The problem arises whenever analysts run KDE over large datasets and then clip the results to regions of interest such as municipalities or protected areas. In that conventional workflow, the search radius, known as the bandwidth, is estimated from the global spatial extent of all input features. A bandwidth tuned to an entire nation may be far too broad for a single city district, smearing density peaks into a blur that no longer reflects local reality. Doing it correctly, by computing KDE separately for each polygon, has always been possible in principle but painfully impractical at scale, requiring analysts to manually clip features, estimate bandwidths, run the analysis, export rasters, and mosaic the results for every single region.

That is the gap the Iterative Density Analysis plugin, described in the journal SoftwareX by Miguel Inácio, Luís Valença-Pinto, and Paulo Pereira, sets out to close. Written in Python using the QGIS Python API, or PyQGIS, the plugin treats each polygon in a boundary layer as an independent analytical unit. For every polygon it extracts the geometry, clips the input features to that boundary, estimates or applies a search radius, computes the KDE raster, and clips the output to the polygon edge. Intermediate maps are written to a user-defined directory and then merged into a seamless mosaic, all without a single line of user scripting.

A key design decision was to build the plugin entirely on native QGIS processing algorithms for clipping, rasterisation, and mosaicking. This avoids external dependencies such as SciPy, which can cause version incompatibilities, and means the plugin inherits bug fixes and improvements directly from QGIS itself. The developers tested it on QGIS 3.34 LTR and 3.40 under Windows 11 and macOS, and because only native algorithms are used, it should run on any platform supported by QGIS 3.x, including Linux. The plugin is distributed under the GNU General Public License v3.0 through GitHub and the official QGIS Plugin Repository.

At the heart of the automation lies Silverman’s rule of thumb, a classic statistical formula that estimates an optimal bandwidth from sample size and data spread. ArcGIS Pro applies this rule automatically, but across the global extent of the data, which is precisely the flaw the new tool corrects. The plugin instead applies the bivariate form of the rule, h = 0.9 multiplied by the estimated standard deviation and by n raised to the power of minus one fifth, to the clipped features within each polygon independently. The spread is measured either by the standard deviation or the interquartile range, whichever is smaller after scaling. When a region contains fewer than two features, the plugin falls back to a bandwidth of three times the output cell size, a small but reasonable radius that scales with resolution rather than leaving the calculation undefined.

Line features get special treatment. Because the standard KDE workflow is built around points, the plugin converts lines, such as road networks or hiking trails, into regularly spaced points before density estimation, sampling along each line at one fifth of the cell size so that point spacing stays finer than the output grid. For very dense networks the spacing coarsens automatically to keep runtime and memory manageable. This ensures that line-density outputs are directly comparable with point-density outputs, a consistency that dedicated line-density algorithms do not always guarantee.

The graphical interface walks users through nine intuitive steps, from selecting the feature type and boundary layer to defining cell size, area units, search radius, and output paths. A progress window tracks each polygon’s calculation, and a detailed log is exported at the end, recording the polygon identifier, the number of features, the Silverman bandwidth, and summary statistics such as the median and mean across all regions. This log is not just bookkeeping: it enables a second, comparative mode of analysis, in which the user takes the mean or median Silverman value across all polygons and re-runs the analysis with that value as a fixed, data-driven search radius applied uniformly everywhere.

The illustrative example in the paper is striking. The authors processed 103 administrative polygons across Lithuania, containing 38,058 tourism infrastructure points from OpenStreetMap, at a 10 metre cell size. The complete plugin workflow finished in 80 seconds on a MacBook Pro with an Apple M2 Max chip. The conventional global KDE-then-clip workflow, tested in ArcGIS Pro on the same hardware, took 182 seconds, while a fully manual per-polygon workflow, extrapolated from the largest municipality, would have consumed roughly 51.5 minutes of user time. More importantly, the maps themselves diverge sharply: within Vilnius municipality, the locally estimated workflow reached peak densities of around 190 points per square kilometre, while the globally derived, clipped output peaked at only about 60, despite both using the identical set of input points.

Why does the difference matter so much? A bandwidth estimated from national data smooths density over vast distances, flattening the sharp local peaks that matter to spatial planners and environmental managers. The locally calibrated surfaces produced by the plugin are also better suited as inputs to downstream methods such as Geographically Weighted Regression, where a globally smoothed density clipped to regions would distort local values and, in turn, regression coefficients. Beyond regression, the authors point to applications in ecosystem services mapping, risk and exposure assessment, hotspot analysis, and suitability overlays, wherever the spatial intensity of points or lines must be characterised region by region.

