Conference Agenda

Land Use (LU) and Land Cover (LC) data allow us to understand the physical and human activities associated with a given land. Thus, LULC is a dynamic and highly researched field. LULC review papers are numerous and provide high-level insights about the proposed approaches, the data used, the study cases, the strengths and limitations, and the identification of new research gaps. Nevertheless, these reviews are not systematic and reproducible. The goal of this work is to propose an ontology to help the research community conduct systematic and shareable literature reviews and comparable analytical analyses of scientific papers. To achieve this, we formalize their metadata, content, strengths, and weaknesses. In particular, we consider the scientific paper as the central element of our ontology and we define formal semantics for all relevant items (data process, LULC life cycle and scientific paper). We hope to open the path to more efficient synthesis, discovery, and reuse of research outcomes from the literature. To facilitate the instantiation process and make it accessible to a broader range of researchers, we designed a tabular-based template. We used our template to simulate the process of conducting a literature review on three use cases: building function, global land cover mapping, and multi-class change detection.

Accurate segmentation of individual tree crowns (ITCs) from remote-sensing imagery is essential for forest monitoring and ecological analysis, yet remains challenging due to overlapping canopies and structural variability. The Segment Anything Model (SAM) shows strong generalization capabilities but requires effective prompting and domain adaptation for remote sensing applications. In this study, we investigate a lightweight fine-tuning strategy using Low-Rank Adaptation (LoRA) to adapt SAM for ITC segmentation on the BAMFORESTS dataset. The impact of different prompting strategies is evaluated, including manually annotated point and bounding box prompts, as well as automatically generated bounding boxes derived from a pre-trained tree detector. SAM is fine-tuned with instance-level ITC masks, enabling prompt-aware segmentation of multiple tree crowns per image. Performance is assessed before and after fine-tuning using standard instance segmentation metrics, including IoU and F1-score. Results show that LoRA-based adaptation improves mask delineation and robustness to prompt variability, with bounding box prompts consistently outperforming point-based inputs. Automatically generated prompts enable a fully automated workflow, although their effectiveness depends on detection quality. Evaluation on an independent validation site with manually annotated ITC labels shows that the fine-tuned LoRA-SAM model achieves performance comparable to manual annotations, while significantly reducing annotation effort. These findings highlight the importance of prompt design in adapting foundation models for remote sensing tasks and demonstrate that parameter-efficient fine-tuning provides a practical pathway toward scalable ITC segmentation.

This research rigorously evaluates the transferability of the pre-trained FLAIR-HUB deep learning model, developed by the French National Institute of Geographical and Forest Information (IGN), in terms of spatial generalizability and multi-resolution robustness, when transferred from its native French domains to the complex urban-agricultural landscape of Iasi, Romania.

The core objective of this investigation is to test the model's performance stability across severe multi-resolution domain shifts and temporal scenarios. The model architecture is applied to orthophotos acquired over Iasi in 2019 (at 0.5 m resolution) and 2024 (at 0.2 m and at a very high resolution of 0.084 m), enabling a comprehensive assessment of cross-resolution and temporal robustness.

A novel validation framework is introduced, combining conventional 2D raster-based evaluation with a 3D point-wise assessment using semantically labeled UAV-derived point clouds. The results demonstrate strong performance for dominant classes such as buildings and herbaceous vegetation, with improved accuracy at higher spatial resolution, while stable classes such as buildings and impervious surfaces show a comparatively robust performance, confirming the model’s capability to consistently represent invariant land cover types. However, performance decreases for heterogeneous and vegetation-related classes due to seasonal variability and class complexity. The 3D validation reveals slightly lower but consistent results, highlighting its role as a more rigorous evaluation approach. Overall, the study confirms the potential of transferring pre-trained semantic segmentation models to new geographic contexts, while emphasizing the importance of spatial resolution, temporal consistency, and validation strategy.

Airborne Laser Scanning (ALS) data have been used to identify giant trees in Taiwan, yet current workflow included volunteers to visually inspect ALS profile images. This study proposed to replace the volunteer-based verification step by applying deep learning to ALS profile images. Candidate treetop locations were first extracted from a Canopy Height Model (CHM) using a 65 m threshold and local maxima filtering. For each candidate, a representative ALS profile image was generated following an automated angle-selection method based on terrain fitting.

An EfficientNetV2-S model was trained using volunteer-labelled profile images from previous nationwide surveys. After label cleaning, a refined dataset was constructed, and a hybrid resampling strategy was applied to address class imbalance. The final model achieved 99.0% overall accuracy, 98.1% precision, and 100% recall on the independent test set, successfully detecting every true giant tree.

To evaluate generalization, the model was applied to 97,487 candidates from the latest national ALS survey. Predictions exhibited a strongly bimodal confidence distribution, demonstrating stable between true and false positives and effectively reducing the manual inspection workload.

This study shows that deep learning can reliably replace crowdsourced verification, enabling scalable, supporting efficient updates of large-scale forest inventories.

We face a unprecedented amount of geospatial data, describing directly or indirectly the Earth Surface at multiple spatial, temporal, and semantic scales, and stemming from numerous contributors, from satellites to citizens. The main challenge in all the geospatial-related communities lies in suitably leveraging a combination of some of the sources for either a generic or a thematic application. Certain data fusion schemes are predominantly exploited: they correspond to popular tasks with mainstream data sources, e.g., free archives of Sentinel images coupled with OpenStreetMap data under an open and widespread deep-learning backbone for land-cover mapping purposes. Most of these approaches unfortunately operate under a "master-slave" paradigm, where one source is basically integrated to help processing the "main" source, without mutual advantages (e.g., large-scale estimation of a given biophysical variable using in-situ observations) and under a specific community bias.

We argue that numerous key data fusion configurations, and in particular the effort in symmetrizing the exploitation of multiple data sources, are insufficiently addressed while being highly beneficial for generic or thematic applications. Bridges and retroactions between scales, communities and their respective sources are lacking, neglecting the utmost potential of such a "global-local loop". In this paper, we propose to establish the most relevant interaction schemes through illustrative use cases. We subsequently discuss under-explored research directions that could take advantage of leveraging available data through multiples scales and communities.

C-band scatterometers have been collecting radar backscatter data since 1991, providing valuable long-term records for environmental monitoring applications such as soil moisture and vegetation dynamics. However, differences in sensor calibration between missions introduce biases that compromise the continuity of these data records. This paper presents the practical implementation and adaptation of Reimer's (2014) rainforest-based inter-calibration approach for ESA's ERS satellites (ESCAT) and MetOp/ASCAT instruments. We implement the method as a modern, open-source Python framework and apply it to the newly complete ERS data record (including ERS-1 data not available in the original study). The resulting calibrated backscatter data record will enable improved long-term monitoring of land surface dynamics with reduced mission-to-mission variability in bias and slope response over incidence angle.

Improved availability and quality of high-resolution satellite imagery allow for reliable airplane detection. Yet, fine-grained classification, especially of commercial airliners, remains a formidable challenge. Besides common difficulties, such as varying image

artifacts and occlusions, the main challenge lies in the strong visual similarity between airliner families. This paper presents a geometry-aware classification that enhances oriented object detectors by integrating absolute measures and geometric features – fuselage length, wingspan, wing sweep angle, engine count, and fuselage width – in the form of priors into a Bayesian maximum a posteriori (MAP) estimation. The proposed pipeline is detector-agnostic by updating class posteriors without retraining the main detector. On the Gaofen Challenge dataset, it results in consistent improvements based on untuned baseline detectors, which outperform the top scores of the sophisticated fine-tuned models. An oracle experiment reveals the potential of the approach with an upper limit of the overall mean Average Precision of up to 0.96 and 0.98 for Gaofen and SuperView data, respectively. Furthermore, the impact of the employed geometric attributes is quantitatively evaluated.

IFC models store detailed building information, but their complex schema and deeply nested property sets make querying difficult for non-expert users and challenging for large language models (LLMs) to handle directly. Current LLM-based approaches are inefficient because prompts often include entire IFC schemas, many properties of which are irrelevant to the user’s query, leading to higher inference costs and potential errors. This paper presents Query2Property, a semantic retrieval system that maps natural language queries to the most relevant IFC properties. By embedding both property descriptions and user queries in a shared vector space, the system retrieves contextually relevant properties for dynamic and concise prompt construction in LLM-driven workflows. Evaluation on 55 representative BIM queries achieves a top-1 accuracy of 87.3% and top-3 accuracy of 100%, demonstrating effective alignment with user intent. Query2Property simplifies LLM-based workflows over BIM data, supporting semantic search and natural language exploration of complex building information.

Semantically rich 3D city models play a vital role in a variety of applications, such as urban planning. Enhancing these models with currently unavailable attributes, such as building storey numbers, can unlock new opportunities to address pressing challenges, including sustainable urban development. In this work, we present an end-to-end pipeline for the automatic estimation of the number of storeys to semantically enrich 3D city models. We employ volunteered geographic information street-view imagery

from Mapillary, using a COCO-pretrained object detection model to identify windows in façade images as key visual indicators for inferring building storey counts. Our detection pipeline, based on the YOLOv3 architecture, estimates storey numbers using an ensemble of clustering methods including Gaussian Mixtures and DBSCAN and enables the automatic augmentation of CityGML-based 3D city models by filling in missing attributes. This enrichment supports advanced applications, such as assessing building-scale energy demand, evaluating vertical urban growth patterns or population density estimations. We validated the feasibility of our approach with unfiltered Mapillary and applied it to a district in the city of Heidelberg, Germany. The paper also includes a detailed discussion of learning process quality, integration workflows, and visualization of the enriched 3D city model. The developed code is available at: https://github.com/hcu-cml/citydb-buildingstoreys-ai.

This study introduces a Spatial constraint One-step Approach (SOA) and Temporal constraint One-step Approach (TOA) to improve the estimation of Terrestrial Water Storage Anomalies (TWSA) from GRACE and GRACE-FO satellite data. Traditional three-step or two-step post-processing methods sequentially apply spectral filtering and leakage correction, often causing signal attenuation, spatial leakage, and reliance on external models.

In contrast, the proposed one-step framework simultaneously estimates all signal components—including trends, seasonal cycles, and non-seasonal signals (NSS)—directly from unfiltered TWSAs within a region of interest. It incorporates full error covariance and models NSS using spatiotemporal constraints: TOA employs a Multi-Order Gauss-Markov process for temporal correlation, while SOA uses spatial covariance functions and a buffer zone to reduce boundary effects. Tikhonov regularization ensures solution stability.

