ISPRS International Journal of Geo-Information

24 pages, 3089 KB

Open AccessArticle

Massively Parallel Lagrangian Relaxation Algorithm for Solving Large-Scale Spatial Optimization Problems Using GPGPU

by Ting L. Lei, Rongrong Wang and Zhen Lei

ISPRS Int. J. Geo-Inf. 2025, 14(11), 419; https://doi.org/10.3390/ijgi14110419 (registering DOI) - 26 Oct 2025

Lagrangian Relaxation (LR) is an effective method for solving spatial optimization problems in geospatial analysis and GIS. Among others, it has been used to solve the classic p-median problem that served as a unified local model in GIS since the 1990s. Despite [...] Read more.

Lagrangian Relaxation (LR) is an effective method for solving spatial optimization problems in geospatial analysis and GIS. Among others, it has been used to solve the classic p-median problem that served as a unified local model in GIS since the 1990s. Despite its efficiency, the LR algorithm has seen limited usage in practice and is not as widely used as off-the-shelf solvers such as OPL/CPLEX or GPLK. This is primarily because of the high cost of development, which includes (i) the cost of developing a full gradient descent algorithm for each optimization model with various tricks and modifications to improve the speed, (ii) the computational cost can be high for large problem instances, (iii) the need to test and choose from different relaxation schemes, and (iv) the need to derive and compute the gradients in a programming language. In this study, we aim to solve the first three issues by utilizing the computational power of GPGPU and existing facilities of modern deep learning (DL) frameworks such as PyTorch. Based on an analysis of the commonalities and differences between DL and general optimization, we adapt DL libraries for solving LR problems. As a result, we can choose from the many gradient descent strategies (known as “optimizers”) in DL libraries rather than reinventing them from scratch. Experiments show that implementing LR in DL libraries is not only feasible but also convenient. Gradient vectors are automatically tracked and computed. Furthermore, the computational power of GPGPU is automatically used to parallelize the optimization algorithm (a long-term difficulty in operations research). Experiments with the classic p-median problem show that we can solve much larger problem instances (of more than 15,000 nodes) optimally or nearly optimally using the GPU-based LR algorithm. Such capabilities allow for a more fine-grained analysis in GIS. Comparisons with the OPL solver and CPU version of the algorithm show that the GPU version achieves speedups of 104 and 12.5, respectively. The GPU utilization rate on an RTX 4090 GPU reaches 90%. We then conclude with a summary of the findings and remarks regarding future work. Full article

19 pages, 2598 KB

Open AccessArticle

DOCB: A Dynamic Online Cross-Batch Hard Exemplar Recall for Cross-View Geo-Localization

by Wenchao Fan, Xuetao Tian, Long Huang, Xiuwei Zhang and Fang Wang

ISPRS Int. J. Geo-Inf. 2025, 14(11), 418; https://doi.org/10.3390/ijgi14110418 (registering DOI) - 26 Oct 2025

Abstract

Image-based geo-localization is a challenging task that aims to determine the geographic location of a ground-level query image captured by an Unmanned Ground Vehicle (UGV) by matching it to geo-tagged nadir-view (top-down) images from an Unmanned Aerial Vehicle (UAV) stored in a reference [...] Read more.

Image-based geo-localization is a challenging task that aims to determine the geographic location of a ground-level query image captured by an Unmanned Ground Vehicle (UGV) by matching it to geo-tagged nadir-view (top-down) images from an Unmanned Aerial Vehicle (UAV) stored in a reference database. The challenge comes from the perspective inconsistency between matched objects. In this work, we propose a novel metric learning scheme for hard exemplar mining to improve the performance of cross-view geo-localization. Specifically, we introduce a Dynamic Online Cross-Batch (DOCB) hard exemplar mining scheme that solves the problem of the lack of hard exemplars in mini-batches in the middle and late stages of training, which leads to training stagnation. It mines cross-batch hard negative exemplars according to the current network state and reloads them into the network to make the gradient of negative exemplars participating in back-propagation. Since the feature representation of cross-batch negative examples adapts to the current network state, the triplet loss calculation becomes more accurate. Compared with methods only considering the gradient of anchors and positives, adding the gradient of negative exemplars helps us to obtain the correct gradient direction. Therefore, our DOCB scheme can better guide the network to learn valuable metric information. Moreover, we design a simple Siamese-like network called multi-scale feature aggregation (MSFA), which can generate multi-scale feature aggregation by learning and fusing multiple local spatial embeddings. The experimental results demonstrate that our DOCB scheme and MSFA network achieve an accuracy of 95.78% on the CVUSA dataset and 86.34% on the CVACT_val dataset, which outperforms those of other existing methods in the field. Full article

