ABSTRACT With the acceleration of global climate change, the evolution of geographical environments has become increasingly complex. This study explores the mechanisms through which topography-climate coupling influences extreme precipitation events in China. Using integrated geographic datasets on precipitation, temperature, and land use/cover, combined with GIS technology and a random forest regression (RFR) model, the spatiotemporal evolution from 1990 to 2020 is analysed. Unlike previous studies that only focused on a single factor or pairwise correlations, this study innovatively quantifies the nonlinear coupling effects of terrain features, climate variables, and land use/cover change (LUCC) on extreme precipitation’s spatial distribution and intensity. The results show that topographic factors, including elevation, slope, and terrain variation, significantly impact the spatial distribution and intensity of rainfall. Extreme precipitation events are spatially clustered, mainly in the southern Qinghai-Tibet Plateau, the southeastern coastal region, and the hilly areas of the middle and lower Yangtze River. These high-risk areas are characterized by complex terrain, steep slopes, and high temperatures. Moreover, the occurrence of extreme rainfall is found to be driven by multi-factor interactions rather than by a single factor. The prediction model demonstrates high accuracy (R2 = 0.85, MSE = 0.00023), providing valuable insights for disaster prevention and geographical environmental research.

ABSTRACT Geographic Information Systems (GIS) offer powerful analytical capabilities for digital humanities research, yet technical barriers and reproducibility challenges limit their adoption by humanities scholars. This paper presents a workflow-based approach that democratizes reproducible spatial analysis by transforming complex GIS operations into accessible, executable workflows using the open-source KNIME platform. Our methodological framework addresses three critical challenges: technical complexity through intuitive workflow design, reproducibility through complete process documentation, and accessibility through browser-based deployment. The approach integrates specialized components for historical data processing – including temporal uncertainty handling and spatial disambiguation – within standardized, shareable workflows that preserve analytical transparency while requiring no GIS expertise. We demonstrate the framework’s effectiveness through a comprehensive case study analysing spatial mobility patterns of pre-modern Chinese literati, where researchers successfully performed complex network analysis, trajectory visualization, and spatiotemporal pattern detection using our workflow-based tools. Results confirm that the approach enables exact analytical replication while significantly lowering technical barriers for humanities researchers. The framework’s modular design supports adaptation to diverse historical spatial research contexts, establishing a replicable methodology for democratizing GIS in digital humanities. This workflow-based approach contributes to more inclusive and reproducible spatial humanities scholarship by making advanced geospatial analysis accessible to researchers regardless of technical background.

ABSTRACT Large Language Models (LLMs) are transforming geospatial artificial intelligence (GeoAI), offering new capabilities in data processing, spatial analysis and decision support. This paper examines the open-source paradigm’s critical role in this transformation. While proprietary LLMs offer accessibility, they often limit the customization, interoperability and transparency vital for specialized geospatial tasks. Conversely, open-source alternatives significantly advance Geographic Information Science (GIScience) by fostering greater adaptability, reproducibility and community-driven innovation. Open frameworks empower researchers to tailor solutions, integrate cutting-edge methodologies (e.g. reinforcement learning, advanced spatial indexing) and align with FAIR (Findable, Accessible, Interoperable and Reusable) principles. However, the growing reliance on any LLM necessitates careful consideration of security vulnerabilities, ethical risks and robust governance for AI-generated geospatial outputs. This paper argues that GIScience advances best not through a single model type, but by cultivating a diverse, interoperable ecosystem combining open-source foundations for innovation, custom geospatial models and interdisciplinary collaboration. By critically evaluating the opportunities and challenges of open-source LLMs within the broader GeoAI landscape, this work contributes to a thorough discourse on leveraging LLMs to effectively advance spatial research, policy and decision-making in an equitable, sustainable and scientifically rigorous manner.

ABSTRACT This paper presents a two-stage modelling framework for land suitability evaluation that integrates geographic information system (GIS)-enabled multi-criteria decision analysis (MCDA) with economic simulation under uncertainty. The first stage applies a hybrid MCDA method combining entropy weighting and the analytical hierarchy process to generate spatially explicit suitability maps incorporating biophysical, social, and sustainability criteria. In the second stage, Monte Carlo simulation is used to evaluate the economic performance of alternative land use scenarios, addressing variability in key input parameters such as yield, cost, and price. Applied in a tropical case study context, the framework enables probabilistic assessment of land allocation strategies and supports more robust decision-making in estate crop planning. By decoupling suitability modelling from deterministic economic assumptions, this approach enhances the transparency, flexibility, and realism of land use evaluation. The integration of spatial MCDA and stochastic simulation demonstrates a transferable method for supporting land use decisions in data-limited but uncertainty-prone environments.

ABSTRACT Machine learning for remote sensing imaging relies on up-to-date and accurate labels for model training and testing. Labelling remote sensing imagery is time and cost intensive, requiring expert analysis. Previous labelling tools rely on pre-labelled data for training in order to label new unseen data. In this work, we define an unsupervised pipeline for finding and labelling geographical areas of similar context and content within Sentinel-2 satellite imagery. Our approach removes limitations of previous methods by utilizing segmentation with convolutional and graph neural networks to encode a more robust feature space for image comparison. Unlike previous approaches we segment the image into homogeneous regions of pixels that are grouped based on colour and spatial similarity. Graph neural networks are used to aggregate information about the surrounding segments enabling the feature representation to encode the local neighbourhood whilst preserving its own local information. This reduces outliers in the labelling tool, allows users to label at a granular level, and allows a rotationally invariant semantic relationship at the image level to be formed within the encoding space. Our pipeline achieves high contextual consistency, with similarity scores of SSIM = 0.96 and SAM = 0.21 under context-aware evaluation, demonstrating robust organization of the feature space for interactive labelling.

