Canopy Height Model Research Articles

Webbing 3D map from UAV images in Da Lat City, Lam Dong Province

Abstract Worldwide, big cities are accelerating to fulfill the establishment of Spatial Data Infrastructure (SDI), which is a key task towards smart city. The mainstay of SDI is to create an important foundation of data capacity to provide necessary and essential information related to the requirements of urban and smart city management. With the advantages of high accuracy, low cost, and flexibility in 3D data creation, UAV (Unmanned Aerial Vehicle) images have been widely applied for the purpose of digital transformation purposes. This article aims to firstly utilize the new opportunities of using UAVs in 3D mapping and sharing 3D map data through the WebGIS platform. Secondly, processing techniques will be described after collecting the UAV images to create orthomosaics, point clouds, and Digital Surface Model (DSM). A classification of ground points was conducted to create a Digital Terrain Model (DTM) in combining with DSM to form a calibrated surface numerical model (Canopy Height Model, CHM). The images map was then digitalized coupled with the object height extracted from the CHM model and assigning attributable data to each object. The results suggested that ArcGIS Online allows users to share information widely, targeting multiple users, and to reduce the cost of GIS software investment in building thematic classes for urban infrastructure management.

Repeated drone photogrammetry surveys demonstrate that reconstructed canopy heights are sensitive to wind speed but relatively insensitive to illumination conditions

ABSTRACT Unoccupied aerial vehicle (UAV) based structure-from-motion (SfM) photogrammetry surveys are becoming a standard tool for ecologists to measure plant structure and biomass in non-forest ecosystems. The reproducibility of SfM survey results under different operational conditions, namely wind speed, sun elevation, and cloud condition, is poorly understood. It is also unclear to what extent commonly applied point-to-grid interpolation of derived point clouds affects inference of vegetation structure. These knowledge gaps limit the use of these methods for measuring and monitoring detection. We captured 61 UAV SfM surveys at the same study area under a range of wind/sun/cloud conditions over a 24-day period during 2021 and used generalized linear mixed effects models to test how the structural reconstructions varied with environmental conditions. Wind speed significantly influenced the canopy height reconstructions, with greater wind speeds reducing mean canopy height. Different plant species exhibited varying sensitivities to wind that are likely related to leaf attributes (size, structure, and density), growth form of the canopy, and physical properties such as limb flexibility. The movement of plants can reduce canopy height estimates derived from photogrammetric surveys, even under relatively low wind speeds. Reconstructed canopy heights were comparatively insensitive to solar elevation variations. Cloud conditions and illumination by direct sunlight had a weak, non-significant effect on reconstructed canopy height, with sunny conditions (generating shadows) resulting in a measurable but marginal reduction in mean canopy heights. When comparing canopy heights of interpolated and discontinuous derived canopy height models the results highlighted that these were insensitive to interpolation in this specific setting. We recommend measuring mean wind speeds throughout surveys where comparisons are to be made between different drone-based SfM surveys of vegetation over either time or space. Care should be taken to ensure that the effects of wind are controlled for so that inferences are valid.

Detection of Individual Corn Crop and Canopy Delineation from Unmanned Aerial Vehicle Imagery

Precise monitoring of individual crop growth and health status is crucial for precision agriculture practices. However, traditional inspection methods are time-consuming, labor-intensive, prone to human error, and may not provide the comprehensive coverage required for the detailed analysis of crop variability across an entire field. This research addresses the need for efficient and high-resolution crop monitoring by leveraging Unmanned Aerial Vehicle (UAV) imagery and advanced computational techniques. The primary goal was to develop a methodology for the precise identification, extraction, and monitoring of individual corn crops throughout their growth cycle. This involved integrating UAV-derived data with image processing, computational geometry, and machine learning techniques. Bi-weekly UAV imagery was captured at altitudes of 40 m and 70 m from 30 April to 11 August, covering the entire growth cycle of the corn crop from planting to harvest. A time-series Canopy Height Model (CHM) was generated by analyzing the differences between the Digital Terrain Model (DTM) and the Digital Surface Model (DSM) derived from the UAV data. To ensure the accuracy of the elevation data, the DSM was validated against Ground Control Points (GCPs), adhering to standard practices in remote sensing data verification. Local spatial analysis and image processing techniques were employed to determine the local maximum height of each crop. Subsequently, a Voronoi data model was developed to delineate individual crop canopies, successfully identifying 13,000 out of 13,050 corn crops in the study area. To enhance accuracy in canopy size delineation, vegetation indices were incorporated into the Voronoi model segmentation, refining the initial canopy area estimates by eliminating interference from soil and shadows. The proposed methodology enables the precise estimation and monitoring of crop canopy size, height, biomass reduction, lodging, and stunted growth over time by incorporating advanced image processing techniques and integrating metrics for quantitative assessment of fields. Additionally, machine learning models were employed to determine relationships between the canopy sizes, crop height, and normalized difference vegetation index, with Polynomial Regression recording an R-squared of 11% compared to other models. This work contributes to the scientific community by demonstrating the potential of integrating UAV technology, computational geometry, and machine learning for accurate and efficient crop monitoring at the individual plant level.