(This article belongs to the Topic Artificial Intelligence Models, Tools and Applications)

► Show Figures

Figure 1

33 pages, 8186 KB

Open AccessArticle

The Threshold Effect in the Street Vitality Formation Mechanism

by Yilin Ke, Jiawen Wang, Shiping Lin, Jilong Li, Niuniu Kong, Jie Zeng, Jiacheng Chen and Ke Ai

ISPRS Int. J. Geo-Inf. 2025, 14(11), 417; https://doi.org/10.3390/ijgi14110417 (registering DOI) - 24 Oct 2025

Viewed by 60

Abstract

Street vitality has become a crucial metric for smart city management. Classical theories qualitatively explain that street vitality originates from the dynamic interaction between people and spatial carriers, yet the threshold effect within this process has not been addressed, leaving a gap in [...] Read more.

Street vitality has become a crucial metric for smart city management. Classical theories qualitatively explain that street vitality originates from the dynamic interaction between people and spatial carriers, yet the threshold effect within this process has not been addressed, leaving a gap in urban research. This study selects South China, one of China’s most vibrant and globally influential regions, introduces dissipative structure theory based on classical theories, and constructs a threshold effect hypothesis model for the vitality formation mechanism. Through energy efficiency conversion of data and a slope-based method for identifying balanced time periods, the periods of supply–demand balance in energy efficiency were identified, the threshold effect in vitality formation was captured, and critical thresholds were measured. The results indicate the following: (1) the hypothesis model is valid; (2) the threshold effect is inevitable and periodic, primarily occurring on workdays from 12:00 to 13:00 and 18:00 to 19:00, and on rest days from 08:00 to 09:00 and 18:00 to 19:00; and (3) the activation threshold is quantifiable and exhibits volatility, ranging from 0.40 to 1.56, varying specifically by city, season, day type, and street type. This study advances the translation of street vitality research from theory into practice and provides theoretical support and strategic guidance for smart city management globally, particularly in developing countries. Full article

(This article belongs to the Topic Digital and Intelligent Technologies and Application in Urban Construction, Operation, Maintenance, and Renewal)

► Show Figures

Figure 1

26 pages, 2890 KB

Open AccessReview

A Review of Google Earth Engine for Land Use and Land Cover Change Analysis: Trends, Applications, and Challenges

by Bader Alshehri, Zhenyu Zhang and Xiaoye Liu

ISPRS Int. J. Geo-Inf. 2025, 14(11), 416; https://doi.org/10.3390/ijgi14110416 - 24 Oct 2025

Viewed by 245

Abstract

Google Earth Engine (GEE) has become one of the most widely used platforms for Land Use and Land Cover (LULC) research, offering cloud-based access to petabyte-scale datasets and scalable analytical tools. While earlier reviews provided valuable overviews of data and applications, this study [...] Read more.

Google Earth Engine (GEE) has become one of the most widely used platforms for Land Use and Land Cover (LULC) research, offering cloud-based access to petabyte-scale datasets and scalable analytical tools. While earlier reviews provided valuable overviews of data and applications, this study synthesizes 72 selected articles published between 2016 and February 2025 to examine the evolution of GEE–LULC research. Results show exponential growth in publications, with Landsat and Sentinel imagery dominating datasets and Random Forest (RF) and Support Vector Machine (SVM) remaining the most common classifiers. Geographically, output is concentrated in China and India, reflecting regional leadership in GEE adoption. Despite its strengths, GEE faces persistent challenges, including memory limits, restricted support for advanced Deep Learning (DL), and reliance on labeled data. Promising directions include integrating few-shot semantic segmentation and hybrid workflows combining GEE scalability with local Graphics Processing Unit (GPU) computing. By bridging platform-focused and application-focused studies, this review provides a comprehensive synthesis of GEE–LULC research and outlines actionable pathways for advancing scalable and Artificial Intelligence (AI)-enabled geospatial analysis. Full article