ABSTRACT This study introduces a novel methodology for the automatic generation of LOD2 3D building models for the historical city of Olomouc (Czechia) using Esri CityEngine. The methodology was demonstrated through three sample areas within Olomouc, addressing key challenges in urban modelling such as roof type classification, building height estimation, and the procedural generation of detailed roof structures. By integrating LiDAR data, orthoimagery, and building footprint datasets, the approach produces accurate spatial representations adaptable to diverse urban settings and represents a new approach tailored to historical Central European cities. The approach explicitly accounts for irregular parcels, heterogeneous roof geometries, and complex reconstruction patterns typical of such historical urban environments. A reusable Computer-Generated Architecture (CGA) script was developed to streamline the modelling process. The study includes an accuracy assessment to validate the reliability of the methodology. Accuracy assessment demonstrated the high reliability of the approach, with median-based ridge and eave height estimation achieving RMSE values of below 0.5 m in residential areas and approximately 1 m in rural areas. Roof type classification achieved accuracies of up to 0.89 for flat and 0.87 for gable roofs, although the accuracy for hip roofs remained lower. The outcome of the proposed approach is a processing pipeline created using Esri Tasks, a workflow automation tool in ArcGIS Pro that guides users through predefined steps, enabling them to apply the methodology to other cities and contribute to novel, scalable solutions, advanced GIS workflows, and enhanced urban planning.

ABSTRACT The relationship between democracy and economic growth has been widely debated since the 1980s, yet economists have not reached a consensus on its impact. This study examines this relationship using a spatial econometric approach across 100 countries from 1990 to 2022, incorporating both economic factors and spatial dynamics, such as geographical proximity and democratic context. A key innovation is the concept of institutional proximity, which posits that countries with similar institutional frameworks experience similar economic growth outcomes. The study provides strong evidence of positive growth spillovers from democracy, where democratic governance in one country enhances economic performance in other democratic nations, highlighting the interconnectedness of political and economic systems. Additionally, it uncovers spillover effects in which democratic countries transfer technology to less democratic regions, thus fostering global economic advancement. Notably, these effects extend beyond geographical proximity, with countries such as Australia, New Zealand, and Japan benefiting from democracy’s influence despite being distant from the epicentres of democracy. This challenges the notion that democracy’s impact is geographically confined, demonstrating its potential to drive global growth across national borders.

ABSTRACT The problem of data-driven location prediction of individual users is based on an effective mining of travel behaviours and motion patterns. In this sense, a central aspect refers to the use of a relevant amount of historical mobility data, required for successfully training machine learning models, especially when involving artificial neural networks. However, such data are sensitive in nature, therefore not easily available and always subjected to privacy-related restrictions on their public share. With the purpose of merging information from different providers without directly sharing geo-private data, we hereby assess the feasibility of decentralized training over multiple data sources, leveraging different unmergeable trajectory datasets stored in separate servers. In particular, we integrate a long short-term memory (LSTM) recurrent neural network framework for location prediction into a federated learning environment, whereby local workers compute the network operations on their data share, and the learning results are progressively synchronized with a parameter server. Variants of federated algorithms are evaluated and compared to separate independent training processes (lower benchmark) and an ideal, but in fact not allowed, centralized training (upper benchmark). By leveraging real-world datasets of sparse non-repetitive mobility traces, our experiments aim to disclose insights on federated learning strategies for advanced trajectory analytic tasks, paving the way to decentralized applications involving multiple geo-private data sources.

ABSTRACT The present article describes an educational resource for the geovisualization of research conducted on the shaping and dissemination of the current administrative divisions in Spain (provinces), designed in the first third of the 19th century. To this end, the digital technology StoryMaps was used, as it allows adding a narrative to the maps, facilitating their correct interpretation in their historical context. A specific methodology is presented for the creation of a digital resource for historical geography. The final product is the result of exhaustive research on historical cartography that comprises 772 maps (whole or an enlargement of a fragment) found in diverse libraries and map libraries, mainly in Spain and France. Our resource allows the visualization, interaction, and analysis of the shaping of provincial divisions. At the same time, it promotes the spread of a cultural heritage of historical cartography which until now has been dispersed and often unknown. Furthermore, it provides a response to the educational need to make the knowledge of Spanish administrative divisions more attractive. The final product is an open digital resource for the transfer of research in academic and educational settings, which allows for the application of some of the Sustainable Development Goals.

ABSTRACT To address the challenges of accurately extracting target features from complex scenes in UAV remote sensing imagery and the susceptibility of small objects to being obscured by noise, this paper proposes a lightweight detection algorithm, RE-YOLO, based on YOLOv8n. First, a multi-scale convolutional module named RFCSConv, which integrates channel and spatial attention mechanisms based on Receptive Field Attention Convolution (RFAConv), replaces the original convolution layers. This enhances feature selection and fusion at multiple scales. Second, the Efficient Squeeze-and-Excitation Module (ESEModule) is introduced into the backbone to strengthen feature representation while reducing computational overhead. Lastly, a composite loss function called Win-IoU, combining Wise-IoU (WIoU) and Inner-IoU, is proposed to dynamically adjust gradient contributions based on anchor quality. Experimental results on the VisDrone2019 dataset demonstrate that RE-YOLO achieves 29.7% mAP@0.5 with only 3.2MB of parameters and a real-time speed of 150 FPS. The algorithm also generalizes well across the HRSID and CARPK datasets, achieving 91.8% and 94.3% mAP@0.5 respectively.