Using airborne LiDAR and enhanced-geolocated GEDI metrics to map structural traits over a Mediterranean forest

Using airborne LiDAR and enhanced-geolocated GEDI metrics to map structural traits over a Mediterranean forest Aaron Cardenas-Martinez a,* , Adrian Pascual b , Emilia Guisado-Pintado a , Victor Rodriguez-Galiano a a Departamento de Geografía Física y An´ alisis Geogr´ afico Regional, Universidad de Sevilla, 41004, Seville, Spain b Department of Geographical Sciences, University of Maryland, College Park, MD, United States ARTICLE INFO Dataset link: Using airborne LiDAR and enhanced-geolocated GEDI metrics to map structural traits over a Mediterranean forest (Original data) Keywords: Spaceborne LiDAR Ecological mapping Geolocation GEDI Forest structure ABSTRACT The estimation of three-dimensional (3D) vegetation metrics from space-borne LiDAR allows to capture spatiotemporal trends in forest ecosystems. Structural traits from the NASA Global Ecosystem Dynamics Investigation (GEDI) are vital to support forest monitoring, restoration and biodiversity protection. The Mediterranean Basin is home of relict forest species facing the consequences of intensified climate change effects and whose habitats have been progressively shrinking over time. We used two sources of 3D-structural metrics, LiDAR point clouds and full-waveform space-borne LiDAR from GEDI to estimate forest structure in a protected area of Southern Spain, home of relict species in jeopardy due to recent extreme water-stress conditions. We locally calibrated GEDI spaceborne measurements using discrete point clouds collected by Airborne Laser Scanner (ALS) to adjust the geolocation of GEDI waveform metrics and to predict GEDI structural traits such as canopy height, foliage height diversity or leaf area index. Our results showed significant improvements in the retrieval of ecological indicators when using data collocation between ALS point clouds and comparable GEDI metrics. The best results for canopy height retrieval after collocation yielded an RMSE of 2.6 m, when limited to forest-classified areas and flat terrain, compared to an RMSE of 3.4 m without collocation. Trends for foliage height diversity (FHD; RMSE =2.1) and leaf area index (LAI; RMSE =1.6 m 2 /m 2 ) were less consistent than those for canopy height but confirmed the enhancement derived from collocation. The wall-to-wall mapping of GEDI traits framed over ALS surveys is currently available to monitor Mediterranean sparse mountain forests with sufficiency. Our results showed that combining different LiDAR platforms is particularly important for mapping areas where access to insitu data is limited and especially in regions with abrupt changes in vegetation cover, such as Mediterranean mountainous forests. 1. Introduction The assessment of three-dimensional (3D) vertical vegetation structure stands as a key element in monitoring terrestrial ecosystems, where canopy height emerges as a flagship indicator for numerous monitoring ecosystem strategies, modelling studies and environmental policies (Bastos et al., 2022; Li et al., 2023). For instance, its significance extends to the estimation of aboveground biomass (AGB), which is a key parameter in the assessment and modelling of global carbon fluxes (Dubayah et al., 2022; Friedlingstein et al., 2022; Ma et al., 2023). Additionally, canopy height plays a pivotal role in characterizing habitat structural heterogeneity as an important factor in explaining biodiversity spatial patterns (Hakkenberg et al., 2023; Marselis et al., 2022; Torresani et al., 2023). Endemic forests represent one of the global biodiversity hotspots and must-preserved ecosystems (Delavaux et al., 2023), but climate change and human pressure are jeopardizing the capability of species to adapt fast enough to resist disturbances due to stand replacement or prolonged heat waves (Anderegg et al., 2015; Hartmann et al., 2018). In the Mediterranean basin, the landscape is undergoing transformations driven by droughts, extreme heat episodes and increasingly recurrent wildfires, impacting carbon fluxes and threatening the habitats of endemic species (Grünig et al., 2023; Moreira et al., 2011; Ruffault et al., 2020). Baseline forest maps over these irreplaceable biodiversity ecosystems could help trace the effects of climate and land-cover change (Goetz et al., 2022; Harris et al., 2021; Potapov et al., 2021; Schimel et al., 2015). * Corresponding author. E-mail address: [email protected] (A. Cardenas-Martinez). Contents lists available at ScienceDirect Science of Remote Sensing journal homepage: