (3SFCA) method is applied to assess emergency shelter accessibility, while the Gini coefficient and Kolm–Pollak equally distributed equivalent (EDE) are introduced to evaluate accessibility inequality across subdistricts and communities. The results show that: (1) compared with the traditional Gaussian two-step floating catchment area (G2SFCA) method, 3SFCA captures clearer accessibility differentiation and stronger local sensitivity; (2) emergency shelter accessibility shows marked spatial differentiation, with pronounced local clustering in the walking mode and a polycentric pattern in the cycling mode; (3) the Gini coefficient reveals distributional imbalance in emergency shelter accessibility, showing a clear “center–periphery” pattern in which urban areas generally exhibit lower inequality than suburban and rural areas; and (4) the Kolm–Pollak EDE emphasizes lower-tail welfare and helps distinguish “low but even” areas from “high but polarized” areas. These findings support equity-oriented strategies for emergency shelter planning and resource allocation. The study provides methodological and practical implications for fine-scale emergency shelter planning and resilient urban governance, contributing to more equitable, inclusive, resilient, and sustainable urban environments.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15090421

Authors: Can Yin Lifu Jin Zhe Zhang

Understanding spatial variation in electric vehicle energy consumption (EC) can support more targeted urban transport analysis, but observational data do not identify the causal effects of urban design. This study analyzes approximately 1700 private EVs operating in Shanghai from October to December 2020, integrating about 29 million high-resolution GPS records with multi-source built-environment data. Segment-level EC was aggregated to 500 m grids and communities. Community functional zones were derived by encoding POI sequences with Qwen2.5-1.5B-Instruct and clustering the resulting embeddings, while Random Forest models and SHAP were used to characterize model-attributed associations with spatial EC variation. At the grid level, mean speed and speed variability received approximately 84% of the native Random Forest importance, whereas the 5D-inspired indicators received the remaining 16%, led by destination accessibility, road density, and bus-stop density. At the community level, the Urban core zone had approximately 9% higher fleet-weighted EC than the Rural zone, while within-zone importance rankings differed between dense and sparse zones. These results describe conditional, location-level associations for the sampled fleet and season.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15090420

Authors: Qian Li Ying Sun Tianqi Guo Xiangfeng Li

Within the “15-min community life circle”, spatial accessibility of basic services is critical to ensuring social equity for people with disabilities. However, existing wheelchair accessibility assessments predominantly focus on static “origin-destination” node configurations, treating the road network as homogeneous and failing to capture how micro-level physical barriers induce macro-level accessibility decay through network connectivity transmission. This study addresses this gap by introducing an “ideal-realistic” scenario comparison framework combined with weighted edge betweenness centrality. This approach backward-traces resource accessibility loss to two dimensions of micro-level road-segment failure, namely topological function loss and service resource loss, shifting the analytical paradigm from resource configuration assessment to network connectivity diagnosis. Taking the central Xinhai Subdistrict of Lianyungang, China, as a case study, this paper constructs a dual-scenario network based on field-surveyed barrier data and establishes a two-dimensional measurement system centered on segment-level function loss and resource loss. The findings reveal that realistic physical barriers cause significant deprivation in wheelchair users’ access to life-circle services. Moreover, this decay is not homogeneously distributed but is highly contingent on the topological role of road segments, exhibiting observed spatial patterns such as skeletal network collapse, critical entrance blockages, and isolated terminal resource clusters. The analysis elucidates how micro-level physical barriers are propagated through spatial connectivity, offering a new perspective for understanding impedance dynamics in existing urban slow-traffic environments. The results can inform precision interventions in pre-standard neighborhood renewal: critical bottleneck segments provide spatial targets for prioritizing resource allocation and context-specific strategies across communities.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15090419