Validation across major river basins and regions like Southeastern China shows that SOA/TOA outperforms conventional filters (e.g., DDK, IPF), reducing errors and improving agreement with mascon products and climate indices. The method also better identifies hydrological extremes (e.g., droughts, floods) and links them to climate drivers like ENSO, enhancing the monitoring and understanding of global water storage dynamics.

Groundwater dependent ecosystems (GDEs) are sustained by direct or indirect access to groundwater, relying on its flow or chemistry for their water needs. These ecosystems span aquatic, terrestrial, and subterranean realms, providing critical ecological functions, maintaining water quality, and supporting biodiversity and Indigenous land use. In Alberta’s boreal region, GDEs are abundant yet remain poorly mapped, limiting understanding of their extent and sensitivity to industrial development and hydrological change. Developing consistent, spatially explicit mapping tools is therefore essential for effective monitoring and management. This research develops and evaluates a remote sensing and machine learning (ML) framework for predicting GDE habitats across boreal Alberta, Canada, as part of a broader provincial effort toward consistent, high-resolution GDE mapping. Multi-sensor Earth observation and geospatial datasets were integrated using ensemble ML modelling to identify groundwater-dependent habitats. Specifically, the study aimed to (1) evaluate the performance of multiple ML algorithms and ensemble approaches for GDE prediction, (2) assess whether aquatic and terrestrial GDEs can be effectively modelled within a unified framework, and (3) identify the most influential environmental and remote sensing variables driving GDE occurrence. The resulting model ensemble achieved high predictive accuracy (AUC = 0.90), with wetland and hydrological variables emerging as dominant predictors. The approach provides a scalable, transferable methodology for regional GDE mapping to support groundwater management, ecosystem monitoring, and cumulative effects assessment across northern Alberta.

This study presents a hydrology-informed deep learning framework for supraglacial lake segmentation on the Greenland Ice Sheet using Sentinel-2 imagery. Traditional approaches to lake mapping rely primarily on spectral cues, which often struggle in regions with weak contrast, shadowing, or surface melt variability. To address these challenges, we incorporate physically meaningful hydrological features—flow accumulation, distance-to-drainage, and surface depressions—derived from high-resolution DEMs to guide the segmentation process. The proposed FiLM-based two-stream U-Net consists of an RGB stream for spectral–textural representation and a hydrology stream encoding surface meltwater routing patterns. Feature-wise linear modulation is applied at multiple levels of the RGB encoder–decoder to dynamically condition spectral features on hydrological context and improve spatial coherence.

Experiments on the SIGSPATIAL 2023 GISCUP dataset demonstrate that this architecture improves segmentation accuracy over a Sentinel-2-only baseline and a simple channel-concatenation model, particularly for small, fragmented, or spectrally ambiguous lakes. The combined use of hydrological cues and deep feature modulation reduces false positives in regions where meltwater is unlikely to accumulate and strengthens delineations along complex lake boundaries. These improvements highlight the value of integrating physically informed geospatial descriptors with modern segmentation networks for robust supraglacial lake detection.

Beyond methodological gains, the results support downstream applications including meltwater routing analysis, supraglacial drainage characterization, and improved understanding of seasonal lake evolution. Ultimately, this framework contributes to more reliable ice-sheet mass balance assessments and sea-level rise projections by enhancing the consistency and physical realism of supraglacial lake mapping at scale.

The rapid retreat and thinning of glaciers in the North-western Himalayas due to climate change have led to a significant increase in the number and size of glacial lakes. These high-altitude lakes, often dammed by unstable moraines, pose a growing threat of Glacial Lake Outburst Floods (GLOFs), which can cause catastrophic flash floods and endanger downstream communities. Accurate estimation of glacial lake bathymetry is crucial for GLOF risk assessment, but direct measurement is challenging due to inaccessibility and harsh conditions. This study presents a methodology for evaluating glacial lake bathymetry using remote sensing data, focusing on the Panikhar glacier lake in Ladakh, India. Time series analysis was conducted to map the lake's water spread from 2015 to 2024 using optical and synthetic aperture radar data. Three approaches were employed to estimate bathymetry: a radiative transfer model (RTM) based on multispectral reflectance, a topographical model using high-resolution digital elevation models, and empirical equations relating lake area to depth. The RTM approach relies on the optical properties of water, while the topographical model leverages the surrounding terrain to infer underwater topography. Empirical equations were drawn from established literature. Results were validated against physical bathymetry survey observations. Among the methods, topographical modeling demonstrated the highest potential for accurate depth estimation, as it directly incorporates the lake's topographic features. This study highlights the importance of integrating remote sensing techniques for effective GLOF hazard assessment in remote, high-altitude regions, offering a scalable solution for monitoring and mitigating risks associated with glacial lakes in the Himalayas.

Wildfires are an increasingly dominant disturbance in boreal and Arctic Canada, a trend projected to continue under a changing climate. The ecological and hydrological impacts of wildfires cascade into the abundant inland lakes in these interconnected northern landscapes, leading to post-fire changes in lake quality and colour. Previous in-situ studies on post-fire lake water quality in boreal regions have yielded inconsistent results, preventing a regional-scale understanding of the prevalence, magnitude, and duration of fire impacts on boreal lakes.

Here, we use harmonized Landsat time series to quantify fire-driven lake colour change and its controls across western boreal Canada. We studied 83 fires that burned 13,968 lakes during 2005 - 2015 and quantified lake colour dynamics through surface reflectance in the red wavelength, a proxy for suspended sediments and turbidity. Using a Difference-in-Difference approach, we found pervasive and long-lasting increases in lake colour driven by fire disturbance, beginning in the first post-fire summer and persisting for at least ten years, indicating sustained elevated suspended sediment concentrations and turbidity regardless of physiographic variations. The magnitude and temporal patterns of these changes varied, with burn severity and physiography as important controls. Severe burns in the Taiga and Shield zones underlain by extensive permafrost led to greater and more prolonged changes in lake colour.

These findings underscore the critical and growing role of wildfires in boreal lake quality change, with important implications for aquatic habitats and water resources in a fire-prone future.

An Open Source Framework for Routing and Event Management in University Campuses

5*S: Space, Surveyors & Students is a collaborative STEM outreach project lead by Maynooth University, in partnership with the Irish National Mapping Agency, Tailte Éireann, Technological University Dublin, Esri Ireland and the Society of Chartered Surveyors Ireland. Funded by Research Ireland and the European Space Education Research Office (Esero) Ireland, these groups have a shared interest to encourage student enrolment on 'geo' courses at university from under-represented groups and also to preempt a looming skills-gap. 5*S provides interactive and engaging educational content and training to teachers and students (12 to 18 years old) who are interested in learning more about satellites, spatial data and SDGs. Leveraging a combination of ArcGIS StoryMaps, a bespoke Augmented Reality app (SatelliteSkill5 - free to download on PlayStore and AppStore) and the National Geospatial Data platform, Geohive - students and teachers are provided with curriculum-focussed content that help teach how to harness the power of spatial data to solve a set of challenges. Framed around the United Nations Global Geospatial Information Management 14 Fundamental Geospatial Data Themes, each core piece of 5*S content topic is tailored to fit into a packed school curriculum and has been trialled in almost 20% of second level schools in Ireland. The learnings from this tailored content have been recorded and evaluated through a series of quantitative and qualitative respondent questionnaires and teacher focus groups/one-on-one interviews. The findings suggest cross curricular potential, value-add for schools and confirm the importance of this for encouraging data literacy and supporting teacher agency.

TorchGeo is a Python library bringing support for geospatial data to the PyTorch deep learning ecosystem. First released over four years ago, TorchGeo has always had strong support for 2D satellite image data. The upcoming TorchGeo 1.0 release will add complete time series support, including 1D through 4D data, requiring a complete rewrite of all GeoDatasets and GeoSamplers. This talk describes the 1.5 years of open source work required to enable full time series support and the backwards-incompatible changes coming to TorchGeo. It also demonstrates the power and simplicity of TorchGeo through a series of case studies: 1D) air pollution, 3D) change detection and land cover mapping, and 4D) weather forecasting and climate modeling. TorchGeo is open source and released under an MIT license, with over 140 built-in datasets, 130 foundation model weights, and 120 contributors from around the world.

The International Society for Photogrammetry and Remote Sensing Student Consortium (ISPRS SC) serves as the official representation of students and young professionals within ISPRS, connecting a global network of more than 900 active members from 64 countries as of November 2025. This paper presents a comprehensive overview of ISPRS SC activities during the 2022-2025 Board of Directors tenure, highlighting significant expansion in educational outreach and capacity building initiatives. Key achievements include facilitating 15 summer schools across seven countries, providing hands-on training in emerging geospatial technologies, and organizing more than 40 webinars through partnerships with 10 ISPRS Working Groups, demonstrating substantial growth from 2 webinars in 2022 to 24 in 2025. The consortium successfully launched 11 Student Chapters worldwide, establishing localized networks that promote inclusive access to geospatial education across diverse regions. Through quarterly publication of the SpeCtrum newsletter, maintenance of active social media presence across four platforms reaching over 10,000 followers, and organization of networking events at major ISPRS symposia, the consortium has strengthened its communication, networking and professional development opportunities. The paper also discusses operational challenges including funding constraints, geographic representation gaps, and Board capacity limitations, while outlining future initiatives including a mentorship program, virtual symposium, and comprehensive Congress 2026 activities. These efforts underscore ISPRS SC's evolving role in developing the next generation of geospatial professionals equipped to address global sustainability challenges.

The densification of GNSS Continuously Operating Reference Station (CORS) networks in mountainous regions is constrained by the high cost of geodetic-grade equipment. Low-cost (LC) multi-frequency GNSS receivers offer a viable alternative, yet their performance in challenging high-altitude Alpine environments remains largely unexplored. This study evaluates the rover-side positioning performance and tropospheric delay estimation capability of a newly installed LC permanent station at Prali (2200~m elevation), in the Alpine region of Piedmont, Italy. The station, based on a u-blox ZED-F9P receiver with a broadband LC antenna and a Raspberry Pi computer, was assessed using Virtual Reference Station (VRS) corrections from the SPIN3 professional CORS network. Six independent two-hour RTK sessions across a full diurnal cycle were processed using RTKLIB in forward-only kinematic mode to emulate real-time conditions. Results demonstrate that the LC station achieves centimetre-level horizontal precision (8--11~mm) with fix rates up to 97\% and time to first fix below 3~minutes under favourable conditions. A diurnal performance variability was observed and characterised across the six sessions. Zenith Tropospheric Delay estimation via CSRS-PPP with 92\% fixed ambiguities yielded physically consistent values (mean ZTD~=~1811~mm, ZWD~=~41~mm), consistent with dry winter conditions at altitude. These results confirm that LC GNSS stations can deliver reliable centimetre-level positioning and meaningful tropospheric products in demanding Alpine environments, supporting their deployment for CORS network densification in regions where geodetic-grade infrastructure is economically or logistically prohibitive.