www.sciencedirect.com/journal/science-of-remote-sensing https://doi.org/10.1016/j.srs.2025.100195 Received 22 October 2024; Received in revised form 10 January 2025; Accepted 10 January 2025 Science of Remote Sensing 11 (2025) 100195 Available online 12 January 2025 2666-0172/© 2025 Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license ( http://creativecommons.org/licenses/by-nc-nd/4.0/ ). Accurately monitoring and understanding all the complexity of terrestrial ecosystem processes, dynamics, and vulnerabilities, as well as defining successful management strategies, largely depend on the availability of timely and high-resolution data about 3D vegetation structure parameters (Grassi et al., 2017; Xu et al., 2021). The evolution of remote sensing tools, spanning from optical and radar images to laser scanners, has significantly enhanced the capability to map forest and tree attributes (Marvin et al., 2016). Aboveground metrics from LiDAR surveys are solid descriptive parameters of vegetation vertical profile and robust proxies to estimate forest structure and AGB at multiple scales (Asner et al., 2014; Beland et al., 2019; Bergen et al., 2009). Moreover, canopy cover metrics and other ecological metrics such as Leaf Area Index (LAI), Plant Area Index (PAI) or Foliage Height Diversity (FHD) are being increasingly used to report climate change effects on vegetation and to inform on forest dynamics (Rishmawi et al., 2021; Schneider et al., 2020; Wang et al., 2022). These structural traits, along with many other vegetation metrics, have been estimated globally using space-borne LiDAR technology through the NASA GEDI mission (Dubayah et al., 2020, 2022). Over the past four years, the GEDI lasers have profiled forest canopies consistently across tropical, subtropical and temperate forests to derive forest canopy height, terrain conditions and the vertical distribution of forests (i.e. GEDI laser footprint data measurements in the completed 2019/23 period exceed the 25 billion). The consolidation of the GEDI mission towards 2030 is a major support for forest ecology and climate science (Goetz et al., 2022). The calibration and validation of GEDI ecological indicators find support in data crossovers of GEDI tracks and high-resolution 3D data, suitable “laboratories” to compare GEDI structural traits and ecological indexes to similar, and sometimes comparable, laser-based metrics derivable from point cloud data either airborne-collected (e.g. Li et al., 2023; Pascual and Guerra-Hernandez, 2023) or terrestrial-based (Calders et al., 2020). Studies on the comparison between GEDI full waveform and ALS point clouds derived metrics can be split into two groups, considering whether they have applied geolocation correction methods or not (Roy et al., 2021; Tang et al., 2023). In sparse, open ecosystems, some studies have used the GEDI simulator (Hancock et al., 2019) to correct the geolocation of on-orbit GEDI measurements through ALS, helping to reduce the 10-m (1-sigma) horizontal uncertainty in the version 2 product (Li et al., 2023; Pascual et al., 2023). However, several studies have omitted this critical step, acknowledging that uncertainty in their results (e.g., Dhargay et al., 2022; Huettermann et al., 2022; Puletti et al., 2020; Wang et al., 2022). The cost of missing proper geolocation correction methods of GEDI measurements is higher in sparse, open forests where a substantial proportion of the ground may be exposed. In these low-cover ecosystems, a horizontal offset of a few meters can significantly impact the accurate retrieval of forest height vertical profiles and cover metrics (e.g., Dorado-Roda et al., 2021; Potapov et al., 2021). Assessing GEDI metrics over sparse forests is important as the GEDI instrument was designed for measuring canopy vertical profiles in closed-canopy ecosystems (conditions of 95%–98% canopy cover, as stated in Dubayah et al., 2020) and not for the monitoring of sparse forest ecosystems. The south of the Iberian Peninsula, within the Mediterranean context, is an interesting mosaic of discontinuous woody vegetation structures that vary greatly in vertical and spatial arrangement and change due to climatic, ecological and management impacts (Gonz´ alez-´ Avila et al., 2023). Moreover, Spain has a consolidated ALS survey program with a high potential for current and future calibrations and validations of GEDI science mission products in sparse forests, among other forest types (Pascual et al., 2023). Most of the studies comparing GEDI to laser point clouds were mostly focused on validating elevation