Authors: Shuqi Cao Guohua Ji

Effective shape representation of building footprints is essential for many geospatial analysis and applications, such as building retrieval, cartographic generalization, and urban morphology analysis. Constrained by scarce and coarse annotations of building shapes, self-supervised learning (SSL) is regarded as a promising paradigm. However, existing SSL approaches primarily infer invariance from augmented views of the same instance without explicitly defined similarity, limiting their ability to capture multi-layered relationships—from geometric regularity to structural layout—across building shapes. Providing reliable and interpretable similarity supervision remains challenging. In this work, we propose Deformation-Induced Self-Supervised Learning (DI-SSL), a framework that explicitly defines similarity through Geometric-Aware Deformations (GAD). GAD constrains deformations along global topology, geometric variation and structural layout, jointly characterizing structural comparability between shapes and ensuring geometrically valid, structurally coherent variants under controlled form deviations. Consequently, similarity is explicitly defined through these constrained deformation processes, yielding an embedding space where structurally similar shapes are consistently organized. Experiments demonstrate that DI-SSL achieves strong performance in retrieval and few-shot generalization. The learned embedding space is well-structured, exhibiting global separability of structural prototypes and local continuity consistent with morphologically meaningful similarities. Ablation studies further highlight the critical role of deformation-induced supervision in shaping this space.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15090418

Authors: Chamseddine Zaki Houssein Taleb Mostafa Rizk Hussein Sheaib Chadia Sawaya Bilel Neji Abbass Nasser

Geographic information systems have traditionally focused on the representation of visual and geometric aspects of geographic phenomena, while auditory information has largely been ignored or reduced to simple numerical attributes. Although numerous spatio-temporal data models have been proposed to describe dynamic geographic processes, few of them provide explicit semantic constructs for representing sound and its evolution over space and time. This paper extends the Modeling of Application Data with Spatio-Temporal Features (MADS) conceptual model by introducing an explicitly defined acoustic semantic dimension. The extension separates source-intrinsic acoustic emission (soundfeature), source–receiver propagation and exposure (AFFECTS and associated EXPOSUREZONE instances), and receiver-centered contextual perception (PERCEPTIONASSESSMENT). These complementary constructs enable acoustic emission, propagation, temporal dynamics, spatial exposure, and perceptual assessment to be represented within a unified spatio-temporal semantic framework. The applicability of the proposed approach is illustrated through a proof-of-concept scenario and representative database queries, demonstrating how the proposed acoustic constructs can be mapped, stored, and queried using standard GIS database infrastructure.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15090417

Authors: Boting Xi Shaohua Wang Siyu Zhao Wenda Wang Liang Long Chang Liu Jingyi Zhou Haojian Liang

Veterinary hospitals are increasingly important neighborhood services, but citywide evidence on their strategic allocation remains limited. Existing GIS-based suitability studies identify favorable areas, but do not determine which limited set of planning nodes to prioritise when existing supply is clustered and underserved areas require explicit protection. This study develops a transparent allocation framework for Beijing by integrating Random Forest (RF)-based observed clinic-location probability, E2SFCA-derived service-gap indicators, and a priority-constrained covering optimisation model. The RF output is interpreted as a proxy for location attractiveness, rather than a direct estimate of demand for veterinary healthcare. The optimisation then protects predefined priority areas and maximises weighted coverage under alternative facility budget scenarios. Additional analyses include a threshold-derived priority set, class-imbalance metrics, catchment-radius sensitivity, key-area-count sensitivity, systematic candidate/key-score coefficient perturbation, district-grouped spatial validation, raw versus calibrated RF allocation sensitivity, and pet-service endogeneity checks. Under the main 5 km catchment and P = 40 scenario, the model achieved complete priority-area coverage and weighted coverage of 0.656. Coverage increased further at higher budgets, so P = 40 is not regarded as a practical optimum, but rather as a diminishing-return scenario. At P = 20 and P = 40, priority-area coverage increased by 5.0 and 2.5 percentage points relative to MCLP, respectively, while aggregate weighted coverage remained comparable. The framework supports strategic planning priority analysis, not direct commercial site selection.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15090416