The plugin is not without limitations, which the authors state candidly. It processes polygons sequentially rather than in parallel, which can inflate runtime and disk usage for hundreds of regions at fine resolution. Silverman’s rule becomes less reliable in polygons with very few features, though the log flags these cases. The line-to-point conversion does not exactly reproduce dedicated line-density algorithms, and clipping features to each polygon means that features just outside a boundary, which would contribute density mass under a standard KDE, are excluded, potentially introducing discontinuities at polygon edges. Only the quartic kernel is currently supported. The developers say future work will explore alternative kernels and bandwidth estimators, Linux validation, and native line-density estimation. Even so, the core contribution stands: a free, transparent, one-click workflow that computes density the statistically correct way, region by region, at a scale that was previously impractical.

Subject of Research: Automated per-polygon kernel density estimation with Silverman's rule in a new open-source QGIS plugin

Article Title: Automated kernel density estimation across multiple polygons: the iterative density analysis QGIS plugin

Article References: Inácio, M., Valença-Pinto, L., & Pereira, P. (2026). Automated kernel density estimation across multiple polygons: the iterative density analysis QGIS plugin. SoftwareX, 36, Article 103145. https://doi.org/10.1016/j.softx.2026.103145

Image Credits: AI Generated

DOI: 10.1016/j.softx.2026.103145

Keywords: kernel density estimation, QGIS, GIS, Silverman's rule, bandwidth selection, spatial analysis, open-source software, PyQGIS, mosaicking, tourism infrastructure, Vilnius, geospatial mapping

Most existing digital tools for hydropower planning stop at site selection. They can highlight a stretch of river with strong flow and a steep drop, and some can estimate the theoretical power available, but they rarely tell the engineer where the intake should be placed, which bank the headrace canal should follow, where the transition to a pressurized penstock should occur, or how the penstock should weave downhill around obstacles. The new algorithm, built as a Python plugin for the open-source QGIS platform, does all of this automatically. For every possible intake along a studied stream, it searches downstream outlet and powerhouse positions, traces candidate headrace routes along both riverbanks, evaluates alternative transition points, routes the penstock in three dimensions across the terrain, avoids restricted zones, and ranks every complete configuration it generates.

The heart of the method is a clever solution to a deceptively hard raster problem: tracing the contour line of a stream point when the digital elevation model contains no cell at exactly that elevation. Rather than attempting a pixel-by-pixel contour walk, the algorithm first divides the entire basin into groups of cells according to which stream cell lies just below them in elevation. This grouping is done in a single pass, which keeps computation fast even for large watersheds. The borders between these elevation groups then serve as approximate contour paths. Because a true contour must cross every line connecting a border cell to a lower neighboring cell, the traced route is guaranteed to stay within one cell width of the real contour — a precision entirely adequate for screening-level planning.

Once the left- and right-bank contours are traced for each intake, the algorithm measures the headrace length along them and pairs each intake with every eligible downstream outlet. Configurations are ranked using a novel metric called the Cost-to-Power Ranking Index, or CPRI, which divides the combined conveyance-cost index of the headrace and penstock by the hydraulic-power proxy formed from nominal discharge and screening head. The screening head accounts for the gradual loss of elevation along the open canal, and the power estimate assumes typical efficiencies of 96 percent for electrical equipment and 90 percent for the turbine. Because the same physical constants apply to every candidate, the index acts purely as a relative comparator, deliberately avoiding any impression that its values represent real monetary costs or final deliverable power.

Early screening uses quick Euclidean estimates of penstock length, but the top candidates then receive a much more rigorous treatment. The algorithm builds a graph whose nodes are raster cells and whose edges are permitted connections between neighbors, with restricted areas removed entirely. Dijkstra’s shortest-path algorithm then finds the minimum-weight penstock route for each shortlisted configuration, accounting for true three-dimensional distance and, where available, spatially varying construction costs. A built-in diagnostic checks whether the shortlist was large enough to certify the final ranking, warning users when more candidates should be retained for rerouting.

