Geomatics

Coastal waters are vulnerable ecosystems that are increasingly affected by sediment and nutrient inputs, as well as pollution resulting from both natural processes and anthropogenic pressures. Monitoring the pattern of chlorophyll-a and turbidity is essential for understanding the dynamics of coastal ecosystems and supporting environmental management. This study investigates the spatial and seasonal patterns of water quality in the coastal zone of Heraklion, Crete (Greece), for the period 2016–2024 using OCEANIDS Data Cube products (GA 101112919). Water quality variability was assessed using satellite-derived spectral indices, including the Normalized Difference Chlorophyll Index (NDCI) and the Normalized Difference Turbidity Index (NDTI). Monthly observations were analyzed using a GIS-based workflow to assess seasonal spatiotemporal variability. The results revealed strong seasonal variability, with higher winter NDCI values in all coastal zones and persistent hotspots concentrated near river estuaries and waters influenced by port activities. Meanwhile, the analysis was used to prioritize the region for further monitoring and to support decision-making for stakeholders. In summary, this study demonstrates the potential of OCEANIDS Data Cube products (NDCI, NDTI) for long-term monitoring of surface coastal water quality in a Mediterranean environment with limited data and supports evidence-based coastal management. Full article

Planning future electric vehicle charging infrastructure requires assessment of spatial suitability, demand, network gaps, and expected accessibility benefits. This study developed a national-scale framework for Bosnia and Herzegovina integrating machine-learning (ML) suitability modelling, post-modelling prioritisation, reproducible candidate selection, and scenario-based population accessibility assessment. A dataset of 188 EVCS locations and spatially balanced pseudo-absences was analysed using 28 spatial predictors. Five classifiers were evaluated through spatial cross-validation, with XGBoost providing the most balanced performance. After feature reduction, the final model retained five predictors: road density, travel time to hotels, travel time to parking, distance to major roads, and travel time to tourist attractions. The priority index combined modelled suitability with population demand, LU/LC opportunity, and the travel-time gap to existing EVCSs. Candidate locations were derived using a deterministic settlement- and road-constrained procedure followed by network-based spacing, and scenarios with 10, 20, 40, 50, and 60 new EVCSs were evaluated. At the 10 min threshold, population coverage increased from 57.9% for the existing network to 65.4% with 10 new EVCSs, 68.7% with 20, 73.8% with 40, 75.6% with 50, and 77.0% with 60. The framework provides a reproducible basis for national EVCS investment screening while distinguishing occurrence-based suitability, strategic deployment priority, and expected accessibility gains. Full article

Ground deformations induced by underground mining significantly affect the geometric condition and serviceability of buildings located in mining areas. Conventional assessments are commonly based on ground deformation indicators, which do not necessarily reflect the actual structural response. This study presents a terrestrial laser scanning (TLS)-based methodology for assessing the three-dimensional deformation of an eleven-storey residential building located in the Legnica–Głogów Copper District (LGCD), Poland. The analysis was performed using a high-density point cloud acquired from ten scanning positions. Following registration and filtering, building geometry was reconstructed and corner positions were determined from 123 horizontal cross-sections. Horizontal displacements, tilt profiles, and rotation about the vertical axis were subsequently analysed within a local coordinate system. The results revealed pronounced spatial variability in both displacement magnitude and direction. The maximum horizontal displacement reached approximately 0.18 m, corresponding to a local tilt of 5.9 mm/m. Corner displacements at the highest common observation level ranged from 8.6 mm to 178.3 mm, indicating that the observed geometry is inconsistent with a simple rigid-body model subjected to uniform tilting. Analysis of geometric changes with height further identified an overall increase in torsional rotation with height, accompanied by local variations. Comparison of TLS-derived geometry with a theoretical mining-induced ground deformation model showed that the measured structural response does not directly reproduce the underlying ground deformation pattern. The largest discrepancies occurred along the building longitudinal axis, indicating that structural stiffness and soil–foundation–structure interaction significantly modify the transfer of ground movements to the superstructure. These results demonstrate the capability of TLS for detailed assessment of mining-affected buildings and provide quantitative insight into the relationship between ground deformation and actual structural response. Full article