Authors: Yan Du Anqi Zhang Mowen Xie Yujing Jiang Hongda Zhang Jingnan Liu

Conventional early-warning methods for geohazards along railways rely largely on single observations, empirical criteria or static analysis, and struggle to meet the demands of corridor-scale screening and dynamic tracking. To address this gap, a railway-oriented Perception–Assimilation–Decision (PAD) closed-loop early-warning framework is proposed. The evolutionary patterns of typical geohazards along railways, including landslides, rockfalls, debris flows and settlement, are reviewed together with the application scope and limitations of multi-source monitoring techniques. Additionally, differentiated assimilation strategies are clarified, with continuous deformation and hydro-mechanical state updating for plastic failure and damage-sensitive evidence and critical-state identification for brittle failure. On this basis, a mechanism–data dual-driven assimilation paradigm and a two-scale PAD organisation, comprising corridor-scale spatial screening and site-scale state updating, are introduced. Recent applications show that data assimilation has shifted from correcting a single monitoring variable toward the dynamic coupling of multi-source observations with physical models. By establishing the logical chain from multi-source perception to state updating and finally to railway engineering response, the PAD framework transforms the traditional anomaly-identification-based warning mode into closed-loop risk management. The results provide a reference for building geohazard early-warning systems and engineering-oriented response along railways.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15090415

Authors: Chima Okoli Timothy Mulrooney Tony Bawo Esimaje Jasmine Allen

University geospatial hubs support research dissemination, teaching, public data access, and community engagement, but their value depends on whether resources can be found, accessed, interpreted, integrated, and reused. This study evaluates 747 publicly listed items in North Carolina Central University’s CoDE Open Data Hub using FAIR principles as an interpretive framework and the Federal Geographic Data Committee Content Standard for Digital Geospatial Metadata (FGDC CSDGM) as the domain-specific measurement basis. Nineteen criteria were scored on explicit criterion-specific 0–2 rules using live ArcGIS item properties, formal metadata XML, and available layer properties. Discoverability was evaluated for a proportionally stratified sample of 153 items through exact-title, keyword, location, item-type, category-browsing, and click-depth tests. The equal-criterion benchmark averaged 28.57 of 38 (75.19%); equal FAIR-dimension weighting produced a mean of 73.26%, and 95.85% of items retained the same descriptive performance band. Independent rescoring of 50 stratified items produced 76.74% exact criterion-level agreement, a linear weighted Cohen’s kappa of 0.587, and an absolute-agreement ICC of 0.715 for total scores. FAIR-aligned scores were highest for Findability (87.25%) and Accessibility (85.96%), followed by Reusability (67.26%) and Interoperability (52.59%). Exact-title and item-type searches retrieved all sampled items, while keyword search was successful for 88.24%. The results distinguish public availability from technical reuse readiness and identify spatial reference, attribute definitions, lineage, limitations, category structure, and navigation depth as priorities. The performance bands are study-specific descriptive summaries rather than FGDC compliance levels or FAIR certification. The study provides an evidence-preserving, type-aware method for applying FAIR principles alongside geospatial metadata standards in heterogeneous ArcGIS Hub catalogs.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15090414

Authors: Ahram Song Jinha Jung Anjin Chang Seula Park

Recent learning-based damage assessment methods using remote sensing data have greatly contributed to the rapid identification and classification of damaged buildings from pre- and post-disaster imagery. However, most outputs are provided in the form of final damage labels or scores. To address this limitation, this study proposes an evidence-aware data model for explainable post-disaster building damage assessment. The proposed model is derived from a component-level roof-damage assessment workflow using optical imagery and post-disaster surface-height data, and is designed to describe the observational evidence, intermediate indices, and decision rules leading to a specific damage grade, as well as its interpretation as a post-disaster change event. Based on the proposed logical data model, a hybrid spatial–graph database was implemented to support both spatial retrieval and the semantic explanation of damage assessment results. The proposed model was evaluated through six competency questions derived from different damage assessment requirements. The results confirmed that the proposed model can support component-level retrieval of damage results, tracing of damage evidence and assessment rules, spatial visualization of damaged building components, and integrated spatial–semantic explanation for individual objects. The results provide a foundation for managing disaster damage assessment processes and results as explainable and reusable structured data.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15090413