► Show Figures

Figure 1

24 pages, 6909 KB

Open AccessArticle

LA-GATs: A Multi-Feature Constrained and Spatially Adaptive Graph Attention Network for Building Clustering

by Xincheng Yang, Xukang Xie and Dingming Liu

ISPRS Int. J. Geo-Inf. 2025, 14(11), 415; https://doi.org/10.3390/ijgi14110415 - 23 Oct 2025

Viewed by 167

Abstract

Building clustering is a key challenge in cartographic generalization, where the goal is to group spatially related buildings into semantically coherent clusters while preserving the true distribution patterns of urban structures. Existing methods often rely on either spatial distance or building feature similarity [...] Read more.

Building clustering is a key challenge in cartographic generalization, where the goal is to group spatially related buildings into semantically coherent clusters while preserving the true distribution patterns of urban structures. Existing methods often rely on either spatial distance or building feature similarity alone, leading to clusters that sacrifice either accuracy or spatial continuity. Moreover, most deep learning-based approaches, including graph attention networks (GATs), fail to explicitly incorporate spatial distance constraints and typically restrict message passing to first-order neighborhoods, limiting their ability to capture long-range structural dependencies. To address these issues, this paper proposes LA-GATs, a multi-feature constrained and spatially adaptive building clustering network. First, a Delaunay triangulation is constructed based on nearest-neighbor distances to represent spatial topology, and a heterogeneous feature matrix is built by integrating architectural spatial features, including compactness, orientation, color, and height. Then, a spatial distance-constrained attention mechanism is designed, where attention weights are adjusted using a distance decay function to enhance local spatial correlation. A second-order neighborhood aggregation strategy is further introduced to extend message propagation and mitigate the impact of triangulation errors. Finally, spectral clustering is performed on the learned similarity matrix. Comprehensive experimental validation on real-world datasets from Xi’an and Beijing, showing that LA-GATs outperforms existing clustering methods in both compactness, silhouette coefficient and adjusted rand index, with up to about 21% improvement in residential clustering accuracy. Full article

28 pages, 31501 KB

Open AccessArticle

A Comprehensive Modelling Framework for Identifying Green Infrastructure Layout in Urban Flood Management of the Yellow River Basin

by Kai Wang, Zongyang Wang, Yongming Fan and Yan Wu

ISPRS Int. J. Geo-Inf. 2025, 14(11), 414; https://doi.org/10.3390/ijgi14110414 - 23 Oct 2025

Viewed by 213

Abstract

The Yellow River Basin faces severe challenges in water security and ecological protection: at the basin scale, complex hydrological processes and fragile ecosystems undermine the water security pattern; at the local scale, waterlogging risks have intensified in Zhengzhou—a core city in the lower [...] Read more.

The Yellow River Basin faces severe challenges in water security and ecological protection: at the basin scale, complex hydrological processes and fragile ecosystems undermine the water security pattern; at the local scale, waterlogging risks have intensified in Zhengzhou—a core city in the lower reaches—impacting the city itself and also exerting negative effects on the basin’s water security. To address this, mapping the scientific layout of green infrastructure (GI) is urgent. However, existing studies on GI layout at the basin-urban scale have certain limitations: neglect of underlying surface spatial heterogeneity, insufficient integration of natural, hydrological and social factors’ synergies, and lack of research on large-scale basins and cities, especially ecologically sensitive areas with complex hydrological processes. To fill these gaps, this study proposes an integrated framework (SCS–GIS–MCDM) combining the SCS hydrological model, GIS spatial analysis, and multi-criteria decision making (MCDM). The SCS hydrological model is refined via localized parameter calibration for better accuracy; indicator weights are determined through the MCDM framework; and green infrastructure (GI) suitability maps are generated by integrating ArcGIS spatial analysis with fuzzy logic. Results show that (1) 6.8% of Zhengzhou is highly suitable for GI, mainly in riparian areas and the Yellow River alluvial plain; (2) sensitivity analysis confirms flooded areas and runoff corridors as key drivers; (3) spatial validation against government-issued ecological control zone plans demonstrates the model’s value in balancing flood safety and socio-economy. This framework provides a replicable application model for GI construction in cities along the Yellow River Basin, thereby supporting urban planners in making evidence-based decisions for sustainable blue–green space planning. Full article