and vegetation relative height (RH) metrics and less, especially over sparse forests, on other ecological indicators i.e., LAI, PAI or FHD. In this study, we aim to assess the capability of on-orbit GEDI data to estimate structural traits in mountainous Mediterranean sparse forests by comparing trends between on-orbit GEDI and simulated GEDI waveforms derived from ALS. By using high-resolution ALS data to correct the geolocation of GEDI waveforms, we explored the relations between structural traits of vegetation, highlighting the role of enhancing the geolocation of GEDI. This approach seeks to offer insights into the uncertainty of GEDI metrics due to geolocation issues in these ecosystems. Other specific objectives addressed in this study are: (1) to explore the effects of sources of uncertainty such as land-cover and topography in the estimations; and (2) to produce wall-to-wall maps of forest structure using ALS surveys as vectors to assess GEDI’s structural traits dominance at the landscape level. 2. Material and methods 2.1. Study area The study was conducted in a Mediterranean forest located in Southern Spain (Malaga province in Andalusia, Spain; 36◦44 ′ N, 4◦59 ′ W). The selected study site was Sierra de las Nieves (SN), a 230-km 2 reservoir of biodiversity protected as a National Park in 2021 (Fig. 1). This area hosts a remarkable diversity of plant species, with up to 1387 taxa (Cabezudo et al., 2022), 79 of them endemic to this region. Among them, the most distinctive species is the Abies pinsapo Boiss., which has in SN its largest population (~5800 ha). The A. pinsapo is a relict species from the Tertiary characterized by its pyramidal shape which can reach up to 30 m in top-of-canopy, and whose presence is typically limited to north-facing slopes over 1000 m above sea level. Here, the species finds the optimal conditions of humidity and temperature: the mean annual temperature in SN is ~11 ◦C and the annual precipitation reaches 1400 mm (M´ endez-Cea et al., 2023), with intense summer droughts from June to October. Below 1000 m above sea level, A. pinsapo mingles with Mediterranean conifers such as Pinus halepensis Mill. and Pinus pinaster in the North-East (Linares et al., 2011). Other species also present in SN include Quercus faginea and Junniperus communis, and understory species such as Ulex perviflorus, Rubus ulmifolius and Salvia rosmarinus. Topography in SN is particularly challenging for the retrieval of vegetation structural traits due to the presence of very steep slopes and canyons that exacerbate the complexity of retrieving accurate profiles of forest vegetation. The structural complexity of the canopy, with large variations in tree species heights and the presence of a dense understory, must also be considered. These conditions represent a challenge for the characterization of structural traits of vegetation with remote sensing techniques and enabled to test the usability of GEDI in complex Mediterranean environments (Dorado-Roda et al., 2021). 2.2. Airborne laser scanning data ALS data acquisition was performed in the northeast of SN, over a mixed-forest area known as “Puerto Saucillo”, characterized by hosting one of the main A. pinsapo forests in the southern Iberian Peninsula (M´ endez-Cea et al., 2023). The altitude of Puerto Saucillo (approximately 1000–1100 m above sea level) marks the transitional zone between the southwestern areas of the study site, where A. pinsapo dominates (Navarro-Cerrillo et al., 2022), and the areas where it is found in mixed forests alongside pine species such as P. halepensis y P. pinaster (de G´ alvez-Monta˜ nez et al., 2024). This makes the study area the lowest-altitude zone within the National Park where A. pinsapo is present (Linares et al., 2011) and a representative area of its forests. The area (10.96 km 2 ) was surveyed in February 2020 using the Leica ALS60 laser scanner (Leica Geosystems AG, Heerbrugg, Switzerland) mounted on a Cessna-337 aircraft. Flight altitude was set to 300 m and 46.3 m/s as nominal flight speed to ensure optimal optical coverage over steep slopes. The maximum scan angle of ±9◦from nadir and about 30% flight strip overlap led to an average point density of 7.7 pt m −2 (Table 1). The resulting ALS data accuracy was 30 cm in the horizontal and 15 cm in the vertical. A. Cardenas-Martinez et al. Science of Remote Sensing 11 (2025) 100195 2 The acquired laser point cloud data was processed using Terrascan (Terrasolid, 2022) and the lidR v3.1. package (Roussel et al., 2020) available in the R statistical software (R Core Team, 2022). Isolated