Surface mining significantly alters land cover and vegetation, making long-term monitoring essential for assessing its environmental impacts. The objective of this study was to analyze the long-term development of the active Dargov quarry (Slovakia) during the period 2009–2025 using Landsat image time series. Changes in vegetation cover and exposed surfaces were assessed using the NDVI, BSI, and NDBI spectral indices in combination with the Mann–Kendall test, Sen’s slope estimator, and the LandTrendr algorithm. The results revealed a gradual decline in NDVI values accompanied by a concurrent increase in BSI and NDBI values within the active quarry, reflecting the expansion of exposed rock and soil surfaces associated with ongoing mining. Statistically significant trends (p < 0.05) were identified in 76.36% of NDVI pixels, 65.45% of BSI pixels, and 63.64% of NDBI pixels within the deposit boundary. The relationships between spectral indices and quarry production were also evaluated using Pearson’s correlation analysis. The spectral indices showed a substantially stronger relationship with cumulative quarry production than with annual production. The LandTrendr algorithm identified the most pronounced vegetation-cover changes primarily during 2019–2024, with 70.91% of the pixels in the Time of Largest Change output within the deposit boundary falling within this period. The proposed methodology provided a comprehensive assessment of the spatial and temporal patterns of changes associated with surface mining at the Dargov quarry and demonstrated its potential to support environmental monitoring, environmental impact assessment (EIA), and mineral resource management. Full article

Super-resolution (SR) of remote-sensing imagery is commonly assessed through spatial fidelity or visual sharpness, although index-based geomatics requires preservation of cross-band spectral relationships. We compare five Sentinel-2 full-band SR configurations (LDSR-S2, SRGAN, SPAN, Mamba, and SWIN) in two hazard-mapping cases: flood-water detection during the 2024 Valencia flood and burn-scar mapping after the 2025 Palisades wildfire. Each model refines four native RGB–NIR bands, while SEN2SR reconstructs the remaining bands to produce a ten-band product at $2.5$ m. We evaluate native-grid reconstruction, the introduction of high-frequency details, and downstream thematic, boundary, and edge-region metrics against a bilinear-interpolation baseline. Dynamic-threshold MNDWI and dNBR detectors are applied independently to each output. Among the learned configurations, SWIN achieves the strongest native-grid reconstruction and task-specific spectral consistency and the strongest fire agreement, but adds the least high-frequency content. Flood full-ROI gains are modest, with LDSR-S2 increasing the F1-score from 0.085 for bilinear interpolation to 0.091. All learned configurations increase flood-edge recall, F1-score, and IoU while reducing edge precision and balanced accuracy. LDSR-S2 gives the strongest final edge F1-score and IoU in both tasks, Mamba gives the lowest learned-model symmetric flood-boundary distance. SPAN yields the best spatial consistency and lowest learned flood-edge spectral error and SRGAN adds the most high-frequency content and the largest combined edge-region gain, alongside the greatest task-specific spectral deviation and the largest symmetric boundary-distance increases. Thus, increased edge activation does not establish uniformly improved delineation or recovered sub-pixel detail. These cases demonstrate feasibility rather than generalization. Operational validation requires more diverse, time-synchronous, high-resolution, spectrally compatible references. Full article