(This article belongs to the Topic Spatial Decision Support Systems for Urban Sustainability)

► Show Figures

Figure 1

21 pages, 13551 KB

Open AccessArticle

A Risk Assessment Method of Three-Dimensional Low-Attitude Airspace Based on Multi-Source Data

by Keli Wang, Wenbin Yang, Yanru Huang, Yuhe Qiu, Wenjiang Huang and Peng Hu

ISPRS Int. J. Geo-Inf. 2025, 14(11), 413; https://doi.org/10.3390/ijgi14110413 - 23 Oct 2025

Viewed by 227

Abstract

The safe operation of low-altitude UAVs is crucial for the effective utilization of low-altitude airspace, necessitating the development of appropriate risk assessment methods to evaluate the associated operational risks. However, current research primarily focuses on two-dimensional risk assessments, with limited focus on assessing [...] Read more.

The safe operation of low-altitude UAVs is crucial for the effective utilization of low-altitude airspace, necessitating the development of appropriate risk assessment methods to evaluate the associated operational risks. However, current research primarily focuses on two-dimensional risk assessments, with limited focus on assessing risks across different heights, thus constraining the ability to guide UAV operations within three-dimensional airspace. In this study, we propose a three-dimensional airspace risk assessment method that integrates multisource data to estimate risks at various altitudes. First, we assess ground impact risks by considering factors such as population density, obstacle environment, and socioeconomic characteristics. Next, we develop a network signal evaluation model to estimate signal loss at various altitudes. Finally, we apply machine learning methods to classify multiple features to determine airspace risks at varying altitudes, resulting in a comprehensive three-dimensional risk map. The results indicate that the majority of the urban area falls within the low-risk category, accounting for approximately 84–87% of the city. High-risk regions are concentrated in central urban areas, with their proportion increasing from 5.9% at 30 m to 9.1% at 300 m. Although the overall trend remains broadly consistent across altitudes, the local variations highlight the necessity of three-dimensional risk evaluation. This three-dimensional risk map can effectively guide safe UAV operations across different altitude layers and provide valuable decision support for flight route planning. Full article

► Show Figures

Figure 1

25 pages, 8387 KB

Open AccessArticle

HFF-Net: An Efficient Hierarchical Feature Fusion Network for High-Quality Depth Completion

by Yi Han, Mao Tian, Qiaosheng Li and Wuyang Shan

ISPRS Int. J. Geo-Inf. 2025, 14(11), 412; https://doi.org/10.3390/ijgi14110412 - 23 Oct 2025

Viewed by 227

Abstract

Depth completion aims to achieve high-quality dense depth prediction from a pair of synchronized sparse depth map and RGB image, and it plays an important role in many intelligent applications, including urban mapping, scene understanding, autonomous driving, and augmented reality. Although the existing [...] Read more.