Authors: Aike Kan Nengsheng Li Xiao Yang Yichen Li Hongzhou Deng

Improving the precise perception of disaster risk and evaluating the effectiveness of warning systems are essential for natural-hazard prevention, mitigation, and spatial risk management. Using Sichuan Province, China, as a case study, this study treats government-issued disaster warning texts within 72 h before accident occurrence as formal risk signals and spatiotemporal information sources. Drawing on the protective action decision model and risk communication theory, we propose an integrated framework combining BERT-CRF-based information extraction, spatiotemporal analysis, and a case-crossover design to evaluate a multi-level warning system through the chain of risk identification, risk communication, protective action, and risk outcome. The results show a clear gradient across warning levels. High-level warnings are significantly associated with lower accident risk, whereas lower-level warnings can identify risky contexts but have limited intervention effects. These findings indicate that disaster warning systems have dual functions of risk identification and intervention, while medium-level warnings may reveal a weak link in the transition from risk perception to protective action. The study provides a geospatial analytical framework for linking official warning information with tourism safety outcomes and offers empirical support for optimizing warning systems and strengthening hazard risk governance in high-risk regions.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15090412

Authors: Hafiz Abdul Wajid Mehtab Khan Asim Farooq Muhammad Abid Danish Farooq Huzaifa Qadeer

Electric vehicles (EVs) are attracting choice and adoption as a travel mode in global transportation systems, driven by technological innovation, economic considerations, and advancing sustainable urban development. The planning, development, design, and location of EV charging stations are challenging tasks for underdeveloped countries such as Pakistan. To propose the site location for an EV charging station, this study comprises Multi-Criteria Decision Analysis and Geographic Information Systems. The study adopts a mixed-methods design, combining qualitative expert input with quantitative spatial analysis to investigate technical, social, and infrastructural criteria. The study considers indicators, including site suitability, population density, interaction between land use and transport, road network, transportation interactions, existing electricity grid infrastructure, and existing fueling and charging stations. This research presents pairwise expert comparisons, indicating that grid capacity (0.28) and demand (0.30) are important factors to consider. A composite suitability Index (CSI) through GIS-weighted overlay was used to classify the area into low, medium, and high suitability zones. According to the CSI, 26% of the area falls in highly suitable areas for EV infrastructure in Islamabad, and 41% falls in medium-to-high suitable areas. The remaining 33% area lies in low-suitability peripheral zones near the hilly region of Islamabad city, which is not considered suitable for EV infrastructure. A total of 67% of the city’s area is suitable for EV infrastructure.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15090411

Authors: Yinlan Huang Jingqiao Fang Shi Chen Tianshuo Xie

Under increasing cropland constraints and pressure to secure oilseed supplies, using winter–fallow cropland for winter rapeseed production can improve annual cropland-use efficiency. Taking the Wanjiang Plain, China, as the study area, this study integrated 10 m winter–fallow cropland maps (2019–2024), 30 m winter rapeseed maps (2000–2022), annual Satellite Embedding features, cropland data, and administrative boundaries. Multi-year occurrence frequency and Getis–Ord Gi* statistics characterized temporal persistence and spatial clustering. A MaxEnt model calibrated with 394 occurrence records from long-term high-frequency rapeseed areas and 26 screened embedding dimensions delineated cropland with present-day land-surface characteristics similar to historically persistent rapeseed locations. This layer was progressively intersected with the historical winter–fallow union and stable winter–fallow cropland. Stable winter–fallow cropland covered approximately 4437 km2, whereas long-term high-frequency winter rapeseed covered only 568 km2, revealing a marked spatial mismatch. Under five-fold spatial cross-validation, the selected linear-feature model with a regularization multiplier of 4 achieved a mean test AUC of 0.904 ± 0.015 and a 10% training-omission rate of 0.108 ± 0.022. Using the model-specific threshold of 0.3096, the final estimates were 6017 km2 of potentially suitable cropland, 2943 km2 of general expansion potential, and 1137 km2 of candidate spatial priority areas for field verification. The last tier was concentrated mainly in Xuanzhou District, Lujiang County, urban Wuhu, Wuwei City, Nanling County, and He County. The framework provides a cautious spatial-screening basis for optimizing winter–fallow cropland use and guiding subsequent field and feasibility assessments.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15090410