Peri-urban forest ecosystems play a crucial role in sustaining biodiversity, regulating climate, and providing essential ecosystem services; however, they are increasingly threatened by rapid urban expansion. Despite advances in remote sensing-based forest monitoring, long-term reconstruction of continuous forest canopy cover dynamics remains challenging in highly fragmented peri-urban landscapes. This study developed a machine-learning workflow to reconstruct forest canopy cover dynamics within the “Suelo de Conservación” of Mexico City between 1994 and 2024 using Landsat imagery and forest canopy cover information derived from the Hansen Global Forest Change dataset. A balanced training dataset comprising 5000 samples distributed across five forest canopy cover classes was used to compare four regression algorithms (Multiple Linear Regression, Random Forest, Gradient Boosting, and Multilayer Perceptron) under five-fold spatial cross-validation. Model performance was evaluated using Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), bias, Pearson’s correlation coefficient (r), coefficient of determination (R²), and Lin’s Concordance Correlation Coefficient (CCC). The best-performing model was applied to generate forest canopy cover maps for 1994, 2003, 2014, and 2024, and forest-cover change was quantified using propagated uncertainty and threshold sensitivity analysis. The reconstructed forest canopy cover maps revealed an initial decline between 1994 and 2003, followed by partial recovery during 2003–2014 and relatively stable forest canopy cover conditions through 2024. Independent comparison with the National Forest and Soils Inventory (INFyS) and Global Forest Watch forest canopy cover products indicated moderate agreement in the spatial distribution of canopy cover while highlighting uncertainties associated with differences in reference datasets and acquisition periods. The proposed workflow provides a transparent and reproducible framework for long-term forest canopy cover reconstruction using freely available satellite imagery and supports forest monitoring and conservation planning in peri-urban landscapes. Full article

In this study, we used the DeepESDL framework to exploit multivariate data cubes for the identification and characterization of frontal zones in the Alboran Sea throughout the year 2023. By operating directly within a unified data hypercube architecture, we seamlessly evaluate the spatial coherence and temporal evolution across highly disparate environmental variables. We adapted the Gradient Histogram Method (GRADHIST) algorithm to run efficiently on data cube tracks of sea surface temperature (SST), sea surface height (SSH), sea surface salinity (SSS), and chlorophyll-a concentration (CHL) fields. This multi-dimensional approach enabled a detailed, synchronized description of the spatial distribution and evolution of fronts. The GRADHIST algorithm proved highly adaptable to the data cube format, utilizing a dynamic threshold that optimizes front detection across fields of fundamentally different physical natures and dynamic ranges. Full article

Smartphone GNSS metadata remains an underexplored source for evaluating navigation-signal integrity in real-world conditions. We investigated GPS behavior recorded by a Samsung Galaxy A72 smartphone across two European flights (EXP03: Rome–Bucharest, n = 377; EXP10: Bucharest–Lisbon, n = 78) and 275 terrestrial reference photographs (Cabo da Roca, Portugal). A total of 730 photographs were analyzed using a seven-indicator taxonomy, conceptually inspired by Receiver Autonomous Integrity Monitoring (RAIM) principles, covering anti-spoofing, anti-sniffing, and anti-tampering checks. GPS capture rates reached 100% (Timestamp Camera) and 83.3% (native camera) up to 11,439 m WGS84, among the highest EXIF altitude profiles reported to date. Velocity spikes (1560–1875 km/h), one at cruise altitude and one during landing, were indistinguishable from GPS spoofing at the EXIF level, and their physical origin is undetermined. Phantom geolocation was absent in flight (0/442) versus 37.3% on the ground, consistent with GPS constellation visibility as the primary factor. In total, 88% (n = 920) of JPEG files lacked the standard EOI marker, generating false positives in integrity validators. Findings are device-specific, derived from non-independent observations, and require replication before generalizing to other GNSS receivers or latitudes. The framework offers a methodological basis for EXIF integrity analysis relevant to EU AI Act Article 10(3) data quality and UAV precision-agriculture georeferencing. Full article

Shoreline determination is fundamental to coastal research, spatial planning, and legal boundary delineation but remains challenging in low-relief coastal areas where small sea-level variations can produce substantial horizontal shoreline displacements. This study compares shoreline determination methods based on tide gauge (TG) observations, LiDAR data, Sentinel-1 synthetic aperture radar (SAR) and Sentinel-2 optical satellite imagery using the low-relief coast of Pärnu Bay, Estonia, as a case study. The comparison was based on shorelines derived from Sentinel-1 and Sentinel-2 imagery acquired on selected common acquisition dates within the 2015–2025 study period, rather than on a temporally continuous annual dataset, and compared with temporally matched LiDAR-derived shorelines extracted from a Digital Terrain Model (DTM) generated from a 2021 LiDAR survey. The LiDAR-derived shorelines were extracted using the mean sea level (MSL) observed at the Pärnu and Häädemeeste TGs at the satellite overpass time, while the satellite-derived shorelines were additionally validated against RTK GNSS measurements. The results demonstrate that the evaluated methods produce substantially different shoreline positions. Sentinel-2-derived shorelines generally corresponded more closely to the temporally matched LiDAR-derived shorelines than Sentinel-1-derived shorelines and most accurately represented the instantaneous land–water boundary during field validation. These findings demonstrate that different shoreline determination methods represent different shoreline definitions. Consequently, shoreline datasets should be interpreted according to their intended purpose rather than treated as directly interchangeable. Full article