The researchers tested the tool on two very different Greek watersheds. The first was the small experimental catchment of the Lykorrema stream on Mount Pendeli in Attica, covering just 7.84 square kilometers of sharp relief between 280 and 950 meters elevation, analyzed at an unusually fine 5-meter resolution. The second was the much larger Inachos sub-watershed of the Sperchios River basin in central Greece, spanning 341.86 square kilometers from 85 to 2,286 meters, analyzed using the 25-meter Copernicus EU-DEM. Runtimes were strikingly practical: the full Pendeli search took 57 seconds on a desktop computer, while even the long branch of the Inachos reach completed in 19 minutes. A modest laptop performed nearly as well, completing the Pendeli analysis in 64 seconds.

Resolution testing revealed how grid size shapes the results. When the top 100 solutions from the 5-meter Pendeli analysis were compared with those from a 10-meter run, 51 matched within 10 meters across all three key points, and 79 fell within 10 meters of the top 500 coarser solutions. The best-ranked layouts agreed within 75 meters. The differences stem partly from the finer grid resolving nearby alternatives that coarser cells merge into one, and partly from terrain detail that only exists at high resolution — a quarter as many cells simply encode a quarter as much information about subtle ridges and gullies.

The tool also proved robust in engineering judgment. Fifty candidate configurations, 30 from Inachos and 20 from Pendeli and chosen for broad spatial spread, were assessed by two specialists with experience on more than 15 small hydropower projects in Greece. They evaluated the plausibility of the intake-canal-forebay-penstock-powerhouse layout, the gentle descending routing of the open headrace, the reduction of penstock length and cost, and the effective use of available head. The verdict: the layouts were technically plausible for preliminary screening and consistent with expected practice. The authors are careful to stress that this validates screening-level reasoning, not detailed hydraulic design.

Several design decisions in the algorithm reveal the messy realities of working with raster terrain. When small hills between the stream and a contour produce isolated rings of candidate cells that a path cannot unambiguously distinguish, the algorithm hugs the downstream direction of the stream, keeping the eventual penstock shorter and cheaper. A subtler puzzle involved rounding: because pixelated elevations rarely match an intake exactly, the intuitive choice is to round downward, giving the water a natural slope. But on steep hillslopes this can make contour cells coincide with downstream stream cells, causing the tracer to cross the river. The team therefore rounds upward instead — choosing the lowest cell higher than the intake — an equally close approximation that structurally prevents the traced contour from ever touching or crossing the stream.

The practical implications reach well beyond Greece. Engineering consultants, developers and public authorities can feed in a digital elevation model, discharge estimates from a flow-duration curve after deducting environmental flow, relative construction costs and exclusion maps for protected areas, and receive a ranked spreadsheet of grouped alternatives complete with coordinates, canal and penstock lengths, cost indices and CPRI scores. Cost rasters can encode rock quality, landslide risk or legal constraints, and a restricted-area raster forbids any crossing of protected cells. The authors position the tool as a bridge between broad regional hydropower mapping and the costly site-specific feasibility studies that follow, and they caution that penstock friction losses, geotechnical conditions, environmental permitting and full economics remain outside its scope. The source code is openly available on GitHub for QGIS 3.34 or later, inviting researchers worldwide to test it in new catchments, refine its screening heuristics and accelerate the hunt for the small, clean hydropower schemes the energy transition will increasingly depend on.

Subject of Research: GIS-based automated preliminary planning of small run-of-river hydropower plants

Article Title: Development of a new simplified algorithm facilitating GIS-based preliminary planning of small hydropower plants

Article References: Soulis, X. K., Soulis, K. X., Tsekouras, G. J., Palli-Gravani, S. N., & Sarantopoulou, V.-E. K. (2026). Development of a new simplified algorithm facilitating GIS-based preliminary planning of small hydropower plants. Results in Engineering, 32, Article 113478. https://doi.org/10.1016/j.rineng.2026.113478

Image Credits: AI Generated

DOI: 10.1016/j.rineng.2026.113478

Keywords: small hydropower, GIS, run-of-river, QGIS, contour tracing, penstock routing, Dijkstra algorithm, digital elevation model, renewable energy, CPRI ranking, hydropower siting, Greece

The research team, led by Jovito P. Permante of Batangas State University – The National Engineering University, together with Erwin Rafael Cabral and Florina Padilla-Santiago, applied what is known as sieve mapping, a weighted overlay approach executed entirely in the open-source desktop GIS package QGIS. Their decision premise was deliberately pragmatic: villages that simultaneously record high absolute COVID-19 case counts and substantial mapped natural-hazard exposure may warrant earlier integrated preparedness attention than villages that score highly on only one of these domains. Crucially, the authors frame this as a hypothesis about co-location and prioritization, not a claim that the coronavirus and natural hazards interact biologically or contribute equally to disaster losses.