Tracking shifts in forest structural complexity through space and time in human‐modified tropical landscapes

Habitat structural complexity is an emergent property of ecosystems that directly shapes their biodiversity, functioning and resilience to disturbance. Yet despite its importance, we continue to lack consensus on how best to define structural complexity, nor do we have a generalised approach to measure habitat complexity across ecosystems. To bridge this gap, here we adapt a geometric framework developed to quantify the surface complexity of coral reefs and apply it to the canopies of tropical rainforests. Using high‐resolution, repeat‐acquisition airborne laser scanning data collected over 450 km2 of human‐modified tropical landscapes in Borneo, we generated 3D canopy height models of forests at varying stages of recovery from logging. We then tested whether the geometric framework of habitat complexity – which characterises 3D surfaces according to their height range, rugosity and fractal dimension – was able to detect how both human and natural disturbances drive variation in canopy structure through space and time across these landscapes. We found that together, these three metrics of surface complexity captured major differences in canopy 3D structure between highly degraded, selectively logged and old‐growth forests. Moreover, the three metrics were able to track distinct temporal patterns of structural recovery following logging and wind disturbance. However, in the process we also uncovered several important conceptual and methodological limitations with the geometric framework of habitat complexity. We found that fractal dimension was highly sensitive to small variations in data inputs and was ecologically counteractive (e.g. higher fractal dimension in oil palm plantations than old‐growth forests), while rugosity and height range were tightly correlated (r = 0.75) due to their strong dependency on maximum tree height. Our results suggest that forest structural complexity cannot be summarised using these three descriptors alone, as they overlook key features of canopy vertical and horizontal structure that arise from the way trees fill 3D space.Keywords: Forest disturbance, LiDAR, logging, recovery, remote sensing, structural complexity

Individual tree detection and crown delineation in the Harz National Park from 2009 to 2022 using mask R–CNN and aerial imagery

Forest diebacks pose a major threat to global ecosystems. Identifying and mapping both living and dead trees is crucial for understanding the causes and implementing effective management strategies. This study explores the efficacy of Mask R–CNN for automated forest dieback monitoring. The method detects individual trees, delineates their crowns, and classifies them as alive or dead. We evaluated the approach using aerial imagery and canopy height models in the Harz Mountains, Germany, a region severely affected by forest dieback. To assess the model's ability to track changes over time, we applied it to images from three separate flight campaigns (2009, 2016, and 2022). This evaluation considered variations in acquisition dates, cameras, post-processing techniques, and image tilting. Forest changes were analyzed based on the detected trees' number, spatial distribution, and height. A comprehensive accuracy assessment demonstrated the Mask R–CNN's robust performance, with precision scores ranging from 0.80 to 0.88 and F1-scores from 0.88 to 0.91. These results confirm the model's ability to generalize across diverse image acquisition conditions. While minor changes were observed between 2009 and 2016, the period between 2016 and 2022 witnessed substantial dieback, with a 64.57% loss of living trees. Notably, taller trees appeared to be particularly affected. This study highlights Mask R–CNN's potential as a valuable tool for automated forest dieback monitoring. It enables efficient detection, delineation, and classification of both living and dead trees, providing crucial data for informed forest management practices.