Small- and medium-sized landslides frequently occur in clusters and exhibit fragmented morphologies, irregular boundaries, and spectral characteristics similar to surrounding roads, bare soil, and sparsely vegetated surfaces, making their automated extraction from remote sensing imagery challenging. Although the Segment Anything Model (SAM) provides strong general-purpose segmentation capabilities, its direct application to landslide mapping is limited by the geoscience domain gap and its dependence on external prompts. This study proposes the Asymmetric Boundary-aware Segment Anything Model (AB-SAM), a parameter-efficient adaptation of SAM for automated landslide semantic segmentation. AB-SAM integrates three task-specific components. First, the offline Multi-Feature Variation-Guided Prompting (MF-VGP) module generates cached auxiliary bounding boxes from registered pre- and post-event images without accessing ground-truth masks. Second, the Asymmetric Feature Augmentation (AFA) strategy combines geometric perturbation, CutMix, and asymmetric dual-branch supervision, in which a Hint-free branch serves as the primary optimization pathway and a lower-weight box-guided branch provides auxiliary spatial supervision. Third, the Boundary-Aware Morphological Prompting (BAMP) module injects trainable boundary-aware morphological information into the largely frozen SAM image encoder. During validation, testing, and application, only the Hint-free branch is retained, enabling inference using post-event imagery without external point, box, or mask prompts. On the fixed, spatially disjoint Zixing test set, AB-SAM achieved an overall accuracy of 96.171%, a precision of 68.149%, a recall of 60.011%, an F1-score of 63.822%, a landslide-class Intersection over Union of 46.867%, and a mean Intersection over Union of 71.452%. Repeated experiments with three random seeds showed low run-to-run variation. Direct evaluation without retraining on the Hokkaido Iburi-Tobu dataset yielded a mean Intersection over Union of 66.136%, providing evidence of cross-region and cross-event transferability. These results demonstrate that AB-SAM provides a practical parameter-efficient framework for automated, hint-free landslide segmentation, although further evaluation across additional regions, sensors, and landslide-size distributions remains necessary. Full article

Accurate identification of the spatial distribution of agricultural greenhouses is essential for regional agricultural management in Japan. However, an efficient method for systematically mapping greenhouse locations onto existing farmland maps has not yet been established, and field surveys over large areas require substantial time and labor. This study aims to develop an automated approach for detecting agricultural greenhouses and integrating their locations into farmland maps by combining PlanetScope satellite imagery with farmland polygon data developed by the Japanese government. To improve the efficiency of the extraction process, farmland polygons were used to restrict the analysis to known agricultural areas, thereby reducing false detections originating from non-agricultural land. Within these predefined regions, three machine learning algorithms—Random Forest (RF), Support Vector Machine (SVM), and Isolation Forest (ISF)—were applied to classify and extract greenhouse features from satellite imagery. After optimizing the hyperparameters of all models, RF and SVM achieved an equivalent peak performance, with an F1-score of 0.86, while ISF reached 0.72. RF was, however, markedly more robust to the polygon-level decision threshold, demonstrating a practical advantage in situations where the threshold cannot be optimized in advance. In addition, an ablation experiment confirmed that without pre-masking with farmland polygons, 81.9% of the pixels predicted as greenhouse were distributed outside the agricultural parcels. The proposed method is expected to serve as an effective approach for efficiently identifying the distribution of agricultural facilities and integrating them with existing farmland information in regions characterized by small and fragmented agricultural fields, such as Japan. Full article