Depth completion aims to achieve high-quality dense depth prediction from a pair of synchronized sparse depth map and RGB image, and it plays an important role in many intelligent applications, including urban mapping, scene understanding, autonomous driving, and augmented reality. Although the existing convolutional neural network (CNN)-based deep learning architectures have obtained state-of-the-art depth completion results, depth ambiguities in large areas with extremely sparse depth measurements remain a challenge. To address this problem, an efficient hierarchical feature fusion network (HFF-Net) is proposed for producing complete and accurate depth completion results. The key components of HFF-Net are the hierarchical depth completion architecture for predicting a robust initial depth map, and the multi-level spatial propagation network (MLSPN) for progressively refining the predicted initial depth map in a coarse-to-fine manner to generate a high-quality depth completion result. Firstly, the hierarchical feature extraction subnetwork is adopted to extract multi-scale feature maps. Secondly, the hierarchical depth completion architecture that incorporates a hierarchical feature fusion module and a progressive depth rectification module is utilized to generate an accurate and reliable initial depth map. Finally, the MLSPN-based depth map refinement subnetwork is adopted, which progressively refines the initial depth map utilizing multi-level affinity weights to achieve a state-of-the-art depth completion result. Extensive experiments were undertaken on two widely used public datasets, i.e., the KITTI depth completion and NYUv2 datasets, to validate the performance of HFF-Net. The comprehensive experimental results indicate that HFF-Net produces robust depth completion results on both datasets. Full article

(This article belongs to the Topic 3D Computer Vision and Smart Building and City, 3rd Edition)

► Show Figures

Figure 1

23 pages, 6498 KB

Open AccessArticle

A Cross-Modal Deep Feature Fusion Framework Based on Ensemble Learning for Land Use Classification

by Xiaohuan Wu, Houji Qi, Keli Wang, Yikun Liu and Yang Wang

ISPRS Int. J. Geo-Inf. 2025, 14(11), 411; https://doi.org/10.3390/ijgi14110411 - 23 Oct 2025

Viewed by 276

Abstract

Land use classification based on multi-modal data fusion has gained significant attention due to its potential to capture the complex characteristics of urban environments. However, effectively extracting and integrating discriminative features derived from heterogeneous geospatial data remain challenging. This study proposes an ensemble [...] Read more.

Land use classification based on multi-modal data fusion has gained significant attention due to its potential to capture the complex characteristics of urban environments. However, effectively extracting and integrating discriminative features derived from heterogeneous geospatial data remain challenging. This study proposes an ensemble learning framework for land use classification by fusing cross-modal deep features from both physical and socioeconomic perspectives. Specifically, the framework utilizes the Masked Autoencoder (MAE) to extract global spatial dependencies from remote sensing imagery and applies long short-term memory (LSTM) networks to model spatial distribution patterns of points of interest (POIs) based on type co-occurrence. Furthermore, we employ inter-modal contrastive learning to enhance the representation of physical and socioeconomic features. To verify the superiority of the ensemble learning framework, we apply it to map the land use distribution of Bejing. By coupling various physical and socioeconomic features, the framework achieves an average accuracy of 84.33 %, surpassing several comparative baseline methods. Furthermore, the framework demonstrates comparable performance when applied to a Shenzhen dataset, confirming its robustness and generalizability. The findings highlight the importance of fully extracting and effectively integrating multi-source deep features in land use classification, providing a robust solution for urban planning and sustainable development. Full article

► Show Figures

Figure 1

23 pages, 4871 KB

Open AccessArticle

Characterization and Modelling of Environmental Crime: A Case Study Applied to the Canary Islands (Spain)

by Lorenzo Carlos Quesada-Ruiz, Nicolás Ferrer-Valero and Leví García-Romero

ISPRS Int. J. Geo-Inf. 2025, 14(11), 410; https://doi.org/10.3390/ijgi14110410 - 22 Oct 2025

Viewed by 227

Abstract

The escalating environmental crisis and the threat posed by environmental crime demand more effective prevention strategies. The predictive mapping of environmental crimes can address this challenge by improving monitoring and response. This study proposes an analysis and modelling of the occurrence of environmental [...] Read more.