Seasonal optimisation of drone‐based photogrammetry in a heterogeneous boreal landscape

AbstractAimsUncrewed aerial vehicles (UAV), or drones, have become more affordable and easier to use, resulting in increased UAV applications in ecology and conservation. However, solar illumination, vegetation phenology and prevailing weather conditions will impact the quality of the derived products to differing degrees. In this study, we investigate how seasonal differences in solar illumination, tree foliage and weather conditions impact the accuracy of digital elevation models (DEM) and canopy height models (CHM) in a heterogeneous boreal landscape.MethodsWe compared DEMs and CHMs derived from drone photogrammetry with DEMs and CHMs produced from a drone‐mounted laser scanner across three seasons with different solar illumination, tree foliage and weather conditions during leaf‐off and leaf‐on seasons. Photogrammetric height models were evaluated across three land‐cover classes consisting of open areas, sparse‐forest and forest. The most accurate CHM for sparse‐forest was produced during summer under overcast conditions, whereas for the forest class, summer under clear skies was best.ResultsStructure from motion (SfM) photogrammetry performed well against the LiDAR survey in most cases with correlations between sampled points of up to R2 = 0.995. Root mean square errors (RMSEs) were <1.5 m in all DEMs and as low as 0.31 m in autumn clear‐sky data over open terrain. CHM RMSEs were somewhat higher in all cases except under winter overcast conditions when the RMSE for sparse‐forest reached 6.03 m.ConclusionsWe have shown that SfM photogrammetry is surprisingly robust to variations in vegetation type, tree phenology and weather, and performs well in comparison with a reference LiDAR data set. Our results show that, in boreal forests, autumn is the preferred season under clear‐sky conditions for DEM generation from SfM photogrammetry across all land‐cover classes, whereas summer is preferred for CHM modelling with a small trade‐off between overcast and clear‐sky conditions over different vegetation types. These results can help potential SfM users in ecology and forestry plan missions and review the quality of products derived from drone photogrammetry products.

Spatial estimation of soil organic carbon, total nitrogen, and soil water storage in reclaimed post-mining site based on remote sensing data

The estimation of Soil Organic Carbon (SOC), Total Nitrogen (TN), and Soil Water Storage (SWS) is crucial in comprehending ecosystem services and environmental sustainability. It plays a crucial role in guiding sustainable restoration strategies and supporting the long-term health of post-mining sites. Remote sensing technology provides valuable tools for modelling and mapping soil properties in reclaimed post-mining sites efficiently and cost-effectively. This study aimed to utilize remote sensing data to estimate SOC, TN, and SWS in a reclaimed post-mining site. Field data was collected from 130 research plots to obtain reference data for SOC, TN, and SWS from the Sonica hard coal post-mine spoil heap. Remote sensing data were: airborne laser scanning (ALS) point clouds and Planets cope satellite imageries. Generalized Additive Models (GAM) were used to develop predictive models. Wall-to-wall predictions of analyzed variables were performed. The results identified topographic and remote sensing indicators that significantly influence SOC, TN, and SWS. Digital Terrain Model (DTM), aspect, and blue spectral band are variables that explain SOC storage, with a significant influence of DTM, ranging from −8 to 18 Mg ha−1. TN was explained by DTM, Canopy Height Model (CHM), blue and Near Infrared (NIR) spectral bands, and Normalized Difference Vegetation Index (NDVI), mainly influenced by NIR and NDVI, ranging from −1.1 to 0.8 and −0.9 to 1.4 Mg ha−1, respectively. The values of Topographic Wetness Index (TWI), aspect, CHM, blue and NIR spectral bands explained SWS, highlighting their importance in assessing soil water dynamics in post-mining landscapes, with TWI and CHM being particularly influential, ranging from −2 to 5.1 and −6 to 2 mm, respectively. However, caution is advised when predicting SOC and TN using remote sensing in post-mining sites due to geogenic carbon considerations.

Use of lidar for monitoring vegetation growth dynamics in reclaimed mine lands in Kentucky