Remote-controlled UAS-based soil sampling offers great potential for scientific research and practical applications in various fields, including agriculture, environmental monitoring, hazardous waste, radiation measurements, chemical plants, snow sampling and glaciology. The added value of automated sampling using a UAS (unmanned aircraft system) compared to manual sampling lies primarily in improved accessibility to hazardous or remote locations, increased operational safety, and the potential to improve efficiency in applications requiring repeated or difficult-to-access sampling. This article introduces two different 3D-printed constructions (grab arm and screw pipe) for UAS-based, remote-controlled sample collection of different soil types. The grab arm construction is based on an adapted version of an open-source CAD from GrabCAD. The screw pipe construction is a new design. During an expedition to Greenland in August 2025, the two constructions were manually evaluated during several days of field work near the Danish and Austrian research stations at the Sermilik Fjord to assess their mechanical sampling performance on challenging Arctic surface materials, including dry and wet sand, gravel, glacial sediment, snow, and glacier surfaces. Because flight testing was not possible during the expedition due to unavailable UAS batteries, these experiments were limited to manual ground evaluation of the constructions. Independent flight tests were subsequently conducted in Austria using a DJI Matrice 300 to evaluate the integration of both the grab arm and the screw pipe with the UAS platform and their operational handling during flight. In Greenland the performance of manual sampling was documented precisely. Further the coordinates of each sampling point were recorded by using GPS, sample images were taken on site, and the collected material was weighed. The results show that UAS in combination with 3D printed constructions is a flexible and location-independent solution for obtaining soil samples of varying composition. In addition to the design, materials, and electronics of the systems, the article also describes the connection to the UAS and the field work results. Full article

Remote sensing is widely recognized as a key technology across a wide range of technical and scientific domains, especially in agriculture. Although satellite data have long supported crop monitoring, their limitations in spatial resolution, revisit frequency and cloud coverage have often constrained their applications. High-resolution satellites, available from the beginning of the 2000s, have improved performance, particularly in the field of precision agriculture, but they remain expensive and inflexible. Unmanned Aerial Vehicles perform better in precision agriculture, offering flexibility and high levels of detail; however, their limited operational areas and short endurance flight times constrain their effectiveness. In this evolving landscape, High Altitude Pseudo Satellites (HAPSs), particularly high-altitude balloons, are emerging as a promising new technology that could fill the gaps between satellite and drone remote sensing. These platforms provide large area coverage with high-resolution imagery and long endurance flights at low operational expenses and ease of deployment. This study investigates the operational characteristics, strengths, and geometric limitations of data acquired by the CubeHAPS^® platform, a high-altitude pseudo-satellite system, as a prerequisite for its application in precision agriculture. Focusing on experimental campaigns conducted in northern Italy in summer 2024 and 2025, the research characterizes platform stability, image block consistency, and photogrammetric quality through internal metrics. The results demonstrate measurable improvements between the two campaigns, attributed to the introduction of a stabilization system in 2025 and establishing the conditions under which the platform can support reliable photogrammetric reconstruction. Full article

Cultural heritage sites are affected by environmental and anthropogenic pressures that require decision-analysis methods capable of combining heterogeneous datasets while accounting for uncertainty. Multi-Criteria Decision Analysis (MCDA), particularly when integrated with Geographic Information Systems (GIS) and Earth Observation (EO) data, is widely used in geospatial analysis involving multiple, often conflicting criteria. This review examines the evolution, application domains, and methodological challenges of MCDA in cultural heritage risk assessment. The literature indicates a predominant reliance on weighting-based methods, especially the Analytic Hierarchy Process (AHP) combined with GIS-based weighted overlay techniques, while uncertainty treatment, temporal monitoring, validation, and multi-threat applications remain limited. Three illustrative applications show that asset-level, regional susceptibility, and historic-urban frameworks address complementary decision needs but differ in their data, expertise, and institutional requirements. Recent developments show a trend to combine MCDA with fuzzy logic, machine learning, and uncertainty modeling, although methodological consistency across these approaches remains uneven. The findings suggest that multi-criteria risk assessment for cultural heritage may depend less on introducing new analytical techniques and more on improving the integration of existing methods. Incorporating repeatable environmental observations, sensitivity analyses, multi-threat assessment, and stakeholder participation may support a more coherent and reproducible approach to heritage-risk assessment. Full article