The technical workflow proceeded in six sequential steps. First, the team assembled a municipal boundary shapefile covering all 46 barangays, municipal records of 2,505 confirmed COVID-19 cases, and ten natural-hazard layers obtained from the Batangas Provincial Government and the Department of Science and Technology–Philippine Institute of Volcanology and Seismology (DOST-PHIVOLCS). The hazard set spanned ground shaking, earthquake-induced and rain-induced landslides, storm surge, flood, tsunami, liquefaction, Taal Volcano base surge, Taal fissure, and volcanic tsunami. Because these layers expressed severity in incompatible ways, from Philippine Earthquake Intensity Scale classes to ordinal flood-depth categories to simple binary susceptibility classes, each category was converted into a common ordinal score.

A distinctive feature of the method was its area-sensitive treatment of hazard coverage. When a barangay contained more than one severity category for a given hazard, each category score was multiplied by the proportion of barangay area that it covered, and the contributions were summed. This retains information about both mapped severity and spatial extent, rather than assigning a single dominant class to an entire polygon. The authors are careful to note, however, that this measures areal exposure, not population exposure. A hazard zone covering a large fraction of a village may contain few residents, while a smaller zone may include most homes, schools, or critical facilities.

Aggregation followed a deterministic weighted-sum formula. The COVID-19 component for each barangay was scaled relative to the highest observed barangay count and multiplied by 100, and the health domain received 50 percent of the decision weight. The natural-hazard domain received the remaining 50 percent, distributed as follows: ground shaking, storm surge, and flood each carried 10 percent; earthquake-induced and rain-induced landslides 5 percent each; and tsunami, liquefaction, base surge, fissure, and volcanic tsunami 2 percent each. The equal 50/50 split is explicitly an a priori screening scenario, chosen to prevent either domain from dominating by design, and the authors stress that it does not imply equal contributions to expected mortality or economic loss. The hazard coefficients reflect locally reported frequency information rather than probabilistic loss estimates.

The results reveal why multi-layer screening matters. Bukal recorded the largest absolute number of confirmed COVID-19 cases, with 125, followed by Mataas Na Bayan with 116, Matingain I with 115, Wawa Ibaba with 108, Ayao-Iyao with 108, and Dayapan with 105. Yet when the natural-hazard components were added, Bukal ranked only 10th in the composite score. The five highest-priority barangays were District IV, Wawa Ibaba, Palanas, Mataas Na Bayan, and Sambal Ilaya. District IV, for example, combined high ground-shaking and earthquake-induced landslide categories with storm-surge, flood, tsunami, liquefaction, base-surge, fissure, and volcanic-tsunami exposure, alongside 99 recorded cases.

The gap between the disease-only ranking and the composite ranking illustrates what the output actually represents. A single COVID-19 layer would place Bukal first; the combined rubric instead identifies where recorded case counts and mapped hazard exposure accumulate most strongly under the specified decision rule. The authors emphasize that this answers a different planning question from either disease mapping or single-hazard mapping, and that the ranking is a relative planning priority under one transparent scoring scenario, not evidence of infection probability or calibrated disaster risk. Terms such as

The choice of the barangay as the analytical unit deserves particular attention, because it shapes what the resulting ranking can and cannot mean. Barangays are the smallest administrative divisions in the Philippine local government system, and they are the units through which municipal planning, emergency response, and health-service delivery are typically organized. By joining both the COVID-19 case records and the hazard layers to a single municipal boundary shapefile, the researchers ensured that information originating from different agencies could be summarized within comparable polygons. This is a practical necessity for multicriteria aggregation, but it also means that the analysis inherits the resolution limits of administrative geography: variations within a barangay, such as differences in building construction, evacuation access, or household density, are invisible at this scale.

The ten hazard layers themselves reflect the unusual geographic position of Lemery within the Taal volcanic region. Several of the hazards considered, including base surge, fissuring, and volcanic tsunami, are specifically associated with caldera volcanism and lake environments rather than with the more commonly mapped fluvial flood and coastal surge processes. Others, such as liquefaction and ground shaking, depend on subsurface conditions that can vary sharply over short distances. Because each hazard is generated by a different physical process and represented in a different data model, the study had to normalize them into a common ordinal scheme before any comparison was possible. This conversion step, while transparent, is also where interpretive judgment enters the workflow, since the mapping from qualitative hazard classes to numeric scores is a methodological choice rather than a measurement.