Surface coal mining in the Appalachian region has led to a significant forest disturbance over time. Evaluating the effectiveness of current reclamation practices in promoting vegetation growth on reclaimed mine sites is a key to understanding how much vegetation has changed in those sites since reclamation. This study employed statewide airborne lidar data to assess changes in lidar vegetation structural metrics on reclaimed mine lands in the Lower Levisa Watershed of Eastern Kentucky between 2011 and 2019 and compare vegetation growth at various reclaimed sites reclaimed in different decades. Eighteen inactive surface mines were selected for the study and categorized into four groups based on the release of their reclamation bonds in different decades. Lidar point cloud data were processed in ArcGIS Pro using filtering and segmentation algorithms to calculate various vegetation attributes from the point clouds, including maximum vegetation height (Hmax), mean height (Hmean), standard deviation of height (HSD), canopy cover (CC), and height percentiles (10, 50 and 75), which were represented as lidar metrics. The process of generating the lidar metrics involved creating Digital Elevation Models (DEMs) and Digital Surface Models (DSMs), calculating Canopy Height Models (CHMs), creating LAS height metrics and generating point statistics rasters to derive these metrics. Change maps for each metric were visually assessed over time, and circular plots with a radius of 12 m were established within each site for further statistical analysis. Significant changes in lidar vegetation metrics were observed between 2011 and 2019 with significant differences among sites reclaimed at different time periods. There was an overall increase in Hmean from 2011 to 2019, with values ranging from 2.4 to 3.8 m. Sites reclaimed in the 1980s experienced an average decrease in canopy cover of −0.5%, while those from the 1990s, 2000s, and 2010s demonstrated increases of 4.9%, 10.1%, and 18.1%, respectively, suggesting that canopy growth rates are higher in younger sites compared to older ones. Vertical variability of the vegetation also increased over time, as indicated by increasing HSD values. Utilizing statewide airborne lidar data allowed for a comprehensive and detailed assessment of vegetation dynamics on reclaimed mine lands. The findings of this study serve as a foundation for future research endeavors focused on vegetation recovery assessment and success in reclaimed mine lands using lidar data.

Development of a computational routine to evaluate the height of cotton plants

Aerial photogrammetry with drones is being frequently used in agriculture. Since computational tools can optimize various personalized workflows for Digital Agriculture, the objective of this study was to develop modules (scripts) of a computational routine called FindPLANT-V1, leaving it personalized for testing on a preliminary basis, through products aerial photogrammetry obtained with a drone, the spatial and temporal variability of the height of cotton plants. To develop the study, an experimental area of ​​500 m2 was selected with irrigated cotton cultivation at the State University of Montes Claros (Unimontes) Experimental Farm, in Janaúba/MG. The FindPLANT-V1 routine has four modules, structured through object-oriented programming with Python 3.8 in the Spyder IDE. The following libraries were used to develop the scripts: Gdal, Imageio, Math, Matplolib, Numpy, OpenCV, OS, OSR, OGR, Pandas, Rasterstats and Scipy. A script was also developed to create the essential directories (inputs and outputs) for the operation of this routine. As a result of aerophotogrammetric geoprocessing, the average ground sample distances (GSD) of classic orthophotomosaics was 1.6 cm. The average GSD of the Digital Surface Model (DSM) and Digital Terrain Model (DTM) was 3.2 cm. Such products, routine inputs, made it possible to estimate the spatial and temporal variability (module 1.ST), the canopy height model (module 2.ProCHM), the zonal statistics of the plots (3.ZonalStat). The results of the routine developed, although partial, made it possible to analyze the spatial variability of cotton plants and quantify the height in different phases of the cycle, presenting flexibility of adaptation for herbaceous crops.

High-resolution mapping of tree species and associated uncertainty by combining aerial remote sensing data and convolutional neural networks ensemble

Mapping tree species diversity is essential for monitoring and managing forest ecosystems. Automating ecoforestry mapping using remote sensing images remains an important challenge due to the tremendous variability in forest covers and the conditions under which images used for classification are acquired. Deep learning algorithms have been increasingly used over the last few years to analyse remote sensing data for mapping tree species. However, most of these studies focus only on a small number of species or a limited area and avoid providing a spatially explicit representation of the uncertainty related to the predictions, rendering them unsuitable for operational use.In this study, we used an ensemble of convolutional neural networks to map forest species and land cover types across a 10,000 km2 area in Quebec, Canada, spanning mixed and boreal forests. We built a georeferenced label database to train and test nine models, which resulted from the combination of three training datasets and three-commonly used convolutional neural network architectures (VGG16, ResNet50v2, Densenet121). These models were trained on multiband aerial photographs and a canopy height model derived from airborne lidar point clouds and used to map the diversity and distribution of tree species and land cover types. The level of agreement among models was used to generate uncertainty maps. The performance of the super-ensemble using 1311 independent forest inventory plots and assessed the extent to which uncertainty maps could serve as a spatially explicit indicator of model performance.The super-ensemble achieved 90% global accuracy. Our results indicated that the performance of the super-ensemble surpassed that of all individual architectures while also showing a positive effect of the canopy height model on performance. The comparison of the super-ensemble map with the proportion of basal area measured in forest inventory plots confirmed the reliability of the model over the study area. Our results also indicated that mapping the inter-model agreement provides a reliable spatially explicit estimate of model performance. The robustness and reliability of the proposed approach support its use in an operational context while also providing a conceptual framework to evaluate the reliability of uncertainty maps.