Over the last few years immersive technologies have experienced rapid advancement providing several solutions in geospatial education such as improving student preparedness, enhancing student learning of theoretical concepts and practical procedures, and even supporting remote learning. However, several educators cannot utilize such immersive technologies because many of the

existing applications are not suitable for geospatial learning. Use of immersive technologies in education often necessitates specialized software and application development with the total investment (in terms of cost and time) becoming a barrier. This project is spearheaded by Working Group V/1 of ISPRS, and it is also supported by the Education and Capacity Building Initiative (ECBI) 2024 grant to provide sample experiences to educators. This project developed two immersive experiences relevant to geospatial education that can be used to enhance lab delivery and learning. The first experience uses a simplified GNSS receiver for

topographic mapping in virtual reality (VR). The second experience uses a tablet and an external GNSS receiver to visualize 3D objects in augmented reality (AR). To design these two applications the research team distributed a global questionnaire to professionals and educators. The questionnaire assisted in understanding the participant’s experience with immersive technologies, their attitude and beliefs towards these tools, and the potential benefits that immersive technologies can bring in education and industry. The results from the VR/AR implementation indicate that interactive environments can effectively support student preparation and reveal common misconceptions in topographic data collection, highlighting their value as both training and diagnostic tools in geospatial education.

The GeoRek project, funded by the DAAD within the German-Ukrainian University Network, aims to strengthen geospatial education in Ukraine through digitalization and international cooperation. Implemented by Jade University of Applied Sciences (Germany) together with Kyiv National University of Construction and Architecture (KNUCA), Dnipro University of Technology, and the University of Bamberg, the initiative develops innovative e-learning tools and micro-credential systems for geodetic reconstruction and high accuracy documentation.

A central element of the project is the VRscan3D - virtual laser scanner simulator — an educational platform that enables realistic training in terrestrial and mobile laser scanning without the expensive equipment. The system supports interactive learning, gamified exercises, and data export for advanced processing.

GeoRek further establishes micro-certificates in key subjects such as terrestrial laser scanning, photogrammetry, and 3D/BIM data processing, aligning with European standards (ECTS, EQF) to promote flexible and lifelong learning. The project’s applied component includes real-life case studies on the digital documentation for reconstruction of war-damaged buildings in Ukraine.

Overall, GeoRek exemplifies how modern digital education can strengthen academic resilience, support reconstruction, and deepen long-term German-Ukrainian cooperation in geospatial sciences.

To enable non-destructive quantitative identification of mixed salts in mural plaster layers, hyperspectral data were collected from Na₂SO₄-CaCl₂ mixed-salt samples. Based on these data, a method integrating spectral preprocessing, feature-set construction, and machine-learning inversion was proposed. First, the original spectra were preprocessed using Savitzky-Golay smoothing and multiplicative scatter correction. A 0.6-order fractional-order derivative (FOD) was then introduced to enhance subtle salt-related spectral features. Subsequently, 30 single-band features were selected using a two-step strategy involving competitive adaptive reweighted sampling for preliminary screening and variable importance in projection for secondary screening. On this basis, dual-band and tri-band spectral indices were further constructed, and a combined-band feature set was formed by integrating the three feature sets. Gaussian process regression (GPR) was used to compare the inversion performance of different feature-input strategies for Na₂SO₄ and CaCl₂ contents. The results showed that the 0.6-order FOD achieved a favorable balance between feature enhancement and noise suppression. Among the evaluated feature-input strategies, the combined-band model showed the best predictive performance for both Na₂SO₄ and CaCl₂. These results indicate that integrating complementary information from feature sets with different dimensions can improve the stability and accuracy of mixed-salt inversion, providing a useful reference for the hyperspectral non-destructive quantification of mixed salts in murals.

Map archives, serving as crucial cultural heritage documenting historical spatial information, face severe challenges in long-term preservation. To evaluate the feasibility of deacidification technology in the conservation of map archives, this study utilized 41 severely acidified early 20th-century map archives as samples. These were treated using a specific non-aqueous deacidification technology, and changes in pH value, color difference (ΔE), and inks stability before and after treatment were analyzed. The results indicate that after deacidification, the paper pH value significantly increased from an average of 4.48 to a range between 8.24 and 8.87. The color change was minimal, with an average color difference ΔE of only 1.62. This study verifies that the deacidification technology is suitable and effective for the deacidification treatment of acidified paper-based map archives, providing a safe and reliable method for preserving their cultural value.

To address the challenges of large initial pose discrepancies in grotto point clouds acquired from multiple sources, complex local geometric structures, significant noise interference, and the tendency of traditional ICP algorithms to fall into local optima, a high-precision point cloud registration method is proposed by integrating feature extraction with the collaborative optimization of coarse and fine registration. This method first performs point cloud preprocessing through voxel downsampling and outlier removal; it then extracts stable feature regions based on normal vector estimation and curvature analysis, and constructs feature representations using FPFH descriptors; building on this, the K-4PCS algorithm is employed to perform coarse registration and obtain optimal initial transformation parameters, followed by fine registration using an improved ICP algorithm combined with KD-tree-based search optimization. The proposed method was validated using the STANFORD DRAGON dataset and the point cloud of the Buddha head statue from Cave 18 of the Yungang Grottoes. The results indicate that the proposed method effectively improves the convergence speed and accuracy of point cloud registration. It demonstrates good stability and applicability in complex cave heritage scenarios and can provide methodological support for the fusion of multi-source point clouds in the digital preservation of cultural heritage.

To address the lack of suitable methods for automatic 2D line drawing generation from grotto wall mesh models, as well as the difficulty of existing methods in balancing structural representation and detail preservation, this paper proposes a line drawing generation method based on depth map and normal map fusion. The method first orthographically projects the 3D model into a depth map and a surface normal map, then constructs an initial line drawing pipeline based on projected edge fusion. A layered optimization strategy is further introduced to improve detail representation and result stability. Experiments on the mesh model of the north wall of Cave 18 at the Yungang Grottoes show that the projected edge fusion method is more suitable for overall structural representation, while the layered optimization method performs better in preserving weak structures and fine details. The proposed method effectively improves the quality of automatic 2D line drawing generation for grotto wall surfaces.

Rock-cut cave temples, such as the UNESCO World Heritage site of Yungang Grottoes, represent invaluable cultural heritage facing severe deterioration. Traditional monitoring methods are often slow, subjective, and inadequate for large-scale, long-term analysis, creating a critical gap in effective conservation.To address this challenge, we developed an automated framework for identifying surface deterioration features on stone carvings using deep learning. Our approach leverages a novel multi-source image dataset, combining historical and modern imagery of the Yungang Grottoes. We propose an enhanced model based on the YOLO architecture, featuring a synergistic semantic and spatial perception mechanism that significantly improves the detection of subtle features like peeling and cracks.The model was trained to recognize three key deterioration types: peeling, crack, and human damage. On-site deployment and testing in the authentic cave environments demonstrated excellent performance, achieving high recognition confidence for cracks (87.5%), peeling (85.2%), and human damage (81.3%). This study provides a powerful new tool for the quantitative monitoring of stone carvings, offering a scientifically-informed pathway for practical and proactive conservation strategies at heritage sites worldwide.

Most of the grotto temples have undergone long-term weathering and multiple repainting campaigns, so accurate identification of the composition and spatial abundance of surface pigments is an important foundation for pigment characterization and conservation research. This study focuses on the dome of Yungang Grotto 7. Data were acquired using a three-dimensional (3D) hyperspectral multimodal digital acquisition system and the Analytical Spectral Devices (ASD) field spectroradiometer. The workflow consisted of two stages: pigment identification and abundance inversion. In the pigment identification stage, a normalized weighted identification method integrating Spectral Angle Mapper (SAM) and the Normalized Difference Spectral Index (NDSI) was proposed based on mineral pigment reflectance curves measured by the ASD field spectroradiometer. In the abundance inversion stage, Fully Constrained Least Squares (FCLS) was applied to estimate pigment proportions in mixed pixels under non-negativity and sum-to-one constraints. The results show that the green pigments are most likely malachite and Paris green, the red pigments are most likely hematite and laterite, and the black pigment is most likely carbon black. The interwoven distribution of Paris green and traditional mineral pigments provides material-science evidence for modern repainting and restoration in this area. Nonlinear mixing may occur on rough and weathered grotto surfaces. However, under the current data conditions, its influence on abundance inversion remains unclear. Therefore, Kernel Fully Constrained Least Squares (K-FCLS) was additionally introduced as a reference nonlinear model for qualitative comparison with FCLS.

Capacity building is a key element in promoting and training in spatial technologies and in fostering a network of early-stage researchers for future collaborations. The first edition of the Earth Sensing Summer School, organised by the University of Padova with support from ISPRS Education and Capacity Building (ECBI) funds, was held from 7 to 13 September 2025 in the Dolomite area of the Alpine region. This article illustrates specific aspects of the organisation and discusses the return on investment in terms of training and networking. It highlights the methodology used for selecting participants and conducting the training, which included a balanced combination of seminars, fieldwork, data analysis, dissemination to peers, and a final defence of results. We discuss the outcomes and feedback from the almost 40 participants and provide ideas for future improvements, aiming to offer insights for fellow researchers who might want to replicate a capacity-building activity of this kind.

This practice paper presents the design, implementation, and first evaluation of Team-Based Learning (TBL) activities in university-level Geomatics courses taught in English to multicultural and international student groups. The study documents a structured pathway for adapting TBL to technically demanding subjects, including GIS suitability analysis, network analysis, remote sensing classification, and heat-risk assessment. Its main contribution lies in showing how a pedagogical model widely discussed in general higher education can be translated into software-based Geomatics teaching while supporting both disciplinary learning and intercultural collaboration. The paper also identifies the main organizational conditions for successful adoption, including team formation, workload calibration, and suitable classroom settings. Results from 12 TBL implementations involving 187 students and 470 total participations show clear benefits of teamwork: average team test performance was markedly higher than individual performance, repeated participation was associated with improved results, and student satisfaction increased after the introduction of TBL. Qualitative evidence further indicates gains in communication, teamwork, and intercultural interaction. Although the first implementation required substantial preparation effort, the approach proved replicable and scalable in subsequent editions, making TBL an effective instructional model for Geomatics education.