The escalating environmental crisis and the threat posed by environmental crime demand more effective prevention strategies. The predictive mapping of environmental crimes can address this challenge by improving monitoring and response. This study proposes an analysis and modelling of the occurrence of environmental crimes in the Canary Islands, a territory of exceptional ecological value and strong tourism and urban sprawl pressures. Four types of illegal activity were examined: buildings and constructions, mining and tilling, solid waste dumping, and liquid waste discharging. A predictive modelling framework based on Random Forest (RF) machine learning algorithms was applied to identify spatial patterns and environmental crime potential. A colour-based environmental crime potential map was generated for each island, showing the likelihood of 0, 1, 2, 3, or all 4 types of environmental crime. Findings reveal that 43.2% of the surface area of the islands could potentially be affected by at least one crime type. Potential occurrences are lower in protected natural areas, in islands with lower population densities and in inland areas compared to coastal regions. The methodology provides a foundation for future research which could assist policymakers and environmental protectors in combating and preventing environmental crimes more effectively and contribute to the preservation of their ecosystems. Full article

► Show Figures

Graphical abstract

23 pages, 7037 KB

Open AccessFeature PaperArticle

Are Sport Clubs Mediating Urban Expressive Crimes?—London as the Case Study

by Rui Wang, Yijing Li, Sandeep Broca, Zakir Patel and Inderpal Sahota

ISPRS Int. J. Geo-Inf. 2025, 14(11), 409; https://doi.org/10.3390/ijgi14110409 - 22 Oct 2025

Viewed by 210

Abstract

The study is referenced by interdisciplinary theories, i.e., routine activity, and social cohesion, to investigate the impacts of sport clubs and events on London’s expressive crimes at varied geographical scales, by utilizing Geographical-temporally weighted regression model. It has identified the spatial patterns of [...] Read more.

The study is referenced by interdisciplinary theories, i.e., routine activity, and social cohesion, to investigate the impacts of sport clubs and events on London’s expressive crimes at varied geographical scales, by utilizing Geographical-temporally weighted regression model. It has identified the spatial patterns of effects from sport clubs’ onto local expressive crimes among London wards, with several boroughs standing out for their being significantly affected. The case study in the home borough of the Hotspur Football Club has further been conducted, by proving the seasonal influences of sports clubs on reducing youth violence within school terms. It was also found disproportional increases in expressive crimes on Premier League match days, especially when receiving the results of draw. The data-driven evidence has generated insights on localized policies and strategies on developing tailored sports to support local young people’s development; pinpointing the optimisation of police forces resources on stop and search practices during sports events in hot spot stadiums. The methodology and workflow had also been proved with high replicability into other UK cities. Full article

► Show Figures

Figure 1

21 pages, 3716 KB

Open AccessArticle

Monte Carlo-Based Spatial Optimization of Simulation Plots for Forest Growth Modeling

by Milan Koreň, Peter Márton, Mosab Khalil Algidail Arbain, Peter Valent, Roman Sitko and Marek Fabrika

ISPRS Int. J. Geo-Inf. 2025, 14(11), 408; https://doi.org/10.3390/ijgi14110408 - 22 Oct 2025

Viewed by 241

Abstract

Accurate placement and geometry of simulation plots are essential for spatially explicit modeling of forest ecosystems. This study introduces a Monte Carlo-based approach for optimizing the spatial alignment of simulation plots with their source polygons, improving their ability to represent stand-level heterogeneity. The [...] Read more.

Accurate placement and geometry of simulation plots are essential for spatially explicit modeling of forest ecosystems. This study introduces a Monte Carlo-based approach for optimizing the spatial alignment of simulation plots with their source polygons, improving their ability to represent stand-level heterogeneity. The method is implemented in GenSimPlot, an open-source Python plugin for QGIS (version 3.30) that automates the generation, placement, and refinement of simulation plots using simple geometric shapes. Monte Carlo optimization iteratively adjusts translation, rotation, and scaling parameters to maximize spatial congruence, thereby enhancing the fidelity of forest growth simulations. A built-in hyperparameter tuning module based on random search enables users to explore optimal parameter settings systematically. In addition, GenSimPlot supports the extraction of qualitative and quantitative environmental variables and terrain from raster datasets, facilitating integration with forest growth models and broader ecological simulations. The proposed approach improves plot representativeness and enables robust scenario analysis across heterogeneous landscapes. Full article

► Show Figures

Figure 1

Journal Menu

Journal Browser

ISPRS Int. J. Geo-Inf., Volume 14, Issue 11 (November 2025) – 12 articles

Further Information

Guidelines

MDPI Initiatives

Follow MDPI