LiDAR points were classified by identifying points with fewer neighbours within a search radius of 5 m, while low points were classified by identifying individual points or groups of points lower than a threshold of 0.5 m within a 2D radius of 5 m. The lidR package was used to identify ground returns and create the digital terrain model (DTM) following standard routines in ALS-based forest inventory (see Guerra-Hern´ andez and Pascual, 2021; Pascual et al., 2020). The Cloth Simulation Filter (Zhang et al., 2016) was applied to derive terrain elevation. Then, an inverse distance weighted algorithm was used to create a 1-m resolution DTM and a fine-grained slope map. The DTM was used to normalize the ALS point clouds to above-ground heights and produce a 0.5-m resolution canopy height model (CHM) (Fig. 2). 2.3. GEDI data GEDI data within the study area was retrieved using the ALS coverage to select co-registered GEDI footprints. Footprint variables included terrain and forest canopy height metrics (GEDI L2A, Dubayah et al., 2021a), canopy cover and density metrics (GEDI L2B, Dubayah et al., 2021b) and the estimates of aboveground biomass density (AGBD) included in the L4A product (GEDI L4A, Dubayah et al., 2022). The protocol for the selection of high-quality GEDI footprints in this study was as follows: waveform fidelity in the selected footprints exceeded 0.95 and was greater than canopy cover. The selection of the GEDI L2A algorithm for ground finding was optimized for each footprint. Footprints were filtered out if the absolute differences between the elevation of the center of the lowest mode relative to reference ellipsoid and the interpolated elevation of the TanDEM-X global DTM used in the GEDI mission exceeded 50 m. We imposed a maximum threshold for L4A biomass estimates (500 Mg ha −1 ) to remove outliers passing the filters due to dense fog and rugged topography that challenge the optimal retrieval of GEDI elevation metrics. After the quality filtering, 862 GEDI footprints were selected. These footprints are contained within 10 GEDI tracks. 2.4. The GEDI simulator The GEDI waveform simulator tool was developed for the pre-launch calibration of GEDI and was used for the calibration of GEDI L4A models (Duncanson et al., 2022). A comprehensive description of the simulator can be found in Hancock et al. (2019). Briefly, the simulator operates by generating virtual waveforms for given footprint locations using discrete point cloud data. In this study, structural metrics were calculated with Fig. 1. Overview of the study area showing different forest types and conditions captured with the airborne LiDAR survey mostly over endemic forest sites dominated by Abies pinsapo and Pinus Halepensis. Table 1 Airborne Laser Scanning sensor specifications and flight parameters. Flight date February 2020 Sensor Leica ALS60 Surveyed area (km 2 ) 10.96 Flight altitude above ground level (m) ~300 Beam divergence (mrad) 0.15 Wavelength (nm) 1064 Overlap (%) ~30 FOV (◦) 20 Point density (pts m −2 ) 7.7 A. Cardenas-Martinez et al. Science of Remote Sensing 11 (2025) 100195 3 respect to ground in full GEDI-like waveform simulation. It should be noted that RH were determined in relation to ground elevation, derived directly from ALS data. This calculation used the center of gravity of points classified as ground returns, rather than finding ground elevation from a Gaussian fit or the lowest inflection point fit to the simulated waveform, as described in Duncanson et al. (2022). This approach allowed mitigating uncertainties related to the interactions between the RH signal and ground identification during the simulation, especially in areas with steep slopes (Liu et al., 2021). 2.4.1. Geolocation of GEDI high-quality footprints The calculation of the offset between GEDI footprints and ALS data followed the approach presented in Blair and Hofton (1999) and implemented in the CollocateWaves tool as part of the GEDI simulator (see Hancock et al., 2019 and the simulator instructions). This method uses the Pearson correlation to find the best affine transformation in X, Y and Z to align the large-footprint GEDI dataset to a small-footprint ALS dataset. In our study, the collocation of GEDI footprints was performed using the following parameter settings as starting point to find the best transformations: geoerror switch with 20 m of expected error and 0.5 m of correlation distance; check cover to remove the footprints with less than 66% ALS coverage; and a GEDI beam sensitivity of at least 0.9 to perform the collocation. The output of the simulator are