The ISPRS Student Consortium (SC) Summer Schools are one of the fundamental initiatives that ISPRS SC jointly organises with interested institutions to advance education outreach and capacity building in photogrammetry, remote sensing, and geospatial information sciences. Since their start in 2004, these programs have provided students and young researchers with immersive learning opportunities, combining technical lectures, hands-on sessions, and cultural experiences. Grounded in Experiential Learning Theory, the Summer Schools emphasise real-world application, reflective observation, and collaboration. This paper explores their evolution, global outreach, and educational impacts. Drawing on recent ISPRS SC Summer Schools, including the BUCEA Summer School 2024 on Smart Cities and the Summer School 2024 on AI for Geospatial Applications, the analysis highlights their integration of theory and practice, networking benefits, and transformative cultural exchange. Challenges such as financial barriers and technological gaps are discussed together with recommendations for sustaining and enhancing these initiatives. This study underscores the critical role of ISPRS SC Summer Schools in fostering a global community of geospatial practitioners to address real-world challenges.

This paper discusses the importance of integrating legal aspects into UAS mapping courses and related curricula providing a framework for integration in an introductory photogrammetric course with sample questions and assignments. Curricula focus is placed on two interrelated issues: first, the extent to which property owners maintain a reasonable expectation of privacy from UAS intrusion within the “immediate reaches” of their airspace; and second, the potential for UAS-mounted sensors to inadvertently capture imagery or point cloud of neighboring properties while operating in compliance with Federal Aviation Administration (FAA) regulations. The discussion concludes by identifying strategies to mitigate these legal and operational challenges, giving students the knowledge and tools to address similar situations in real-life scenarios, and ensuring that aerial surveying practices respect both regulatory compliance and property rights.

Remote sensing (RS) and especially earth observation (EO) have been used extensively for decades in environmental monitoring, infrastructure asset management, urban planning, emergency response, mapping and many others. The pace of technology

advancements in big data, cloud computing, Geospatial AI (GeoAI) and Geospatial Foundation Models (GeoFM) causes a paradigm shift on how to and who can maximize the potential of remote sensing technology. This paradigm shift challenges traditional geomatics education and pedagogical methods. Additionally, the gap between geomatics graduates’ skills and market needs is widening. The pace of disruptive technology advances like GeoAI and GeoFM often outpaces developments in geomatics education content or suitable pedagogical methods and formats. To address these skills gaps in geomatics courses and courseware, an initiative has been developed

between the Canadian Space Agency and Carleton University, involving more than a dozen different partners spanning industry, government, academia and NGOs. We have been gathering information through qualitative and quantitative techniques to obtain

insights about the soft and hard skills that are valuable and/or lacking in contemporary geomatics graduates, to forecast trends and future needs, and plan how to optimize the introduction of new technology and techniques into the educational content. Based on the mapped feedback, university-level geomatics courses are being redeveloped and updated, and novel course modules, mini-courses and micro credential programs are being developed and tested.

The WG Africa project is a collaborative initiative bringing together 12 national institutions from 8 European countries. Its objective is to support and strengthen the use of Copernicus data and services in Africa through a training-of-trainers program funded by the European Commission under the Framework Partnership Agreement on Copernicus User Uptake (FPCUP) and implemented in French, Portuguese, and English. To widely support the Copernicus products uptake, the primary goal is to collaborate with African academic and private-sector trainers by integrating Copernicus-based modules into their training programs or curricula. This initiative complements other capacity-building efforts in space-based Earth Observation in Africa, such as GMES & Africa and the Global Gateway European initiative.

Geospatial education in the UK is facing a critical decline, despite the increasing relevance of 3D reality capture and spatial technologies across sectors. While industry recognises the value of geospatial skills, the absence of coordinated national policy or incentives has led to the closure of key undergraduate programmes. Notably, Newcastle University closed its geospatial UG programme in 2023. The University of East London remains the only UK institution offering a dedicated undergraduate surveying degree, supplemented by an industry-linked apprenticeship. To address the skills gap, several universities now offer postgraduate conversion courses in geospatial science, primarily within geography or environmental science departments.

Outreach has emerged as a vital strategy to raise awareness and inspire future talent. GeospatialUK.org, developed at Newcastle University with industry support, provides accessible resources on careers, study pathways, and classroom activities aligned with UK education curricula at high school level. Its exercises—ranging from mapping hazards, wildfires, census data to GNSS-based calculations—bridge advanced research with school-level learning. It also offers insight into geospatial relate careers and links to possible job opportunities. The platform has gained international traction and continues to attract users.

This paper highlights the urgent need for national coordination in geospatial education and showcases GeospatialUK.org as a scalable model for outreach. Without intervention, the UK risks a shortage of skilled geospatial professionals, undermining its capacity to address pressing societal challenges

Recreational fishing supports economies, wellbeing, and connection to the marine environment but can pressure fish stocks. Traditional monitoring in New Zealand is costly, sporadic, and self-reported. This study evaluates integrating mobile phone data (MPD) and satellite-based object detection (YOLO on Sentinel-2 and sub-meter imagery) to improve monitoring. MPD provides temporal coverage but is biased, while satellite imagery offers spatial validation but provides only snapshots. Combining these datasets mitigates biases and gaps, enabling more accurate, representative estimates of fishing activity. This is the first study to integrate these approaches, demonstrating the potential of hybrid methods for scalable, cost-effective recreational fisheries monitoring.

The aggregation of building footprints is a key task of cartographic generalization, which is an important topic in geoinformation science. It has been approached from various angles, ranging from heuristics and optimization algorithms to machine learning. Given a set of input polygons that represent the building footprints, the task is to generate a set of polygons that provide a coarser representation of the input. The problem has applications in the visualization of settlement areas in small-scale maps, as well as settlement classification and analysis. A popular solution approach is to construct a subdivision of the plane and then build a solution by selecting faces from the subdivision. Often, a triangulation is used for the subdivision. However, this can cause the orientations of the boundary edges in the solution to differ drastically from the input polygons, which leads to a loss of information about the underlying settlement structure. We explore an alternative method that constructs the subdivision by extending the input building edges, thereby automatically preserving their orientations. To make the approach scalable to large instances without substantially decreasing the solution quality, we propose different methods of reducing the complexity of the subdivision. Our experimental evaluation on real-world data shows that our method is able to aggregate towns containing up to approximately 10 000 building footprints while preserving input edge orientations much better than state-of-the-art methods.

Timely updates of lane-level dynamic information are crucial for intelligent vehicle path planning and driving safety. Most existing crowdsourced map update methods lack sufficient analysis of the reliability and uncertainty of perception results, making it difficult to ensure the accuracy of map updates. We propose a novel method for updating lane-level dynamic information in HD maps based on crowdsourced data. First, a hybrid modelling multi-object detection method is used to reliably perceived lane markings and traffic cones. To address the issues of false detection and missed detection in single-vehicle perception, a multi-vehicle probabilistic fusion algorithm is proposed, which explicitly models perceptual uncertainty to effectively mitigate the impact of missed and false detections, enabling accurate, robust, and real-time detection of dynamic information such as temporary lane closures.To validate the effectiveness and accuracy of the proposed method, we conducted experiments in the Intelligent and Connected Vehicle Demonstration Zone in Wuhan.Experiments comparing single-vehicle and multi-vehicle fusion results demonstrate the effectiveness of the proposed method in enhancing detection performance.

This study uses machine learning and geospatial science to investigate how climate change may affect the foraging and habitat suitability of little penguins Eudyptula minor in Australia and New Zealand. An innovative modeling approach was followed here to identify favorable climatic conditions for the species across both regions. The model trained on Australian occurrence data was projected to New Zealand, and vice versa, to assess cross-regional habitat suitability and potential range shifts under changing climate conditions. This is to further evaluate adaptive potential and determine whether transoceanic relocation would be feasible in the event of local extinction. The study evaluated habitat suitability using the ML model and climate variables from the WorldClim dataset. The findings showed that the healthy habitat of little penguins is significantly shaped by temperature-related bioclimatic variables, especially temperature annual range. According to the models, the habitat suitability of little penguins varies between the two nations, with Australia offering the little penguins of New Zealand less hospitable conditions. But the New Zealand is predicted to offer relatively better habitat to Australia-based little penguins. This study offers vital information for conservation strategies by highlighting the possible changes in penguin populations brought on by climate change. A promising tool for comprehending how the climate affects marine ecosystems is provided by this study.

This paper presents an advanced computational framework developed to support decision-making for the placement of the underground Einstein Telescope, a third-generation gravitational-wave observatory. The system aims to automate the search for an optimal location through a multi-criteria analysis approach. Because the ET is extremely sensitive to environmental noise sources—including seismic, thermal, and anthropogenic vibrations—its design prioritises underground construction. This strategy, also adopted for the Japanese KAGRA detector and in contrast to surface-based observatories such as LIGO and Virgo, minimises interference from surface activities while ensuring subsurface stability.

The proposed methodology integrates Geographic Information System (GIS) data, incorporating a Digital Surface Model (DSM) to spatially represent relevant factors. The dominant site-selection criteria were identified and weighted according to their scientific and strategic importance in collaboration with the ET scientific community. An interactive parametric model was developed to interface directly with the GIS data, enabling evaluation of key factors and providing real-time analytical feedback on placement scenarios. Using an evolutionary algorithm combined with a composite fitness function, the system balances competing objectives and delivers optimised solutions, offering a robust decision-support tool for the early planning stages of the Einstein Telescope project.

Although the Sardinia site is currently considered a preliminary case study, the methodology is generalisable and applicable to other candidate sites to host ET

In fractured rock slopes, the geometric configuration and spatial arrangement of unstable rock blocks are fundamentally governed by the intersection of multiple joint sets. The mechanical weakening along these joints markedly reduces the integral strength of the rock mass and establishes potential kinematic release boundaries. This study establishes an in-situ hazardous-rock detection and characterization framework utilizing high resolution three-dimensional point cloud acquired under realistic topographic conditions. This method first examines the spatial interaction between joints and slope morphology, and incorporates explicit kinematic criteria to automatically identify structural combinations capable of different failures. Consequently, the spatial positions and distribution patterns of potentially unstable blocks are delineated within the point cloud. Subsequently, point cloud differencing is employed to achieve volumetric extraction and statistical classification of block sizes, enabling quantitative characterization of block volume and elevation across the source areas. Representative blocks are then selected as initial release elements, with their actual geometrical and volumetric attributes incorporated into rockfall simulations. This allows for the computation of key kinematic parameters including rockfall frequency, bounce height, velocity, and kinetic energy. Overall, the presented approach delivers a scalable pathway for rapid detection, quantitative assessment, and hazard evaluation of structurally controlled rockfalls in complex mountainous terrain. The results provide technical support and decision insights for the safe operation and disaster-resilient planning of transportation infrastructure.