Effective control and management of the encroaching and invasive Vachellia karroo (V. karroo) in grassland and savanna biomes depends on accurate information about its spatial distribution, making remote sensing approaches essential for mapping the extent of affected areas. Although Sentinel-2 satellite data are widely used for land use and land cover applications, they often lack the spatial details required to distinguish woody species like V. karroo. The fusion of Sentinel-2 data with high-resolution UAV imagery offers a promising approach to enhance spectral information for species-level discrimination. This study evaluated UAV, Sentinel-2, and fused UAV–Sentinel-2 imagery for discrimination of V. karroo in grassland and savanna biomes of the Eastern Cape, South Africa. Field data and imagery were collected in October 2022 and classified using Random Forest (RF) and Support Vector Machine (SVM) algorithms to distinguish V. karroo. The findings showed that V. karroo was more prevalent in the savanna biome. SVM marginally outperformed RF in classifying V. karroo in the grassland biome, achieving overall accuracies ranging from 68.9% to 97.4%, compared to 57.8% to 97.4% for RF. Among the datasets, the fused UAV–Sentinel-2 images yielded the highest classification accuracy, with an overall accuracy of 97.4% and a kappa coefficient of 0.96. The UAV images also demonstrated high classification accuracy, with an overall accuracy of 91.67% and a kappa coefficient of 0.77, confirming its value for fine-scale mapping and reference data support. In contrast, the Sentinel-2 images produced lower classification accuracy, with an overall accuracy of 84.6% and a kappa coefficient of 0.75, mainly due to their coarser spatial resolution. Classification was more challenging in the savanna site, where mixed vegetation structure increased confusion between V. karroo and grass. These findings show that fused UAV–Sentinel-2 images can improve species-level discrimination, while UAV and Sentinel-2 data remain complementary for fine-scale mapping and broader monitoring of bush encroachment in grassland and savanna ecosystems. Full article

Digital platforms for cultural heritage are increasingly expected to document the geometric and material properties of sites. Furthermore, such platforms are increasingly expected to preserve and communicate the intangible practices and community knowledge that give these places cultural significance. In this context, this paper presents a hybrid virtual reality platform for the documentation, communication, and dissemination of both tangible and intangible heritage in underground wine landscapes. The framework integrates 360° panoramic imagery, 360° videos, lightweight object visualisations, and georeferenced 3D point clouds within a unified interface adapted to different device capabilities. The system’s key contributions include seasonal navigation, participatory recording of community practices, multiscale representation of artefacts and architecture, and a guided narrative system that improves orientation for non-expert users. The platform has been validated through the underground wineries of Baltanás (Spain), thereby demonstrating a novel approach that brings together metric documentation and immersive storytelling. This enhances accessibility, facilitates heritage interpretation, and ensures the digital preservation of living cultural practices in complex heritage settings with broad public dissemination potential. Full article

Parcel area sits at the intersection of urban planning, land administration and land surveying. It underpins development intensity, floor area allocation, minimum lot thresholds, land readjustment and value capture mechanisms. Yet discrepancies between modern measurements and ownership titles are usually evaluated through fixed tolerance formulas rather than quantified confidence intervals. While coordinate precision is routinely specified, the uncertainty of the derived parcel area is seldom expressed explicitly, limiting the traceability of planning calculations based on cadastral geometry. This paper presents a variance-based formulation for estimating parcel area uncertainty from boundary coordinates. Using the Gauss area function and first-order propagation, vertex precision is translated into parcel-level confidence intervals based on horizontal RMS parameters commonly reported in cadastral practice, including documented transformation accuracy. The Greek cadastre provides an illustrative case combining a national GNSS infrastructure, a unified reference system and formula-based area screening embedded in statutory workflows. Illustrative examples show how area uncertainty varies with parcel geometry and measurement origin. Absolute uncertainty increases with parcel size and boundary elongation, while relative uncertainty decreases with parcel size. A Monte Carlo analysis of the error-correlation structure shows that the diagonal, independent model is not a universal bound: depending on the structure of the transformation error and on parcel geometry it may either overstate or understate the true area uncertainty, by factors between about 0.4 and 3.5 in the cases examined. The results clarify how coordinate precision propagates into regulatory-relevant area values and support more transparent interpretation of area discrepancies in planning and land administration contexts. Full article