waveforms for each footprint and three-dimensional correction factor by intersecting orbit to correct the geolocation of the footprints (Fig. 3). To measure the impact of geolocation correction we compared pairs of RH metrics (on-orbit versus simulated GEDI) for RH98 and two GEDI L2B metrics. RH98 was used as a proxy of canopy height, being one of the GEDI L2A RH metrics used in GEDI L4A models due to its importance in aboveground biomass density estimation (i.e., this predictor is used in GEDI biomass models over Europe and many World regions as discussed by Kellner et al. (2023). The GEDI L2B metrics included in our study were FHD and LAI. These metrics are extracted from each GEDI waveform and are based on the directional gap probability profile derived from the L1B waveform (Tang and Armston, 2019). Their inclusion in our study reflects their role as structural traits that represent different main attributes of forest structure. FHD describes the distribution of foliage density across vertical canopy layers (MacArthur and MacArthur, 1961; Valbuena et al., 2012) and serves as an indicator of the vertical stratification and structural complexity of the canopy (Atkins et al., Fig. 2. Overview of the research experiment using GEDI high-quality on-orbit data and airborne LiDAR data. The resulting LiDAR point cloud data was used to map canopy height and slope terrain. A. Cardenas-Martinez et al. Science of Remote Sensing 11 (2025) 100195 4 2023). Meanwhile, LAI is defined as the projected leaf area within a canopy per horizontal ground area (Asner et al., 2003), providing a measure of canopy area and density (Atkins et al., 2023). It should be noted that while on-orbit GEDI measures PAI (projected plant area within a canopy per horizontal ground area), it is replaced in the GEDI simulator by LAI. Nevertheless, both metrics are closely related, given that the trunk and branches barely contribute to the total plant area surface (Kucharik et al., 1998). Taking this into consideration, we treated both metrics as comparable, as done previously by Pimmasarn et al. (2020) and Huettermann et al. (2023). Furthermore, although the GEDI Simulator allows for the calculation of LAI by height layers, this study considered the measurement of LAI for the entire vertical column. The widespread use of FHD and LAI in studies focused on the estimation of structural and functional traits of vegetation is well-documented for both ALS (e. g. Schneider et al., 2017; Zheng et al., 2021, 2022) and for GEDI (e. g. Boucher et al., 2020; Dhargay et al., 2022; Hirschmugl et al., 2023; Schneider et al., 2020). This broad applicability allowed us to assess GEDI’s performance for both traits in comparison with other ecosystems. The geolocation enhancement was evaluated considering land-cover heterogeneity and terrain steepness in SN. We used the 10-m V200 landcover product from the European Space Agency (Zanaga et al., 2022) now operational in GEDI (L4B biomass, Dubayah et al., 2023) to classify footprints into forests, shrublands and grasslands and the NASA Shuttle Radar Topography Mission (SRTM) digital elevation model to calculate terrain slope. 2.4.2. Simulation of GEDI metrics at landscape-level The GEDI simulator was also used to simulate GEDI-like waveform metrics over wall-to-wall tiles of ALS data. The surveyed area using ALS was gridded into 25-m tiles (14,575 tiles) to simulate waveforms and predict on-orbit conditions for the study area. The prediction was achieved by establishing relationships between on-orbit and simulated conditions through the comparison of values for the same GEDI metrics (RH98, FHD, and LAI). These relationships were captured within the high-quality GEDI footprints used for the analyses. 2.5. Structural traits using ALS data To further compare GEDI products, we used ALS-derived traits to show differences in data distribution and issues when predicting GEDI structural traits over complex forest structures in steep conditions. Three canopy-related structural traits widely used in describing forest structural diversity and measurable through ALS (e.g., Gelabert et al., 2020; Schneider et al., 2017; Zheng et al., 2021) were applied in this study. We selected the 98th percentile height (P98; indicative of canopy height), FHD (as previously described in Section 2.4.1 as indicative of the vertical distribution of canopy layering) and LAI (also described in Section 2.4.1 as the projected surface area of plant material per unit ground area), structural traits representative of height, canopy structural complexity and vegetation density respectively (Atkins et al., 2023; Valbuena et al., 2020). The ALS P98 was compared to GEDI RH98 and used, along with FHD and LAI, in the subsequent spatial prediction of GEDI structural traits at the landscape level (See Section 2.6.). To derive P98, the ALS point cloud was first filtered to retain only the first returns using the filter_poi function, and the 98th quantile height was then calculated using the grid_metrics function available in lidR. To estimate FHD, we used the function FHD published in leafR R package v0.3.5 (Almeida et al., 2021). FHD was retrieved from abundances considered as per-voxel relative leaf area density (LAD) values by applying the Fig. 3. Figure of 3 sub-figures showing non-ALS-collocated positions (on-orbit) and ALS-collocated positions with the corrections after using the simulator (enhanced on-orbit). The ALS-based canopy height model is presented in the background. The 2D mean on-orbit geolocation correction was 8m. A. Cardenas-Martinez et al. Science of Remote Sensing 11 (2025) 100195 5 Shannon–Weiner diversity function, as described in MacArthur and MacArthur (1961): FHD = − ∑ i pi*ln pi(1) where pi is the proportion of the total vegetation (in this case, ALS returns) that is in the ith layer, structured in 2-m vertical intervals. Finally, LAI was calculated in leafR using the method proposed by Almeida et al. (2019), based on the application of the MacArthur-Horn equation (MacArthur and Horn, 1969) to LAD for each 1-m voxel populated with vegetation: ⎧ ⎪ ⎨ ⎪ ⎩ LAD =ln(pulsesin pulsesout)*1 K LAI =∑LAD (2) where pulsesin and pulsesout represent the pulses that entered each voxel and passed through it, respectively. K represents the Beer-Lambert Law extinction coefficient and depends primarily on the foliage distribution and orientation and the thickness of leaves and forest canopy (Kamoske Fig. 4. Workflow diagram showing all steps in the methodology applied. From top to down: data collection using UAV LiDAR, retrieval of GEDI observations, filtering of ALS point clouds and generation of products, collocation of GEDI footprints using ALS to correct geolocation error, ALS-simulation of GEDI metrics at landscape scale and evaluation of gridded maps for three indicators: forest canopy height, Foliage Height Diversity (FHD) and Leaf Area Index (LAI). A. Cardenas-Martinez et al. Science of Remote Sensing 11 (2025) 100195 6 et al., 2019; Weiss et al., 2004). K value should be adjusted in order to calibrate estimated LAI into an independent LAI measurement as recommended by Almeida et al. (2019), based on the use of field measurements. Following these considerations, we employed a limited set of six LAI field measurements collected in June 2023 using a LAI-2200C Plant Canopy Analizer during a field campaign to provide a calibration for K. The total explained variance (R 2 ) was 0.757 between LAI ALS and LAI field (see Supplementary Figure s1). Taking into account that this formula assumes that each LiDAR pulse is vertically incident, we imposed a maximum of 10◦off-nadir view in the point cloud (~75% of the points had less than 5◦off-nadir view), considering the recommended values in Liu et al. (2018) and the limitations of our LiDAR survey. Traits were obtained for the entire ALS flight area at a 25-m resolution corresponding with the GEDI footprint. 2.6. Accuracy assessment For the assessment of enhanced geolocation, we used linear models and calculated the absolute and relative root mean squared error (RMSE), total explained variance (R 2 ), and bias between distributions of simulated GEDI and on-orbit GEDI metrics (i.e., RH98, FHD and LAI). The ALS-based benchmark metrics at shot level were also assessed using on-orbit GEDI as reference. For the estimation of RH98 and FHD, all available on-orbit GEDI footprints were used. Meanwhile, since LAI is not only a structural but also a biophysical trait, it is more sensitive to changes due to climate and forest disturbances (Heiskanen et al., 2013; Wu et al., 2018). Therefore, linear models derived from the GEDI footprints corresponding to the closest years (2019 and 2020) to the ALS flight were tested for estimating LAI. The spatial predictions at 25-m resolution using ALS-based estimates and predicted on-orbit GEDI values derived from linear models using simulated GEDI and on-orbit GEDI footprints were similarly assessed at landscape level. To visualize tendencies of GEDI structural traits in the study area, values of forest canopy height - RH98, FHD and LAI were rescaled between 0 and 1, using a min-max normalization approach to capture the dominance of each indicator and its spatial variation across SN. We used an RGB colour composite of the structural traits. Thereby, red areas were defined as values of RH98 >0.5, FHD <0.5 and LAI <0.5; green areas as RH98 <0.5, FHD >0.5 and LAI <0.5; and blue areas as RH98 <0.5, FHD <0.5 and LAI >0.5. Finally, small white areas resulting from the combination of high values for each structural traits were defined as RH98, FHD and LAI >0.75. A workflow diagram summarizing all the steps followed in the methodology is shown in Fig. 4. 