In the application of target detection for marine oil spills, deep learning methods are gradually replacing traditional remote sensing image recognition approaches. While complex models designed for higher accuracy may compromise recognition speed, they often fail to meet the rapid response requirements of terminal device applications (Chai et al, 2025). Therefore, developing a lightweight detection model that balances high accuracy and real-time performance is crucial for enhancing marine oil spill emergency response capabilities (Liang et al, 2024). Based on the yolov8m model, this study introduces GSConv (Li et al, 2024) lightweight convolution and CCFM (Guo et al, 2025) cross-scale feature fusion module, which significantly improves the adaptability of multi-scale target detection and recognition accuracy in complex backgrounds while maintaining model lightweightness, thereby offering a novel and effective solution for marine oil spill target detection.

During the commissioning phase of ESA's Sentinel-2C, tandem images with Sentinel-2A were acquired with a delay of 30 seconds. We present a novel, automated method for labeling moving vehicles in Sentinel-2 images, leveraging the temporal offset between these tandem acquisitions. We propose a filtering process that isolates pixels corresponding to vehicles that moved between the two acquisitions. We generate a training dataset based on this process, removing the need for a large manual labeling phase. The dataset is used to train a standard deep-learning-based vehicle detection model. Experimental results, as well as a validation study using ground-truth data from California, highlight the quality of the proposed labeling method, and show that a vehicle detection model can be successfully trained from quasi-simultaneous acquisitions.

With the rapid development of aerial imaging technology, aerial target detection has become a research hotspot with broad applications in intelligent transportation, agricultural monitoring, and military surveillance. However, the performance of aerial detection models is often degraded under adverse weather conditions such as fog, sandstorms, and low illumination. In such environments, aerial images typically suffer from reduced contrast and color distortion, which significantly affects the model’s ability to accurately identify targets. To this end, a Lightweight All-in-One Aerial Detection Enhancer Under Adverse Weather (LAD-Enhancer) has been proposed. The designed enhancer processes and restores degraded aerial images, thereby enhancing the detection model’s ability to perceive potential targets. Unlike conventional image restoration models, LAD-Enhancer integrates detection labels as additional supervision during training to ensure that enhancement is detection-oriented rather than purely visual. Furthermore, a three-stage training strategy and a Mixture of Experts (MoE) framework are employed to adaptively classify and process images captured under different degradation conditions. Experimental results demonstrate that, with an increase of fewer than 3K parameters, the proposed LAD-Enhancer significantly improves detection performance under adverse weather conditions while maintaining almost unchanged performance on clear-weather images.

Currently, the state-of-the-art methods for detecting small unmanned aerial vehicles (UAVs) continue to struggle in complex urban settings due to several persistent challenges, namely, frequent target occlusion, high similarity in thermal radiation signatures between UAVs and their surroundings, and the inherently low visual saliency of small UAV targets, all of which contribute to degraded detection performance. To tackle these issues, this paper introduces a novel multi-modal feature fusion collaborative detection (MFFCD) framework grounded in learnable spatial mapping. The architecture consists of three key components: firstly, a multi-branch parallel feature extraction module (MBPFE) that simultaneously processes infrared, visible, and radar range-azimuth images, complemented by a feature fusion module (FFM) designed to enhance both intra-modal and inter-modal feature interactions; then, an adaptive spatially-aware dynamic detection head module (DDH) that dynamically recalibrates feature weights to strengthen target representation and boost detection accuracy; and a feature collaborative enhancement module (FCE) that employs a learnable affine transformation to align and fuse multi-modal features, thereby producing more robust and reliable detection outcomes. Extensive experiments show that the proposed MFFCD framework substantially outperforms existing methods under challenging urban conditions, achieving a 56.89% gain in Mean Average Precision (mAP) for small UAV detection.

Infrared and visible remote sensing image fusion is crucial for improving scene perception in complex environments, but existing autoencoder-based methods suffer from insufficient information interaction between modalities, inadequate deep feature fusion, and ineffective loss functions in extreme scenarios. To address these issues, this study proposes a Differential Feature Enhancement and Cross-modal Fusion (DFECF) method. The DFECF adopts an end-to-end architecture consisting of dual-stream encoders, cross-modal fusion modules, Transformer global perception modules, and decoders. Specifically, the Differential Enhancement (DE) module extracts differential information between infrared and visible features, combined with spatial and channel attention to enhance feature representation. The cross-modal fusion module adaptively integrates deep features based on channel attention, adjusting feature weights according to scene characteristics. The Transformer module supplements the global receptive field to capture long-range feature dependencies, and a joint loss function is designed to optimize fusion performance. Experimental results on public datasets show that the proposed method outperforms existing state-of-the-art methods in both subjective visual effects and objective evaluation metrics, especially in extreme environments such as strong light and thick smoke. It effectively improves the integrity of scene perception and provides high-quality data support for practical applications such as forest fire prevention, mining area monitoring, and autonomous driving.

This research addresses the need for accurate updates to the seabed datasets in coastal areas under environmental and human pressure. It uses Google Earth Engine (GEE) to develop a cloud-based application for Satellite-Derived Bathymetry (SDB) of the Irish bays using Sentinel-2 and Landsat-8 imagery. For the validation, the OPW Pilot Coastal Monitoring and INFOMAR datasets were used. The research refines semi-empirical algorithms and introduces an Earth Engine App (EEA) using the JavaScript API specifically tailored for and non-specialist public use. The methodology employed included pre-selecting high-quality satellite images based on the higher R-squared and lower RMSE to ensure reliability and better performance. In the initial phase, 18 bays were assessed, and the results showed that five bays (Dublin, Dungarvan, Portrane, Rosses, and Tramore) performed better across the evaluated metrics. he development and use of this application support a wide range of marine applications, especially for capacity building, as part of the pilot research led by Maynooth University and Geological Survey Ireland (GSI).

Accurate estimation of methane fluxes across Arctic wetlands is essential for understanding carbon–climate feedbacks, yet remains difficult due to sparse ground measurements, strong spatial heterogeneity, and the coarse resolution of most existing bottom-up inventories. To address these limitations, we develop a high-resolution methane flux modeling framework that integrates multisensor Earth observation data with a geofoundational deep-learning approach. The study leverages 30 m Harmonized Landsat–Sentinel (HLS) imagery, together with environmental predictors from SMAP and ERA5, and daily eddy-covariance methane flux measurements from Arctic sites after 2015. Following data filtering and quality control, the dataset comprises more than 7,600 daily observations from 45 wetland sites across northern high latitudes. A hybrid model architecture is constructed by combining the Prithvi geospatial foundation model for HLS feature extraction with a lightweight feature-wise attention encoder processing 48 auxiliary environmental variables. Fused latent representations are used to predict daily methane flux at 30 m resolution. The model demonstrates strong performance on an independent test set, capturing key spatial and temporal patterns of methane emissions. By enabling fine-scale flux estimation far beyond the resolution of conventional 0.1°–0.5° inventories, the framework offers new opportunities for detailed Arctic methane monitoring and improved characterization of wetland-driven emissions.

To plan nature restoration of fluvial corridors on a national level an inventory of existing man-made levees is mandatory. We suggest an automatic method for a river-wise extraction of levees from a high resolution terrain model based on profiles perpendicular to the river axis. In this course we present a method to cover corridors with non overlapping profiles with a given maximum distance. Levee detection is based on a mathematical formulation of the protective function of levees. In an evaluation of 150 km river length distributed over nine different rivers in Austria the method detected 98% of manually extracted levees, and 68% of their length.

Temperatures in urban areas are rising due to the climate change. Together with increasing urbanization and densification reducing cooling green spaces in cities this leads to so called urban heat islands (UHI) with increased surface- and air-temperatures in urban areas relatively to the surrounding areas. Since high temperatures are the reason for many exceed deaths municipalities are forced to protect their citizens. Satellite earth observation allows to monitor the development of urban heat islands to warn inhabitants early from dangerous heat. An other important way is increasing the resilience of cities to heat waves. For this we developed a simple but efficient method for the simulation of urban surface- and air-temperatures from single very high resolution stereo satellite images. In this paper we present the improved workflow for the simulation of urban temperatures together with the calibration and validation. Further we compare the results to in-situ-measurements in the city of Memmingen in southern Germany, to LandSat thermal mapper imagery and existing works on urban heat islands. Additionally we show how modifying the digital twin e.g. by adding trees or water areas allow the simulation of different scenarios to support decision-makers on their path towards resilient cities.

Scandinavia is facing climate change, with mean temperatures projected to rise by 2-4°C. To prepare Swedish forests for this challenge, the Swedish tree breeding program aims to selects trees adapted to a range of biotic and abiotic conditions. Key variables in this selection process include spring phenology, damage, and overall tree vitality.

Traditionally, these data have been collected through manual field assessments, a resource-intensive approach that constrains both the number of trees that can be evaluated and the frequency of measurements. Remote sensing offers an alternative: high-resolution multispectral drone imaging enables the scoring of greater numbers of trees in less time, captures multiple data points across the growing season, and reduces the risk of human error through algorithmic measurement.

This project aims to develop methods suitable for integration into the Swedish tree breeding program by using multispectral drone imagery to assess spring phenology, shoot damage, and vitality in young Norway Spruce. Field campaigns were conducted during spring 2023 and 2024. Bud flush is modeled from spectral values of tree crowns, using manual assessments of a subset of trees as training data. To capture the full progression of bud flush at high temporal resolution, images were acquired before the vegetation season and up to twice weekly during the period of most rapid development. The same modeling framework is applied to assess damage and vitality.

Antarctic ice shelves regulate ice sheet mass balance through their "buttressing effect", with major implications for global sea level rise. This study focuses on the Nansen Ice Shelf in Victoria Land, East Antarctica, which exhibits complex topography and sensitivity to environmental changes. Previous research has primarily centered on its significant collapse event in 2016; however, systematic evolutionary patterns over longer timescales remain unclear. This study integrates multi-source remote sensing observations from 1948 to 2025 to systematically reconstruct changes in the Nansen Ice Shelf's geometric characteristics (crevasse width, area) and dynamic parameters (ice flow velocity). Findings reveal distinct activity differences between the northern and southern regions of the ice shelf, closely linked to their respective boundary conditions and structural features.