Global Navigation Satellite Systems (GNSS) have emerged as a reliable and cost-effective tool for estimating atmospheric precipitable water vapor (PWV), particularly in regions with limited meteorological instrumentation. In this study, the performance of three tropospheric mapping functions—the Global Mapping Function (GMF), Niell Mapping Function (NMF), and Vienna Mapping Function 1 (VMF1)—was evaluated for PWV estimation using GPS observations collected during the 2009–2011 period in southwestern Mexico, a region characterized by high atmospheric variability and frequent extreme weather events. GPS data from three stations (TECO, COL2, and PENA) were processed using the GAMIT/GLOBK 10.71 software, and the resulting PWV estimates were validated against independent radiosonde observations and the European Centre for Medium-Range Weather Forecasts (ECMWF) Fifth-Generation Reanalysis (ERA5) data. The results show that GPS-derived PWV successfully captures the seasonal variability of atmospheric water vapor, with maximum values during the summer rainy season. High correlations were obtained with both radiosonde and ERA5 data, particularly at the TECO station (R = 0.95–0.99), where RMSE values ranged from 3.27 to 5.46 mm and BIAS values from $-2.73$ to $-1.47$ mm. In contrast, larger discrepancies were observed at COL2 and PENA, mainly due to horizontal separation and altitude differences relative to the radiosonde site, highlighting the importance of spatial representativeness during validation. Among the evaluated mapping functions, no single model consistently outperformed the others across all stations, years, and reference datasets. Nevertheless, GMF and NMF generally exhibited more stable and consistent performance, whereas VMF1 showed greater variability under the adopted processing strategy. Additionally, a clear relationship was identified between PWV and precipitation records, indicating that increases in PWV coincided with periods of intense rainfall and suggesting its potential as an indicator of atmospheric conditions favorable for precipitation events. Overall, this study shows that GPS-derived PWV can reproduce the seasonal variability of atmospheric water vapor under the adopted processing strategy and demonstrates the importance of mapping function selection and spatial representativeness for accurate PWV estimation. Full article

InfraRed Thermography (IRT), a non-invasive, non-contact, and non-destructive testing (NDT) technique, has become an established tool in the assessment of a building’s behavior and energy performance. However, the inherent low spatial resolution of thermal infrared (TIR) cameras has led recent work to fuse thermographic and geometric data to generate accurate 3D representations of buildings encapsulating temperature information. Whilst existing data fusion methods have relied on sensors in fixed relative orientation (RO), the co-registration of independent TIR and RGB blocks using ground control points (GCPs), or the reprojection of TIR images onto additional geometric or parametric models, approaches that directly match multi-modal images remain limited. In principle, if multi-modal tie points were available, it would be possible to directly align the RGB block with the TIR block; however, such matching is extremely challenging due to the substantial differences in radiometric properties. The main contribution of this paper is to demonstrate the applicability of off-the-shelf deep learning-based image matching algorithms, originally trained on mono-modal datasets, to multi-modal matching tasks for InfraRed Thermography 3D-Data Fusion (IRT-3DDF). We conduct a comparative evaluation of the principal algorithms developed in recent years, with particular emphasis on 3D accuracy and computational efficiency, under the hypothesis that, owing to the inherently local nature of the problem they address, these algorithms can generalize from a mono-modal training domain to a multi-modal application domain. The results are benchmarked against existing hand-crafted open-source multi-modal reference methods. Importantly, the proposed method is fully-automatic, obviating the need for sensor pre-calibration, manual co-registration, or associated positioning information. Results demonstrate that DL-based image matching, using pre-trained neural networks outside of their expected training domain, provides a viable approach for IRT-3DDF capable of co-registering blocks of multi-modal images across varying scales, settings, sensors, and subjects. Our results indicate accuracy in 3D is up to seven times better than multi-modal hand-crafted algorithms, while hand-crafted mono-modal methods fail to co-register images in their entirety. Full article