A Hybrid Method for Individual Tree Detection in Broadleaf Forests Based on UAV-LiDAR Data and Multistage 3D Structure Analysis

Individual tree detection and segmentation in broadleaf forests have always been great challenges due to the overlapping crowns, irregular crown shapes, and multiple peaks in large crowns. Unmanned aerial vehicle (UAV)-borne light detection and ranging (LiDAR) is a powerful tool for acquiring high-density point clouds that can be used for both trunk detection and crown segmentation. A hybrid method that combines trunk detection and crown segmentation is proposed to detect individual trees in broadleaf forests based on UAV-LiDAR data. A trunk point distribution indicator-based approach is first applied to detect potential trunk positions. The treetops extracted from a canopy height model (CHM) and the crown segments obtained by applying a marker-controlled watershed segmentation to the CHM are used to identify potentially false trunk positions. Finally, the three-dimensional structures of trunks and branches are analyzed at each potentially false trunk position to distinguish between true and false trunk positions. The method was evaluated on three plots in subtropical urban broadleaf forests with varying proportions of evergreen trees. The F-score in three plots ranged from 0.723 to 0.829, which are higher values than the F-scores derived by a treetop detection method (0.518–0.588) and a point cloud-based individual tree segmentation method (0.479–0.514). The influences of the CHM resolution (0.25 and 0.1 m) and the data acquisition season (leaf-off and leaf-on) on the final individual tree detection result were also evaluated. The results indicated that using the CHM with a 0.25 m resolution resulted in under-segmentation of crowns and higher F-scores. The data acquisition season had a small influence on the individual tree detection result when using the hybrid method. The proposed hybrid method needs to specify parameters based on prior knowledge of the forest. In addition, the hybrid method was evaluated in small-scale urban broadleaf forests. Further research should evaluate the hybrid method in natural forests over large areas, which differ in forest structures compared to urban forests.

Geometric vs spectral content of Remotely Piloted Aircraft Systems images in the Precision agriculture context

In the last years the agricultural sector has been evolving and new technologies, like Unmanned Aerial Vehicles (UAV) and satellites, were introduced to increase crop management efficiency, reducing environmental costs and improving farmers’ income. MAIA-S2 sensor is presently one of the most performing optical sensors operating on a Remotely Piloted Aircraft Systems (RPAS); given its spectral features, it aims at supporting a scaling process where monoscopic satellite data (namely Copernicus S2) with high temporal and limited geometric resolution can be integrated with stereoscopic data from RPAS having a very high spatial resolution. In this work, data from MAIA-S2 sensor were used to detect the effects of different fertilization types on corn with reference to a test field located in Carignano (Piemonte region, NW-Italy). Different amounts of top dressing fertilization were applied on corn and an RPAS acquisition operated on 14th June 2021 (corresponding date to the corn stem elongation stage) to explore if any effects could be detectable. Three spectral indices, namely Normalized Difference Vegetation Index, Normalized Difference Red Edge index and Canopy Height Model, computed from at-the-ground reflectance calibrated MAIA-S2 data, were compared to evaluate the correspondent response to the different fertilization rates. Results show that: (i) NDVI poorly detect N-related differences zones; (ii) NDRE and CHM reasonably reflect the different N fertilization doses; (iii) Only CHM proved to be able to detect crop height and, consequently, biomass differences that are known to be induced by different rates of fertilization.

Evaluating the efficacy of sampling acquisition paths for mapping vegetation structure using terrestrial mobile laser scanning.