The area-adjustment procedure addresses a common pitfall in overlay analysis. If a hazard class covering only a small sliver of a barangay were counted with the same weight as one covering nearly the whole village, the composite score could be distorted by cartographic artifacts or boundary effects. Multiplying each category score by the proportion of barangay area affected preserves the spatial extent of exposure within the polygon. The authors nonetheless caution that areal coverage is not the same as population exposure, a distinction that matters greatly in practice: a hazard zone may blanket a large but sparsely settled portion of a village, while homes, schools, and critical facilities cluster in a smaller, more vulnerable pocket.

The weighting scheme also merits scrutiny. Assigning 50 percent of the decision weight to the health domain and 50 percent to the natural-hazard domain was an a priori choice designed to prevent either concern from dominating by construction. Within the hazard domain, the coefficients were distributed according to locally reported frequency information, giving ground shaking, storm surge, and flood 10 percent each, the two landslide types 5 percent each, and the remaining five hazards 2 percent each. These weights encode qualitative judgments about how often each hazard is expected to affect the municipality, not probabilistic estimates of loss. The additive combining rule then simply sums the scaled health component and the weighted, area-adjusted hazard scores, without modelling any interaction or joint probability between hazards or between hazards and disease.

The divergence between the disease-only and composite rankings illustrates the value of the combined view. Bukal led the municipality in absolute confirmed COVID-19 cases with 125, yet placed only tenth once hazard exposure entered the calculation. Conversely, barangays such as District IV and Sambal Ilaya rose in the composite ranking because they combined meaningful case counts with exposure to multiple hazards, including several of the volcanic and coastal threats specific to the Taal region. This kind of reordering is precisely what multi-risk assessments are intended to reveal: concentrations of concern that remain hidden when hazards and health data are examined separately.

The authors are explicit that the output is a relative planning priority under one transparent scoring scenario, not an estimate of infection risk or calibrated disaster risk. Raw case counts are not population-normalized incidence, and the study does not model disease dynamics, mobility, or health-system capacity. Similarly, the hazard scores describe mapped susceptibility and severity classes rather than expected losses. The ranking therefore answers a first-stage screening question: where should integrated preparedness attention, field validation, and resource planning begin?

The researchers outline several directions for strengthening future analyses. Population normalization would convert absolute case counts into incidence measures, allowing comparisons across barangays of different sizes. Sensitivity analysis would test how the ranking responds to alternative weights and scoring rules, addressing the known conditionality of composite indices. Adding vulnerability and exposure indicators, such as building stock, demographics, and critical facilities, would move the assessment closer to a genuine risk framework. Modelling hazard interactions, for example the way earthquakes can trigger landslides, and adopting finer-resolution spatial data would further refine the picture. Until such refinements are made, the study stands as a demonstration that routinely available agency datasets, standard GIS operations, and explicit assumptions can be combined into a defensible starting point for local multi-hazard and public-health planning.

Subject of Research: GIS-based spatial prioritization of overlapping natural-hazard exposure and COVID-19 case counts in Lemery, Batangas, Philippines

Article Title: GIS-based spatial prioritization of natural-hazard exposure and recorded COVID-19 case counts in Lemery, Batangas, Philippines

Article References: Permante, J. P., Cabral, E. R., & Padilla-Santiago, F. (2026). GIS-based spatial prioritization of natural-hazard exposure and recorded COVID-19 case counts in Lemery, Batangas, Philippines. Frontiers in Earth Science, 14, Article 1970276. https://doi.org/10.3389/feart.2026.1970276

Image Credits: AI Generated

DOI: 10.3389/feart.2026.1970276

Keywords: GIS, multi-hazard mapping, COVID-19, spatial prioritization, QGIS, Lemery Batangas, Taal volcano, disaster risk reduction, sieve mapping, barangay-level analysis, weighted overlay, geoinformatics

In a study published in Earth Science Informatics, Alexander Cotrina-Sanchez and Michele Torresani of the Free University of Bozen-Bolzano, together with Leonel Corado of the Universidade de Évora, present GEDIMetrics, an open-source plugin for the widely used geographic information system QGIS. The tool integrates the entire GEDI processing chain, from discovering and downloading raw data files to filtering, clipping, and fusing four complementary data products into a single analysis-ready dataset, all through a graphical interface that requires no programming. The plugin is freely available through the official QGIS Plugin Repository and on GitHub under the GNU General Public License.