Authors: Zhong Sun Yukun Zhang Wei Wang

In response to the escalating challenges of global climate change and extreme meteorological events, this study addresses the critical issue of low disaster resistance in socialist-built heritage (SBH, i.e., built cultural heritage constructed during China’s socialist construction stage from 1949 to 1978). Focusing on Henan Province, China, with meteorological observation datasets spanning 1990–2023 and 1362 SBH sites constructed from 1949 to 1978, we innovatively construct a four-dimensional risk assessment model integrating Meteorological–Environmental Hazard (H), Exposure (E), Vulnerability (V), and Ecological Resilience (R). By synthesizing Spearman correlation analysis, Variance Inflation Factor (VIF) diagnostics, and piecewise regression, this research elucidates the spatial differentiation and driving mechanisms of meteorology-induced compound disaster risks. Key findings identify Zhengzhou and Kaifeng as core high-risk zones and precisely quantify a critical hazard threshold of 0.4 with a 95% breakpoint confidence interval [0.372, 0.428], passing the Chow test p < 0.001 and validated via historical heritage damage records (R2 = 0.78). Variable-importance analysis proves that extreme precipitation (32.7%) and annual strong-wind days (28.1%) are the dominant meteorological–environmental drivers, alongside the significant buffering effect of ecological resilience. Ultimately, this study refines the disaster risk assessment framework for SBH, providing robust theoretical support and a decision-making basis for adaptive conservation strategies in similar regions. The average disaster risk of SBH in Henan is 0.421 ± 0.153; for each 0.1 growth of H-E-V coupling term above the threshold of 0.4, disaster risk rises by 37.2%.

ISPRS International Journal of Geo-Information doi: 10.3390/ijgi15090409

Authors: Ratovoson Robert Andriambololonaharisoamalala Petra Helmholz Ivana Ivánová Dimitri Bulatov Eriita Jones Susannah Soon Yongze Song

Climate change and rapid urbanisation are intensifying the urban heat island effect, increasing thermal stress, degrading air quality, and leading to rising energy demand. Monitoring neighbourhood-scale heat requires Land Surface Temperature (LST) observations at fine spatial and temporal resolutions, yet satellite thermal products are limited by revisit frequency, acquisition time, and cloud cover. This study developed a novel approach integrating satellite-derived land cover characteristics with continuous contact-based temperature measurements from low-cost LoRaWAN sensors and geostatistical modelling to generate hourly LST maps at 10 m resolution. The technique provides communities with simpler, affordable methods for measuring heat islands and supporting mitigation strategies. Observations from 52 locations across Curtin University’s Bentley campus in Perth, Western Australia, were combined with land cover indices. Empirical Bayesian Kriging captured spatial and temporal urban heat patterns with a root mean square error of approximately 3 °C, representing a bias of near 1 °C. Predictions were consistent with Landsat-derived LST, revealing persistent heat retention over asphalt and cooler conditions associated with vegetation. Integrating satellite-derived predictors with ground measurements provides continuous fine-scale information to identify local heat hotspots and inform targeted mitigation. Unlike satellite data, these low-cost ground measurements could be collected with the help of urban practitioners, developers, and academic institutions.