Mobile Laser Scanning (MLS) is rapidly increasing in popularity for the capture and quantification of vegetation structure in natural landscapes. However, understanding the optimal implementation for forest capture in the field is still limited, particularly regarding the choice in acquisition path and length followed during data capture. How variable is the characterisation of vegetation structure when using different acquisition paths, and is this difference likely to be significant for most users? In this study we compared four acquisition path designs commonly cited in the literature to determine the importance of this choice for users when capturing vegetation structure. MLS point clouds were systematically captured to repeatedly survey study plots in a closed-canopy forest ecosystem in south-eastern Australia. Digital elevation models, canopy height models, and vertical voxel occupancy profiles were derived to illustrate the sensitivity of the path's length and configuration to variability in the quantification of vegetation structure. We found strong agreement between the acquisition path designs that increased as path length increased. No significant differences were found in digital elevation models derived at 75% and 100% path lengths. However, differences were observed in two and four of the 25% and 50% path length plots respectively. Significant differences in canopy height model data were only found in one plot at the 50% path length. Mean differences to the reference digital elevation models was 0.09 m across all plots, and 0.4 m for the canopy height models. Voxel occupancy profiles showed the greatest agreement between the understory and canopy where the mean difference between acquisition paths across all plots was 5.72%. Our findings have significant implications for the use of MLS from an operational perspective, as they illustrate the variability and reliability of their use in natural forested landscapes. Users can choose to balance their choice in acquisition path design with path length to meet requirements such as efficiency or ease of data capture in the field.

A new method for individual treetop detection with low-resolution aerial laser scanned data

In the past decade, the use of three-dimensional forest information from airborne Light Detection and Ranging (LiDAR) has become widespread in forest inventories. Accurate Individual Treetop Detection (ITD) and crown boundary delineation using LiDAR data are critical for obtaining precise inventory metrics. To address this need, we introduced a novel growing tree region (GTR)-driven ITD method that utilizes canopy height models (CHM) derived from very low-resolution airborne LiDAR data. The GTR algorithm consists of three key stages: (i) preserving all height layers through incremental cutting and stacking of CHM; (ii) employing a three-layer concept to identify individual treetops; and (iii) refining the detected treetops using a distance-based filter. Our method was tested in five temperate forests across Central Europe and was compared against the widely-used local maxima (LM) search combined with an optimized variable window filtering (VWF) technique. Our results showed that the GTR method outperformed LM with VWF, particularly in forests with high canopy density. The achieved root mean square accuracies were 74% for the matching rate, 19% for commission errors, and 27% for omission errors. In comparison, the LM with the VWF method resulted in a matching rate of 71%, commission errors of 20%, and omission errors of 31%. To facilitate the application of our algorithm, we developed an R package called TREETOPS, which seamlessly integrates with the lidR package, ensuring compatibility with existing treetop-based segmentation methods. By introducing TREETOPS, we provide the most accurate open-source tool for detecting treetops using low-resolution LiDAR-derived CHM.

Indicators for monitoring reduced impact logging in the Brazilian amazon derived from airborne laser scanning technology

Monitoring reduced impact logging (RIL) activities in sustainably managed forest areas in the Amazon is a costly and complex, yet crucial endeavor. One viable monitoring option is the use of airborne laser scanning (LiDAR), which enables estimating forest structure parameters over large areas in a reduced timeframe with high accuracy. In this study, we propose and assess the applicability of five monitoring indicators for RIL based on Light Detection and Ranging (LiDAR) data acquisition in areas under forest concession. Five Annual Production Units (APUs) were investigated within the Forest Management Unit (FMU) III of the Jamari National Forest, located in the Southwest of the Brazilian Amazon. These sites were surveyed by LiDAR in 2014 and 2015 (one year after the exploration). Digital Terrain Models (DTMs), Surface Models (DSMs), and Canopy Height Models (CHMs) were generated for each APU. The proposed indicators were: i. Detection and identification of crown removal in dominant and co-dominant trees above 30 m; ii. Gap detection in the forest canopy; iii. Impacts of Reduced Impact Logging on the Understory; iv. Changes in the vertical canopy profile; and v. Affected areas within Permanent Preservation Areas (PPAs) and restricted areas. There was a 3.95% reduction in the occurrence of taller canopies after RIL, and a higher occurrence of small gaps (λ > 1), with λ values (2.34) being higher in the area with the oldest logging history (APU 1). Gini coefficient values in all APUs were below 0.5, indicating a low intensity of disturbances in the forest canopy. The shape (γ) and scale (β) parameters of the understory and canopy were not significantly correlated with variables related to selective logging. Restricted areas were considered for the allocation of roads, trails, log landings, and places with slopes equal to or >15%, and the indices of areas affected by RIL in PPAs and restricted areas were <2%. The proposed indicators using LiDAR data show great potential for monitoring managed areas in the Amazon and can be utilized by concession companies and government oversight.