The technical challenge the plugin addresses is deceptively simple to describe but notoriously tedious in practice. GEDI’s laser footprints, roughly 25 meters in diameter and spaced about 60 meters apart along the satellite’s track, capture the vertical distribution of returned laser energy. From these waveforms, NASA derives several distinct products: Level 2A provides canopy height and ground elevation metrics, Level 2B characterizes vertical canopy structure including canopy cover, plant area index, and foliage height diversity, Level 4A delivers aboveground biomass density estimates from globally calibrated models, and Level 4C offers the Waveform Structural Complexity Index, a machine-learning-derived measure of three-dimensional canopy complexity. Each product, however, is distributed as independent HDF5 files organized by laser beam, and the products must be aligned using a unique identifier called the shot number before any joint analysis becomes possible.

Existing tools each solved only fragments of this puzzle. NASA’s EarthData Search portal handles discovery and download but performs no subsetting, filtering, or cross-product integration. Programming libraries such as rGEDI and pyGEDI offer flexible scripted processing but demand coding expertise. Command-line frameworks like GEDI-Pipeline introduced automation, yet none provided a graphical workflow suitable for non-programmers, and none supported automated multi-product merging. GEDIMetrics closes that gap with a five-stage pipeline: granule discovery through NASA’s Common Metadata Repository API, authenticated download from EarthData, variable extraction with quality filtering, spatial subsetting to a user-defined region, and footprint-level merging of the selected products, followed by export to standard geospatial formats such as GeoPackage and GeoParquet.

The engineering behind the plugin reflects careful attention to both performance and reproducibility. Spatial filtering proceeds in two stages, beginning with a fast bounding-box pre-filter and followed by precise point-in-polygon clipping using Shapely geometric operations, an approach that substantially reduces processing time for granules covering far more area than the target region. Because data availability can differ across products due to variations in temporal coverage, algorithm convergence, or quality criteria, the merging stage uses Level 2A as the base product and offers configurable join strategies: an inner join retains only footprints present in all selected products, guaranteeing cross-product consistency, while an outer join preserves maximum spatial coverage at the cost of null values. Output files embed metadata recording the full processing configuration, including selected products, temporal range, region geometry, quality thresholds, and plugin version, ensuring that every analysis can be reproduced exactly.

Quality filtering, a critical step for GEDI data, is handled through an intuitive interface. Users can set minimum thresholds for product-specific quality flags, exclude footprints degraded by cloud or atmospheric effects, restrict analysis to land surfaces, and apply a global minimum beam sensitivity threshold, defaulting to 0.90 following criteria established for broad-scale applications. The sensitivity threshold can be raised for particular ecosystems, for example to 0.96 for temperate forests or higher for dense tropical canopies where the laser signal struggles to penetrate to the ground. A logging panel reports the number of granules identified and the estimated data volume before download begins, allowing users to adjust their parameters before committing to lengthy retrievals.

To demonstrate the plugin’s capabilities, the team tested GEDIMetrics across two contrasting forest ecosystems of roughly 100 square kilometers each: a temperate mixed forest in northeastern France and a humid tropical rainforest in the Madre de Dios region of southeastern Peru, an area partially overlapping the Tambopata National Reserve where intact forest faces mounting pressure from illegal gold mining. Using data from March 2019 to April 2023 with default quality settings, the temperate site yielded 96 granules and 5,939 forested footprints, while the tropical site produced only 23 granules and 1,712 footprints, a disparity that reflects the International Space Station’s orbital geometry, which delivers more frequent coverage at mid-latitudes than near the equator.

The case study also revealed striking asymmetries in product-level data availability that cannot be inferred from footprint counts alone. Retention of the Level 2B canopy structure product was moderate at the temperate site, at 60.8 percent of forested footprints, but nearly absent at the tropical site, at just 0.9 percent. The biomass product showed the inverse pattern, with 7.2 percent retention in France versus 87.2 percent in Peru. Despite these differences, the merged datasets painted a coherent ecological picture: the tropical forest exhibited taller canopies, with a 98th-percentile height of 24.05 meters compared to 18.86 meters at the temperate site, higher aboveground biomass density of 157.49 versus 139.51 megagrams per hectare, and greater structural complexity. Most strikingly, footprints inside the Tambopata National Reserve showed consistently taller canopies, 38 percent higher biomass density, and higher structural complexity than those in the surrounding buffer zone, a quantifiable signature of the degradation pressure from gold mining at the reserve’s edges.