Flood mapping is one of the most important applications of Synthetic Aperture Radar (SAR) because it can monitor the earth's surface under all-weather, day-and-night conditions. While SAR intensity has been widely used for flood mapping, the potential and added value of interferometric coherence, especially its temporal behavior in different environments, remains mostly unexplored. In this study, we assess the potential and added value of interferometric coherence from Sentinel-1 time series for flood mapping in three contrasting regions: the urban area of Valencia (Spain), the arid region of Sistan and Baluchestan (Iran), and the agricultural area of Hannover (Germany). Our analysis of multi-temporal coherence shows that coherence provides clear flood indicators in arid regions through strong temporal decorrelation, but its performance is less reliable in vegetated and urban areas. In agricultural regions, pre-flood (baseline) coherence is inherently low due to vegetation phenology and temporal decorrelation, making any additional decrease due to flood inundation often indistinguishable. In urban areas, coherence generally remains stable, with only slight decreases observed in specific cases; therefore, the detectability of flooded areas using coherence-based approaches is limited in both agricultural and urban environments. In contrast, coherence in arid regions is high before flooding and drops significantly during flood events, making floods easy to detect in such regions. These findings demonstrate that, for flood mapping, interferometric coherence is a valuable but environment-dependent indicator, with the highest benefit seen in arid regions where intensity-based methods are limited.

Arctic sea ice thickness has been declining over recent decades due to climate change, making accurate prediction increasingly critical for environmental monitoring and climate modeling. Microwave remote sensing combined with machine learning has emerged as a promising approach for estimating sea ice thickness. This study investigates the prediction of lab-grown sea ice thickness, ranging from 27 to 47 cm, using time-series backscatter data collected from surface-based Ku- and C-band scatterometers in three polarizations (VV, HH, and HV). A Random Forest model was applied to the time series, incorporating Normalized Radar Cross-Section (NRCS) values and statistical features (mean and standard deviation) across various temporal variables (lead and lag times). The model achieved high prediction accuracy, with the lowest error recorded at RMSE = 0.03 cm. Feature importance analysis using the Permutation Importance method revealed that co-polarized C-band features (C-VV and C-HH) were the most influential in predicting sea ice thickness. These findings underscore the potential of integrating microwave remote sensing with Random Forest models to enhance sea ice thickness prediction and provide valuable insights for future research and real-time monitoring in Arctic regions.

Operational flood monitoring demands both accurate extent delineation and quantitative depth estimation, yet existing research addresses these objectives separately. This study presents an integrated SAR-to-depth framework combining state space model segmentation with DEM-based geometric depth estimation to deliver comprehensive flood intelligence from Sentinel-1 SAR imagery and digital elevation models. We propose CS-Mamba, a hierarchical U-Net architecture incorporating selective state space

mechanisms, achieving 79.79% mean IoU on 10 European flood events from the KuroSiwo benchmark while surpassing CNN baselines and outperforming RSMamba by 7.37 percentage points. Test performance exceeding validation confirms robust cross-event generalization to unseen disasters. Controlled experiments establish that deep learning predictions achieve sufficient accuracy for operational depth estimation, with CS-Mamba flood masks showing ±2% agreement with reference annotations across four global DEMs despite conservative extent delineation. This agreement enables integrated pipelines without manual annotation, while systematic DEM comparison identifies Copernicus and MERIT as optimal choices. The complete framework delivers three-class flood masks and pixel-wise depth maps at operational resolution, bridging the traditional gap between extent mapping and quantitative assessment for emergency response.

This contribution introduces TGSD-Net, a temporal variation-guided self-supervised network designed to improve despeckling of polarimetric SAR (PolSAR) imagery without the need for clean reference data. The method leverages consecutive multi-temporal observations to create pseudo training pairs and incorporates a lightweight temporal change detection prior, allowing the network to exploit temporal redundancy while remaining robust to land-cover variations. TGSD-Net further integrates auxiliary polarimetric decomposition features and a spatiotemporal information fusion module to enhance structural and scattering representations. The approach is tailored for multi-temporal SAR scenarios, where speckle, temporal variation, and heterogeneous land-cover types pose significant challenges. Experiments on real PolSAR datasets show that TGSD-Net achieves strong noise suppression while preserving edges, textures, and physical scattering properties. The results demonstrate the potential of self-supervised temporal learning to advance PolSAR image restoration and support downstream remote sensing applications.

Current Remote Sensing applications demand multi-source, multi-sensor data fusion. Multi-source, multi-sensor data fusion provides useful information integrated for quick and better interpretation, understanding and effective decision-making. Data fusion of Synthetic Aperture Radar (SAR) data of Earth Observation Satellite-04 (EOS-04) and Optical Multispectral (MX) data of Sentinel-2 are current topic of interest in this paper. SAR and Optical MX which includes active and passive remote sensing technologies belong to different mechanisms of wave interaction due to widely separated and non-overlapping regions of the electromagnetic spectrum. In this paper, a novel approach to the re-implementation of Wavelet, Brovey, Fast Intensity Hue Saturation (FIHS), Frequency filtering, and Pure pixel data fusion methods is presented. The presented novel approach emphasises modulation-based fusion technique with proper normalization and scaling of both the input datasets. Fusion results of presented fusion methods are evaluated visually as well as quantitatively with quality metrics. The quality metrics demonstrate the ability of the presented novel approach to fuse optical spectral information into SAR data effectively to generate improved high-resolution SAR-coloured fused products.

Natural disasters demand rapid damage assessment to guide humanitarian response. Here, we investigate whether medium-resolution Earth observation images from the Copernicus program can support building damage assessment, complementing very-high resolution imagery with often limited availability. We introduce xBD-S12, a dataset of 10,315 pre- and post-disaster image pairs from both Sentinel-1 and Sentinel-2, spatially and temporally aligned with the established xBD benchmark. In a series of experiments, we demonstrate that building damage can be detected and mapped rather well in many disaster scenarios, despite the moderate 10m ground sampling distance. We also find that, for damage mapping at that resolution, architectural sophistication does not seem to bring much advantage: more complex model architectures tend to struggle with generalization to unseen disasters, and geospatial foundation models bring little practical benefit. Our results suggest that Copernicus images are a viable data source for rapid, wide-area damage assessment and could play an important role alongside VHR imagery. We release the xBD-S12 dataset, code, and trained models to support further research.

In recent years, technological advancements, particularly in artificial intelligence (AI), are changing various fields and spurring new research. This study focuses on the use of AI in cartography and historical studies. It is part of the PRIN project "Crafted in Stone / Recorded on Paper," which aims to document the heritage of small Italian municipalities by creating an open-access database. The research discovered significant documents in Gandino, Italy, including a large-scale map and a 139-page textual register from the mid-eighteenth century. These documents come from land surveyors who measured municipal boundaries and properties using physical landscape markers. The original surveying method, although lost, shares similarities with modern land descriptions.

The study seeks to generate new maps from these textual registers using AI capabilities, aiming to replicate a historical mapping effort from the 1700s. Initial tests with an AI model involved reading the register, computing measurements, and creating coordinate tables. The results showed promise despite some inaccuracies. The goal is to develop an interdisciplinary method that graphically reconstructs information from written documents, enhancing access for historical and territorial analysis. The research will also explore further AI models and larger case studies to achieve this aim.

The analysis of buildings from aerial imagery is a fundamental task for urban planning and disaster response, yet it traditionally requires a suite of specialized models for tasks like segmentation, detection, and semantic querying. The advent of generalist Vision-Language Models (VLMs) offers a new paradigm, but their adaptation to the specific, high-resolution remote sensing domain remains a significant challenge. This paper proposes and investigates a novel methodology for adapting a general-purpose VLM,

Google’s PALIGEMMA2, to function as a unified geospatial building analyzer. The core of this contribution is a data-centric pipeline that converts single-modality annotations (building polygons) into a rich, multi-task instruction-tuning dataset (16,500 samples) spanning segmentation, detection, Visual Question Answering (VQA), and captioning. A rigorous study is conducted to answer three critical questions: (1) Can a single instruction-tuned VLM outperform specialized models in a multi-task setting? (2) What are the synergistic benefits of multi-task learning? (3) How data-efficient is this adaptation process? The results demonstrate that the unified model significantly outperforms the zero-shot PaliGemma2 baseline and strong single-task fine-tuned variants on three out of four tasks, while remaining competitive on the fourth. A strong synergistic effect is found: multi-task training on both visual localization and semantic tasks improves performance on individual localization tasks. Furthermore, the analysis shows that high performance can be achieved with a surprisingly small instruction dataset. This work provides a complete methodology for efficiently adapting VLMs to multi-task geospatial analysis, suggesting a new path towards generalist models in remote sensing.

Foundation models have achieved remarkable success across various domains due to their ability to learn generalizable representations from large-scale, unlabeled datasets. In the geospatial domain, several foundation models have been developed to leverage the abundance of unlabeled remote sensing data and support Earth observation tasks across diverse regions and sensor types. However, the geolocation-dependent characteristics of remote sensing data introduce unique challenges in adapting these models to region-focused applications. By conducting a comprehensive empirical analysis across diverse geographical regions and tasks, we explore whether incorporating regional information during pretraining or fine-tuning improves performance on region-specific downstream tasks. We show that regional representation learning, as well as regional adaptation of features extracted from a globally trained foundation model, is beneficial when the region-specific performance of the downstream tasks is of interest. To this end, we also propose a regional adaptation to the globally trained foundation models to balance global diversity with regional representation learning for improved performance.

Label noise in the sense of incorrect labels is present in many real-world data sets and is known to severely limit the generalizability of deep learning models. In the field of remote sensing, however, automated treatment of label noise in data sets has received little attention to date. In particular, there is a lack of systematic analysis of the performance of data-centric methods that not only cope with label noise but also explicitly identify and isolate noisy labels. In this paper, we examine three such methods and evaluate their behavior under different label noise assumptions. To do this, we inject different types of label noise with noise levels ranging from 10 to 70% into two benchmark data sets, followed by an analysis of how well the selected methods filter the label noise and how this affects task performances. With our analyses, we clearly prove the value of data-centric methods for both parts – label noise identification and task performance improvements. Our analyses provide insights into which method is the best choice depending on the setting and objective. Finally, we show in which areas there is still a need for research in the transfer of data-centric label noise methods to remote sensing data. As such, our work is a step forward in bridging the methodological establishment of data-centric label noise methods and their usage in practical settings in the remote sensing domain.