Forest canopy height modelling based on photogrammetric data and machine learning methods

AbstractForest topographic survey is a problem that photogrammetry has not solved for a long time. Forest canopy height is a crucial forest biophysical parameter which is used to derive essential information about forest ecosystems. In order to construct a canopy height model in forest areas, this study extracts spectral feature factors from digital orthophoto map and geometric feature factors from digital surface model, which are generated through aerial photogrammetry and LiDAR (light detection and ranging). The maximum information coefficient, Pearson, Kendall, Spearman correlation coefficients, and a new proposed index of relative importance are employed to assess the correlation between each feature factor and forest vertical heights. Gradient boosting decision tree regression is introduced and utilised to construct a canopy height model, which enables the prediction of unknown canopy height in forest areas. Two additional machine learning techniques, namely random forest regression and support vector machine regression, are employed to construct canopy height model for comparative analysis. The data sets from two study areas have been processed for model training and prediction, yielding encouraging experimental results that demonstrate the potential of canopy height model to achieve prediction accuracies of 0.3 m in forested areas with 50% vegetation coverage and 0.8 m in areas with 99% vegetation coverage, even when only a mere 10% of the available data sets are selected as model training data. The above approaches present techniques for modelling canopy height in forested areas with varying conditions, which have been shown to be both feasible and reliable.

Leveraging optimal UAV-lidar features for two CNN approaches to delineate mangrove tree crowns

Effective individual tree crown delineation plays a critical role for assessing mangrove quality and health. However, mangrove tree crowns often form large, interconnected clusters in dense coverage areas, making it difficult to separate them individually. Moreover, many mangrove trees have multiple large branches that result in irregular tree shapes and create internal gaps inside their crowns. Unmanned Aerial Vehicle (UAV) lidar data can potentially overcome these challenges, but a comprehensive assessment of UAV-lidar-derived features is still missing, as well as how to best use them for dense mangrove forests. In this study, we set forth two objectives: (1) to derive optimal features for mangrove individual tree crown delineation through an exclusive examination of all the UAV-lidar features; (2) to develop effective methods that can best incorporate these optimal UAV-lidar features. To achieve the first objective, we extracted 224 features from UAV-lidar data based on three groups of attributes: height, intensity, and point density. Seven spatial scales ranging from 0.2 to 0.5 m were tested in this process. For the second objective, we applied two state-of-the-art Convolutional Neural Networks: Mask Region-based Convolutional Network method (Mask R–CNN) and Ultralytics You Only Look Once version 8 (YOLOv8). At last, the derived three optimal features are: canopy height model at 0.20 m, coefficient of point cloud height variation at 0.25 m, and ground point percentage at 0.25 m. Comparing our methods with traditional methods that only use canopy height model, we found that integrating Convolutional Neural Networks and the optimal UAV-lidar features improved the accuracy by more than 13%. Mask R–CNN was better for dense mangroves, while YOLOv8 was excellent for sparse and short mangroves. The derived optimal UAV-lidar features further enhanced the detection of short trees, densely clumped trees and trees with irregular shapes. To conclude, we developed three novel features based on UAV-lidar data that are especially suitable for individual tree crown delineation in dense mangrove forests. Using these features, we demonstrated that Convolutional Neural Networks can achieve high performance. We hope that our methods will facilitate various mangrove forest studies that rely on accurate individual tree crown delineation.

Validation of beyond visual-line-of-sight drone photogrammetry for terrain and canopy height applications

Aerial drones enable detailed three-dimensional terrain and vegetation canopy reconstructions supporting a diverse set of environmental monitoring applications. Small drone-derived photogrammetric datasets (<5 km2) often extend field-scale observations and enable calibration and validation for satellite-based mapping. Applications usually require data over much larger areas or in remote areas such as the Arctic, yet large-area drone datasets are rare due to aviation regulations usually stipulating that missions adhere to within-visual-line-of-sight (VLOS) conditions. Terrain and vegetation canopy reconstruction based on longer-range drone missions requires evaluation to justify future beyond-visual-line-of-sight (BVLOS) data initiatives. This work analyzed BVLOS drone photogrammetric datasets for terrain and vegetation height applications along a forest-tundra ecotone (FTE) spanning 396 km in northern Canada. Specifically, we: 1) performed validation assessments of derived terrain models and canopy height models, and 2) investigated remote sensing applications supported by BVLOS data in a FTE experiencing the effects of climate change. The BVLOS data covered in excess of 550 km2, increasing conventional drone data coverage by two orders of magnitude. Elevation validation indicated that strong agreement with GNSS benchmarks, Airborne Laser Scanning (ALS) and VLOS drone data (0.15–0.19 m RMSE) is possible, and that vertical RMSEs <2x the image spatial resolution is achievable. To validate vegetation canopy height modelling capabilities, comparisons with NASA's Land, Vegetation, and Ice Sensor (LVIS) full-waveform ALS observations indicated satisfactory agreement for height metrics (R2 = 0.7 to 0.8, RMSEs <1.0 m). Canopies were reconstructed at fine spatial scales, which in turn supported a unique comparison of vegetation height distributions between six land-cover classes and ecoregions along the FTE, and identified discrepancies in how national and international canopy height models represented the FTE. Detected terrain subsidence, including retrogressive thaw slumping and ice-wedge degradation, illustrated how climate-driven permafrost thaw is modifying landscapes and placing infrastructure at increased risk, which requires intensive monitoring. These applications demonstrated how BVLOS-derived data enables broader scale assessments of regional vegetation structure and terrain changes across larger spatial extents than previously possible with VLOS drones. Mapping opportunities will arise in the environmental remote sensing discipline as drone systems are increasingly operated beyond visual range.