At the temperate site, the plugin’s integrated dataset captured seasonal phenology with unexpected clarity. Canopy cover and plant area index rose sharply during the April-to-October vegetative period, consistent with the flush of deciduous foliage, while canopy height remained essentially stable across seasons, confirming that height metrics are largely insensitive to phenological state. The most dramatic seasonal signal appeared in the median height percentile, which jumped from 1.27 meters in the leafless months to 6.36 meters during the growing season, reflecting the increased interception of laser energy by mid-canopy leaves. This ability to distinguish genuine structural change from seasonal foliage dynamics has direct implications for disturbance monitoring and carbon accounting.

The researchers are candid about current limitations. Processing is single-machine and sequential, which can make large regions slow to analyze, and the plugin does not yet correct GEDI’s footprint geolocation uncertainty of roughly 10 meters, though its modular design allows integration with external correction tools. Future development priorities include polygon-level summary statistics, diagnostics quantifying footprint loss at each filtering stage, in-plugin visualization of vertical foliage profiles, and a hybrid local-cloud workflow leveraging NASA’s evolving cloud-native archives. The team also envisions extending the plugin to ICESat-2 canopy height data, which would push coverage beyond GEDI’s latitudinal limits into boreal and high-latitude forests. For now, GEDIMetrics stands as a quietly transformative piece of software: by collapsing a multi-step, script-heavy workflow into a transparent graphical process inside the world’s most widely used open-source GIS, it puts the vertical architecture of the world’s forests within reach of anyone with a computer, a QGIS installation, and a free NASA EarthData account.

Subject of Research: An open-source QGIS plugin for accessing, filtering, and fusing NASA GEDI spaceborne LiDAR forest structure data products

Article Title: GEDIMetrics: a QGIS plugin for accessing and integrating multi-product GEDI spaceborne LiDAR data

Article References: Cotrina-Sanchez, A., Torresani, M., & Corado, L. (2026). GEDIMetrics: a QGIS plugin for accessing and integrating multi-product GEDI spaceborne LiDAR data. Earth Science Informatics, 19(11), Article 205. https://doi.org/10.1007/s12145-026-02254-z

Image Credits: AI Generated

DOI: 10.1007/s12145-026-02254-z

Keywords: GEDI, spaceborne LiDAR, QGIS, forest structure, aboveground biomass, canopy height, remote sensing, open-source software, geospatial data, NASA EarthData, tropical forest, forest carbon

The original release of YOLO-MOD already offered something genuinely useful: a general-purpose framework for running YOLO object detection models inside standard QGIS workflows. Users could perform multi-class object detection using models in both PyTorch and ONNX formats, visualize the results as georeferenced vector layers overlaid on their maps, and export annotated datasets in the YOLO format for training future models. Basic post-processing utilities, such as layer merging, canvas tiling, and previewing exported annotations, rounded out an end-to-end workflow. But there was a significant constraint. The first version could only detect and classify five predefined object categories: ships, aircraft, helicopters, airports, and storage tanks. Anyone wanting to detect something else, from solar panels to wind turbines, was simply out of luck.

The updated version 2.1.1 removes that ceiling entirely. Users can now load and apply their own trained Ultralytics YOLO models, compatible with YOLOv8 and later architectures, with an arbitrary number and type of object classes. The technical mechanism behind this is elegantly simple. When a model stored in the PyTorch .pt format is loaded through the plugin’s interface, YOLO-MOD automatically extracts the object class names embedded in the model and dynamically generates the appropriate configuration controls in the user interface for each class. This means researchers can configure visualization and processing options for their custom classes exactly as they would for the built-in standard models, without ever touching a line of plugin code. The same workflow now applies to virtually any detection task based on horizontal bounding boxes.

The implications for applied remote sensing are considerable. A coastal monitoring team could train a model on harbor infrastructure and vessel types specific to their region, load it into QGIS, and immediately begin scanning aerial imagery for those objects. An airport safety analyst could build a detector for ground support equipment. An environmental agency could train a network to flag illegal dumping sites or deforestation edges. In each case, the results appear directly as color-coded bounding boxes on georeferenced layers, superimposed on basemaps such as Bing Aerial, meaning detections land in the same coordinate system as every other dataset in the project. That integration eliminates the clumsy export-import cycles that have long characterized deep learning workflows in the geospatial domain.