Extracting road networks from high-resolution imagery remains a significant challenge in geomatics, particularly in fragmented rural landscapes. The big difficulty is the spectral similarities between unpaved tracks and agricultural backgrounds that can lead to classification errors. This study proposes an automated geospatial pipeline based on the YOLOv11 architecture. Specifically, the approach is made on the optimization of the multi-class road detection in the rural areas of Kosina and Markowa, two villages in Poland. To reduce the computational effort, due to large-scale 9000x9000 px orthophotos and to improve the detection of small-scale features, Slicing Aided Hyper Inference (SAHI) strategy was integrated. High-resolution imagery has been decomposed into optimized tiles, ensuring feature continuity across boundaries and preventing GPU memory overhead. The instance segmentation model was trained on a custom-annotated dataset, with seven labels (categories) such as internal paved roads, rural tracks, and railway infrastructures. Therefore, a high level of robustness has been achieved reaching a mean Average Precision value (mAP@0.5) of 0.90. A confusion matrix reveals quantitatively that the pipeline effectively distinguishes between complex classes and low omission rates. As a result, the generated outputs are converted into interoperable GeoJSON format ensuring their integration into GIS environments. In conclusion, the experimental result demonstrates that the framework is valuable for emergency response logistics and urban planning. It offers a scalable and near real-time solution for updating national topographic databases.

Outbreaks of the European spruce bark beetle (Ips typographus) have intensified across Central and Northern Europe due to droughts, storms, and other extreme climatic events. Resulting Norway spruce mortality has reduced growing stock and weakened forest carbon uptake, creating an urgent need for rapid, operational tools for early detection. Pre-emergence detection, i.e. identifying infested trees before brood emergence, is particularly valuable, yet field surveys remain too slow and costly at large scales. UAV-based optical remote sensing offers high-resolution, flexible, and timely mapping at the single-tree level, allowing detailed observation of spectral changes soon after attack.

Despite many recent UAV studies, the reliability and transferability of pre-emergence detection remain unclear. Differences in sensor types (multispectral vs. hyperspectral), band configurations—especially in the red-edge and green-shoulder regions—and analytical approaches have produced inconsistent results. Many models are developed within single sites and often lack standardized accuracy metrics or cross-site validation, limiting insights into robustness under varying ecological and climatic conditions.

To address this, we compiled six UAV datasets from four major outbreak regions—southern Sweden, southern Finland, the southeast Alps in Italy, and Czechia—covering multispectral and hyperspectral campaigns at the single-tree level. Using these harmonized data, we compare machine-learning models for classifying tree health based on spectral features and vegetation indices. A central focus is transferability. We test models across regions using cross-regional, joint, and leave-one-region-out schemes to quantify generalization across contrasting climates, outbreak phases, and stand structures. The results reveal consistently informative spectral regions and modelling strategies, offering practical guidance for operational early-warning systems.

Natural and anthropogenic disturbances are impacting the health of forests worldwide. Monitoring forest disturbances at scale is important to inform conservation efforts. Here, we present a scalable approach for country-wide mapping of forest greenness anomalies at the 10 m resolution of Sentinel-2. Using relevant ecological and topographical context and an established representation of the vegetation cycle, we learn a predictive quantile model of the normalised differential vegetation index (NDVI) derived from Sentinel-2 data. The resulting expected seasonal cycles are used to detect NDVI anomalies across Switzerland between April 2017 and August 2025. Goodness-of-fit evaluations show that the conditional model explains 65% of the observed variations in the median seasonal cycle. The model benefits most from the local context information during the green-up period. The approach produces coherent spatial anomaly patterns and enables country-wide quantification of forest browning. Case studies with independent reference data from known events illustrate that the model reliably detects different types of disturbances.

The exponential spread of the bark beetle (Ips typographus L.) outbreaks across Europe in recent years has led to heightened interest in remote sensing-based detection. This increase is closely linked with ongoing climate change, which has led to rising temperatures, prolonged dry periods, and increasing frequency and intensity of both biotic and abiotic disturbances. These conditions created a favourable environment for bark beetle proliferation, resulting in larger and more widespread infestations. Effective detection and management of these outbreaks is crucial for forest officals, necessitating the implementation of monitoring systems that complement traditional ground-based efforts. At present, remote sensing approaches for bark beetle detection mainly rely on optical data to identify changes in spectral reflectance of vegetation. In this study, we utilised annual Sentinel-1 synthetic aperture radar (SAR) composites from 2021 to 2023 for infestation detection. A Random Forest classification model was applied to distinguish between healthy and infested forest areas. Additionally, vegetation indices were incorporated to evaluate and compare the results. A reference dataset was used to validate model performance. Our results show that the Sentinel-1 based approach achieved lower accuracies (max. overall accuracy: 0.78), compared to Sentinel-2 (max. overall accuracy: 0.87). Despite this, the Sentinel-1 data proved valuable as a tool for bark beetle infestations detection, especially in scenarios where optical data may be unavailable or limited. These results underscore the importance of integrating SAR data into remote sensing workflows to improve the detection of bark beetle outbreaks.

Forest ecosystems are increasingly threatened by biotic and abiotic disturbances that are intensifying under a changing climate. Accurate detection of tree stress is essential for effective forest management, as stress strongly increases vulnerability to damaging agents such as pests, pathogens, and fire. Tree water functioning is a key indicator of physiological status, yet traditional field-based methods for monitoring water transport – such as sap flow measurements – require costly instrumentation and can only be applied to a limited number of trees.

Hyperspectral remote sensing offers a powerful means to upscale forest health monitoring, but its effectiveness depends on robust spectral indicators that reliably reflect physiological change. Green-Shoulder Indices (GSI), which leverage reflectance features in the 490–560 nm region linked to carotenoid dynamics, have been previously used to monitor tree health. Because carotenoids are closely tied to photosynthetic regulation, stress responses, and canopy vitality, GSI have emerged as promising indicators of health decline. Notably, they have shown strong performance in detecting Norway spruce trees in the early stages of bark beetle infestation.

This study investigates how GSI can be further strengthened as indicators of forest hydraulic functioning by integrating hyperspectral drone imagery with continuous sap flow monitoring. By linking canopy spectral responses to internal water transport dynamics, we aim to advance GSI as operational tools for large-scale forest health surveillance and disturbance detection.

Carotenoids regulate photoprotection and respond early to stress, but their retrieval from canopy reflectance is often unstable because green-band signals are confounded by canopy structure, illumination/view geometry, and covariance with chlorophyll. This study proposes and evaluates the sensitivity of green-shoulder indices (derived from 490–550 nm bands) to carotenoid content in vegetation. We use FluSAIL simulations to generate canopy reflectance under wide-ranging biochemical and structural conditions and benchmark multiple green-region indices (490–560 nm, including PRI-type formulations) for their sensitivity and stability to carotenoids. We then transfer the best-performing index–carotenoid relationship to independent real-world datasets with pigment measurements at both leaf and canopy scales (ANGERS, LOTUS, CABO) to test generalization beyond the simulation domain. Results showed that a curvature-based green-shoulder index provided the most consistent carotenoid sensitivity, with the strongest and most stable VI–Car relationships across varying chlorophyll–carotenoid coupling, LAI, and sun–sensor conditions. Validation on measured spectra confirms that green-shoulder indices can predict carotenoid content with high accuracy and improved transferability compared with conventional green indices.

Non-stand replacing (NSR) disturbances—low- to moderate-severity events causing single-tree mortality or canopy thinning—are driven by agents such as drought, insects, pathogens, low-intensity fire, wind, and snow. Their variable duration, frequency, and extent challenge detection using medium-resolution optical imagery because changes are spectrally subtle and spatially complex. We developed a framework to detect NSR disturbances in boreal forests on a sub-annual basis using Sentinel-2 (S2) time series. Key methods include the spectral normalisation of monthly cloud-free composites via weighted multidimensional medians (medoid and geometric median), as well as improvements to the sensitivity and robustness of the HILANDYN algorithm. Observation weights are based on spectral distance measures (Euclidean distance and Spectral Angle Mapper), normalised using an adaptive sigmoid function. Normalisation reduced seasonality patterns by 41.4%, leaving only 13.7% of the tested time series with a significant seasonal pattern. Validated on more than 10,000 points, the best F1 and F2 scores were 0.75 and 0.72, respectively, when using seven S2 variables. These metrics increased to 0.80 and 0.81, respectively, when including detections in the subsequent vegetative season. The geometric median outperformed the medoid, and the optimal spectral indices varied by agent, e.g., NBR for canopy removal, red-edge indices for wind and snow damage. While the framework effectively maps natural and anthropogenic NSR events, reducing detection lag at high latitudes remains a priority.

3D Gaussian Splatting (3DGS) has revolutionized real-time rendering with state-of-the-art novel view synthesis, but its applicability to accurate geometric measurement remains limited. Compared with multi-view stereo (MVS)-based point clouds or mesh models, 3DGS provides superior visual quality and completeness, while existing measurement approaches still rely on stereoscopic workstations or direct measurements on incomplete and inaccurate reconstructed geometry.

As a novel view synthesizer, 3DGS reproduces source views and smoothly interpolates intermediate viewpoints, enabling users to intuitively identify congruent points across multiple views. By triangulating these correspondences, accurate 3D point measurements can be obtained. Inspired by traditional stereoscopic measurement, the proposed approach removes the need for stereo workstations and biological stereoscopic capability, while naturally supporting multi-view measurements for improved accuracy.

We implement a web-based application to demonstrate this proof of concept using UAV-based aerial datasets. Experimental results show that the proposed method achieves measurement accuracy comparable to or better than traditional stereoscopic measurement approaches while operating entirely on non-stereo workstations. In particular, the proposed method consistently outperforms direct mesh-based measurements, achieving RMSEs of 1–2 cm on well-defined points. On challenging thin structures, the proposed method reduces RMSE from 0.062 m to 0.037 m, and successfully measures sharp corners where mesh-based methods fail entirely.

The source code and documentation are open-source and available at: https://github.com/GDAOSU/3dgs_measurement_tool.

Traditional 3D Gaussian Splatting (3DGS) methods initialize Gaussian primitives from Structure-from-Motion point clouds, resulting in loosely distributed representations that lack geometric constraints and metric accuracy. This limitation severely restricts their application in architectural heritage preservation, where millimeter-level precision and practical editability are essential requirements.

This paper introduces Gaussian Texturing, a novel framework that fundamentally transforms how Gaussians relate to geometry by directly binding 3D Gaussian primitives to precisely measured mesh surfaces—essentially "texturing" surfaces with Gaussians. Our approach comprises three key innovations: (1) a constrained optimization framework that maintains tight Gaussian-surface coupling throughout training, preventing geometric drift while preserving photorealistic rendering quality; (2) engineering-oriented editing tools enabling geometry-based material replacement, region editing, and mesh-driven deformation; and (3) seamless integration with professional heritage preservation workflows.