Mapping global mangrove canopy height by integrating Ice, Cloud, and Land Elevation Satellite-2 photon-counting LiDAR data with multi-source images

Large-scale and precise measurement of mangrove canopy height is crucial for understanding and evaluating wetland ecosystems' condition, health, and productivity. This study generates a global mangrove canopy height map with a 30 m resolution by integrating Ice, Cloud, and Land Elevation Satellite-2 (ICESat-2) photon-counting light detection and ranging (LiDAR) data with multi-source imagery. Initially, high-quality mangrove canopy height samples were extracted using meticulous processing and filtering of ICESat-2 data. Subsequently, mangrove canopy height models were established using the random forest (RF) algorithm, incorporating ICESat-2 canopy height samples, Sentinel-2 data, TanDEM-X DEM data and WorldClim data. Furthermore, a global 30 m mangrove canopy height map was generated utilizing the Google Earth Engine platform. Finally, the global map's accuracy was evaluated by comparing it with reference canopy heights derived from both space-borne and airborne LiDAR data. Results indicate that the global 30 m resolution mangrove height map was found to be consistent with canopy heights obtained from space-borne (r = 0.88, Bisa = −0.07 m, RMSE = 3.66 m, RMSE% = 29.86 %) and airborne LiDAR (r = 0.52, Bisa = −1.08 m, RMSE = 3.39 m, RMSE% = 39.05 %). Additionally, our findings reveal that mangroves worldwide exhibit an average height of 12.65 m, with the tallest mangrove reaching a height of 44.94 m. These results demonstrate the feasibility and effectiveness of using ICESat-2 data integrated with multi-source imagery to generate a global mangrove canopy height map. This dataset offers reliable information that can significantly support government and organizational efforts to protect and conserve mangrove ecosystems.

Improving the Estimation of Structural Parameters of a Mixed Conifer–Broadleaf Forest Using Structural, Textural, and Spectral Metrics Derived from Unmanned Aerial Vehicle Red Green Blue (RGB) Imagery

Forest structural parameters are crucial for assessing ecological functions and forest quality. To improve the accuracy of estimating these parameters, various approaches based on remote sensing platforms have been employed. Although remote sensing yields high prediction accuracy in uniform, even-aged, simply structured forests, it struggles in complex structures, where accurately predicting forest structural parameters remains a significant challenge. Recent advancements in unmanned aerial vehicle (UAV) photogrammetry have opened new avenues for the accurate estimation of forest structural parameters. However, many studies have relied on a limited set of remote sensing metrics, despite the fact that selecting appropriate metrics as powerful explanatory variables and applying diverse models are essential for achieving high estimation accuracy. In this study, high-resolution RGB imagery from DJI Matrice 300 real-time kinematics was utilized to estimate forest structural parameters in a mixed conifer–broadleaf forest at the University of Tokyo Hokkaido Forest (Hokkaido, Japan). Structural and textual metrics were extracted from canopy height models, and spectral metrics were extracted from orthomosaics. Using random forest and multiple linear regression models, we achieved relatively high estimation accuracy for dominant tree height, mean tree diameter at breast height, basal area, mean stand volume, stem density, and broadleaf ratio. Including a large number of explanatory variables proved advantageous in this complex forest, as its structure is influenced by numerous factors. Our results will aid foresters in predicting forest structural parameters using UAV photogrammetry, thereby contributing to sustainable forest management.