The update also tackles one of the most practical frustrations of working with automated detection: false positives. A new Edit Detections tab extends the plugin’s post-processing capabilities with interactive tools for removing individual detections from an existing vector layer. Users can select one or more detected objects, highlight them visually on the map, and permanently delete them without rerunning the entire detection process. This matters more than it might sound. In operational settings, analysts routinely spend hours cleaning up automated results before they can be used, and every detection run that must be repeated from scratch wastes computational resources and time. Being able to surgically remove spurious detections within QGIS itself keeps the analytical workflow inside a single environment.

The editing workflow is designed to feel native to QGIS users. Analysts select features using the familiar QGIS Selection Toolbar, for instance with the Select Features by Area or Single Click tool, and the plugin’s Selection Editing panel reports how many features are ready to be removed with a single button press. The layer merging functionality from the previous version has been retained and enhanced alongside these new tools, so refining a large multi-part detection result remains straightforward. The developers describe this as a way to let detection results be refined before further analysis or export, which is precisely the point in the pipeline where most real-world projects stall.

The third major enhancement concerns training data preparation, often the most labor-intensive phase of any machine learning project in remote sensing. The Canvas Tiling tool, which automatically splits the current QGIS map extent into image tiles suitable for training object detection models, has been upgraded with an interactive preview of the tile grid before any files are generated. Users configure tile width and height, the overlap percentage between adjacent tiles, the padding color for regions outside the available map content, and the output directory. Rather than committing immediately, a Preview and Start Tiling button first displays the proposed grid, letting the user verify the layout and then confirm or cancel the operation. Areas beyond the map content are automatically padded in the user-defined color, white by default, ensuring that every generated tile has identical dimensions regardless of the shape or size of the selected extent.

Under the hood, the plugin is built entirely in Python and designed for QGIS 3.x, with testing reported on versions 3.40 and 3.42 under Windows 11 using the embedded Python environment. Dependencies include PyTorch, with CUDA-enabled builds recommended for GPU acceleration, ONNX Runtime for flexible CPU and GPU inference, and NumPy. The software is distributed under the GNU General Public License version 3.0, versioned with Git, and publicly available through its GitHub repository, keeping it fully open source. That licensing choice matters for public agencies and research groups in developing regions, for whom commercial geospatial AI platforms can be prohibitively expensive and whose data policies sometimes forbid cloud-based processing altogether.

The broader context is a field moving fast. Object detection in remote sensing has been propelled by the growing availability of high-resolution imagery and by advances in detection algorithms since the original You Only Look Once architecture was introduced in 2016. Yet although several QGIS plugins now support selected deep learning tasks, the developers note that flexible frameworks for multi-category object detection remain relatively limited. YOLO-MOD’s approach of dynamically adapting its interface to whatever model the user supplies is a notable contribution to closing that gap, particularly for teams that lack the software engineering resources to build custom integrations.

What makes this release interesting beyond its immediate feature list is what it signals about the direction of geospatial software. The division between the GIS, where spatial data lives, and the machine learning environment, where models run, has been eroding for years, and tools like YOLO-MOD represent the pragmatic middle path: rather than rebuilding GIS capabilities inside a deep learning framework, they embed the models directly into the analyst’s existing environment. For the growing community of researchers tracking ships in shipping lanes, aircraft at airfields, and infrastructure across vast landscapes, the ability to load any YOLOv8 or later model, clean up its outputs interactively, and tile fresh training imagery from the same map canvas constitutes a genuinely streamlined loop, one that could shorten the cycle from data collection to working detector from weeks to days.

Subject of Research: A YOLO-based multi-category object detection plugin for optical remote sensing imagery integrated into the QGIS geographic information system.

Article Title: Version [ 2.1.1 ] – [ YOLO-MOD: YOLO-based multi-category object detection plugin for maritime, aviation, and infrastructure objects in optical remote sensing images in QGIS ]

Article References: Ciecholewski, M., & Strzelecki, M. (2026). Version [2.1.1]- [YOLO-MOD: YOLO-based multi-category object detection plugin for maritime, aviation, and infrastructure objects in optical remote sensing images in QGIS]. SoftwareX, 35, Article 103002. https://doi.org/10.1016/j.softx.2026.103002

Image Credits: AI Generated

DOI: 10.1016/j.softx.2026.103002

Keywords: YOLO-MOD, object detection, QGIS, remote sensing, deep learning, YOLOv8, geographic information systems, Ultralytics, satellite imagery, maritime detection, infrastructure detection, open-source software