Experimental validation on MipNeRF360 benchmarks and custom architectural datasets demonstrates that our method achieves millimeter-level geometric precision while maintaining competitive rendering metrics. Unlike traditional "bind-after-training" approaches, our direct surface binding paradigm eliminates intermediate reconstruction steps, ensuring accuracy from source data. Real-world applications in heritage documentation and architectural design confirm the method's practical value, successfully bridging the gap between photorealistic visualization and engineering-grade geometric accuracy for professional applications.

In this study, we develop a Structured framework for Gaussian Splatting (3DGS) with LiDAR integration (Structured-Li-GS). It is a lightweight Gaussian Splatting pipeline that leverages LiDAR–inertial–visual SLAM. Structured-Li-GS achieves high-quality 3D reconstructions with fewer Gaussians by training on accurate, dense, colorized point clouds. Gaussian primitives are anchored using sub-sampled point clouds, and their ellipsoidal parameters are initialized from local surface geometry. Our training strategy integrates a comprehensive set of loss terms, including photometric, flattening, offset, depth, and normal losses—guided by the dense point cloud, enabling accurate reconstruction without Gaussian densification. This approach produces up-to-scale, high-fidelity results with a moderate model size. For experimental validation, we develop a custom hardware-synchronized LiDAR–camera handheld scanner. Experiments on both benchmark datasets and our real-world in-house dataset demonstrate that Structured-Li-GS surpasses state-of-the-art methods while using fewer Gaussians.

True Digital Orthophoto Maps (TDOMs) are essential for urban analysis and map updating, traditionally generated through photogrammetric workflows involving aerial triangulation, DSM construction, and orthorectification. Recently, 3D Gaussian Splatting (3DGS) has emerged as an alternative approach using differentiable volumetric rendering. While both methods depend on acquisition geometry, they follow fundamentally different reconstruction processes, potentially producing distinct representational characteristics. Systematic comparisons under controlled conditions remain limited.

This study generates photogrammetric and 3DGS-based TDOMs from four UAV datasets acquired over the same area with varying resolution (2.51–5.8 cm GSD), image count, and oblique view proportion (0–75%). All datasets were preprocessed through common SfM to obtain identical inputs. We evaluate differences through inter-method agreement (PSNR, SSIM, LPIPS), detail preservation (gradient magnitude, high-frequency energy), and spatial distribution patterns (boundary–interior separation).

Results show 3DGS systematically smooths fine-scale texture with gradient ratios of 0.58–0.89 and high-frequency energy reductions of 2.5–55× relative to photogrammetry. Oblique view proportion emerges as the dominant divergence factor: oblique-dominant datasets show lowest agreement (PSNR 15.15) despite larger image counts, while nadir-only datasets achieve higher similarity (PSNR 26.73). Difference maps reveal 2–3 times higher discrepancies along boundaries than interiors. Visually cleaner 3DGS boundaries are byproducts of overall smoothing rather than superior reconstruction. These findings establish that the two methods are complementary—photogrammetry preserving texture fidelity and 3DGS providing structural regularity—with acquisition geometry critically influencing performance characteristics.

Efficient and robust 3D scene representation is crucial in fields such as robotics, autonomous driving, and augmented reality. While RGBimagesprovidevaluable content for 3D reconstruction, other modalities like thermal or depth can enable additional information on the 3D environment. Lately, Novel View Synthesis (NVS) methods like Gaussian Splatting (GS) have started using multiple modalities to further boost their performance. But fusing or combining those multi-modal data can make the process slower and bring in additional challenges. Therefore, our project aims to use single modality based on thermal infrared domain, by removing the reliance on visible light, as much as possible. We propose a method Thermal-to-Depth Gaussian (TDg), that uses only thermal images and depth estimation in its architecture to derive the radiance fields. Mainstream methods relying heavily on RGB images, perform poorly in visually degraded environments, such as low-light conditions, fog, smoke, or extreme weather. Contrary to this, infrared cameras can detect objects’ inherent thermal radiation and provide a robust perception, suitable regardless of lighting and weather conditions. But despite their promise, thermal images are inherently characterized by low contrast, sparse texture, and non-uniform brightness distribution. So current approaches still rely heavily on paired RGB images for supervision or joint optimization, failing to establish a truly independent and purely thermal-based Gaussian representation system. Therefore, the core innovation of our work is to prepare a self contained Thermal GS framework that uses only thermal image inputs. We design a thermal-guided depth estimation module, Thermal-to-Depth (TDg), providing explicit and reliable constraints for geometric optimization.

With the rapid advancement of satellite, aerial, and UAV platforms, the daily volume of remote sensing data collected over megacities has grown exponentially. However, only a limited portion of this data can be transformed into usable products in time. Current production workflows remain lengthy and poorly automated, which fails to meet the increasing demand for high-precision and high-timeliness remote sensing products in city management, environmental monitoring, and emergency response.

To address this gap, this study proposes the construction of an standardized, efficient and reusable foundational database system consisting of three key components: image control point database, interpretation sample database, and spectral information database.

The image control point database establishes a unified geometric reference for multi-source data; The interpretation sample database provides large-scale, high-quality labeled data for deep learning-based image analysis; and the spectral database offers standardized spectral features for accurate classification and parameter inversion.

Together, the three databases form a collaborative mechanism that links geometric accuracy, semantic understanding, and spectral consistency, thereby building a complete chain from analysis-ready data (ARD) production to rapid information extraction. Using Shanghai as a case study, this paper presents the design, construction, and collaborate applications of the three databases, demonstrating their effectiveness in supporting refined and sustainable megacity governance.

Disaster Risk Management benefits from innovative techniques including AI and Multi Sensor Fusion. The Fireguard Approach uses such technologies to improve the Wildfire Management works in Saxony, Eastern Germany by supporting standing efforts in Early Warning, Disaster Response and Monitoring. Unmanned Aerial Systems (UAS) play a vital role in providing real-time information via a 5G network to a central information management system that delivers geospatial information to response teams. This study highlights the potential of combining UAS, AI, geospatial solutions and existing data for real-time wildfire monitoring and risk assessment systems. The preliminary study successfully shows the potential of the provided solution to enhance Wildfire early detection, response and monitoring to address immediate and long-term needs of response teams.

This paper presents a multi-agent geospatial application that enables users to interact with spatial data through natural language, called MapEcho Copilot. The system integrates large language model (LLM) reasoning, semantic embedding search, and spatial analytics within a unified architecture. A vector embedding database is constructed to index diverse open-source geospatial datasets. Upon receiving a user query such as “show all tennis courts in downtown of Toronto” or “find habitats of grizzly bears in Canada” the system performs semantic retrieval to identify relevant datasets, followed by geospatial filtering and reasoning through specialized agents via an interactive and friendly interface. The multi-agent framework coordinates between semantic understanding, data retrieval, and spatial computation layers to deliver map-based responses in real time. The Results demonstrate the system’s ability to process both semantic and geospatial queries with high accuracy and interpretability, providing an intuitive bridge between natural language and spatial intelligence.

Point clouds have proven over the years to be a suitable spatial representation of scenes and objects at varying scales and levels of complexity, making them widely used across several scientific domains and applications. Advancements in sensor technology, computer vision, and data science have produced high‑resolution point clouds and advanced analytical approaches, leading to broader adoption for spatial information extraction to support decision making. However, traditional point cloud management systems for organizing and distributing data throughout the point cloud lifecycle often create significant duplication at each stage. This causes data fragmentation as multiple copies and versions are scattered across different processing steps, workgroups, and storage locations, further limiting cross‑domain applications.

In this paper, we propose a unified point cloud data management (PCDM) approach that supports the principles of findability, accessibility, interoperability, and reusability (FAIR) across domains at scale. The proposed approach aims to support diverse point cloud retrieval for cross-domain analysis by leveraging a single, reusable PCDM system built on a shared data model. Our approach improves on existing frameworks and provides a foundation for point cloud data management and data spaces.

This contribution presents a rigorous mathematical and computational examination of confidence ellipses applied to bivariate spatial distributions, with a specific focus on their implementation in the open-source LFTools plugin for QGIS. Confidence ellipses are widely used in geography, environmental sciences, public health, criminology, and spatial statistics to summarize central tendency, dispersion, and directional trends of point-based datasets. Although conceptually well established, their practical reliability depends on correct numerical implementation and statistical consistency—an aspect rarely evaluated in detail.

The study first revisits the formal mathematical foundations of confidence ellipses, including covariance-matrix geometry, eigen-decomposition, and Chi-Square-based scaling for different confidence levels. It then analyses the computational workflow adopted in LFTools and validates its correctness using 100,000 simulated Gaussian random points, demonstrating near-perfect adherence (<0.05% deviation) to theoretical confidence intervals.

To assess performance on real-world data, the method is applied to the Groundwater Well Database for Brazil (GWDBrazil), comprising more than 350,000 groundwater wells. Confidence ellipses at the national and regional levels reveal strong anisotropy, clustered patterns, and non-Gaussian spatial structures, confirming both the robustness of the tool and the complexity inherent to real geospatial phenomena.

Results indicate that the LFTools implementation is mathematically sound, statistically reliable, and suitable for scientific applications. The study highlights the relevance of reproducible open-source tools and outlines future directions involving spatial–temporal extensions, non-parametric approaches, and multi-scale territorial analysis, applicable in Brazil and worldwide.

Through theoretical analysis and empirical research, this study thoroughly examines the theoretical foundations and practical applications of zero-defect sampling inspection schemes, revealing significant differences between inspecting large lots as a whole versus splitting them into sub-lots in terms of consumer's risk control. The findings indicate that although the zero-defect sampling scheme (Ac=0) adopted in the GB/T 24356-2023 standard shifts quality control from "post-production spot checks" toward "in-process prevention", it exhibits notable deficiencies in controlling consumer's risk, resulting in an unacceptably high level of risk for consumers. Empirical analysis demonstrates that, for large lots with relatively poor quality, e.g., when the product's defect rate is 10%, the inspection plan (100, 10, 0) still carries a 33.3% probability of erroneously accepting the lot, which significantly exceeds the risk level typically acceptable to consumers and thus imposes excessive quality risk on them. Furthermore, the study reveals that inspecting small lots or subdividing large lots benefits producers, highlighting an imbalance in the current standard's risk allocation mechanism. These insights provide more reliable theoretical support and practical guidance for quality management of surveying and mapping products.