3. Results 3.1. Effect of geolocation correction to estimate GEDI canopy height Distributions of ALS-collocated and non-collocated (i.e., on-orbit positions of the footprints) showed Pearson correlation values above 0.85 for 98.7% of the footprints (See Table 2). The average 2D distance between the pair of enhanced and non-enhanced geolocated footprints was 9.4 m (median), 3.4 m (mode, the lower offset correction for the intersecting GEDI L1B tracks) and 8.0 m (mean). To compute these summaries, we averaged the geolocation offset of high-quality footprints intersecting the study area. The ALS-collocation of GEDI observations reduced the RMSE in forest canopy height (RH98) by almost a meter and increased the R 2 from 0.52 to 0.62 (Table 3). Thus, RMSE in GEDI RH98 using the ALScollocated method was 4.36 m when comparing simulated vs on-orbit GEDI. Without geolocation correction, the value increased to 5.3 m (Table 3, Figure s2). Given that many studies have compared ALS height percentiles to GEDI energy-based relative height metrics, it is relevant to show that, regardless of the geolocation correction method, comparing ALS P98 to GEDI RH98 produced large discrepancies. Our values for RMSE and bias were similar, above 7 m and 5 m, respectively, in both situations. The quantile distributions showed an improvement in the alignment between simulated GEDI and on-orbit GEDI on the right-side (i.e., above the 95th quantile) of the RH98 spectrum when using data collocation (Fig. 5). The comparison of GEDI RH98 versus ALS P98 confirmed the afore-described null effect and revealed a shorter domain of ALS percentiles compared to both simulated and on-orbit GEDI. 3.2. Impact of land-cover and slope on GEDI canopy height estiomation More than 90% of the selected footprints were identified as forested areas according to the 10-m ESA land-cover product implemented in GEDI. Collocated waveforms over forests showed systematically lower errors in canopy height estimation compared to footprints ranging over shrublands or grasslands (Fig. 6). Specifically, ALS-GEDI collocation over forested areas led to a decrease in RMSE from 5 m to 4.1 m. Meanwhile, for grasslands and shrublands, the RMSE decreased from 7.6 m to 6.3 m. Slope estimates from the SRTM global product highlight the substantial improvement in GEDI accuracy for forests in flat conditions (slope below 10◦). Here, the RMSE was down to 2.6 m and the R 2 reached 0.83 when using the ALS-GEDI collocation (Fig. 7). Towards moderate (10–30◦) and steep conditions (above 30◦) the error in the retrieval of canopy height was 4.2 m and 5 m, respectively. The absence of GEDI geolocation correction showed a decline in R 2 by 10 points for flat conditions and up to 20 points for steep conditions, where the RMSE reached 6.4 m, marking a 20% increase compared to ALS-collocated conditions (5.1 m) as shown in Fig. 7. It is noteworthy that most observations (~70%) for this study area range in moderate terrain steepness (10–30◦). 3.3. Estimation and wall-to-wall mapping of GEDI structural traits The agreement between on-orbit and simulated GEDI estimates for Table 2 Geolocation offsets (ALS versus on-orbit GEDI). The numbers of each GEDI track ID indicate the Julian Date and specific hour of acquisition. GEDI L1B Track ID dX Easting dY Northing dZ Elevation Pearson correlation GEDI footprints (n) 2019170142546 8.00 −5.00 0.00 0.921 197 2019338194614 8.00 2.00 0.00 0.879 76 2020052055355 2.12 −2.88 0.13 0.667 4 2020060024632 6.00 11.00 0.00 0.936 184 2020188065523 −5.29 −12.24 −0.27 0.940 57 2020314215842 2.65 2.10 0.23 0.943 207 2021197193255 8.00 −7.00 0.00 0.858 49 2022021233617 11.15 13.41 0.10 0.372 7 2022144225451 −1.35 0.89 0.39 0.896 75 2022336121243 1.00 −2.00 0.00 0.886 6 Table 3 Performance of GEDI at estimating forest canopy height showing the effect of GEDI geolocation correction using the GEDI simulator. Fitting statistics were computed for pairs of on-orbit and simulated relative height 98 (RH98) and height percentile 98 for ALS distributions. Structural indicator Fitting Statistic ALS-collocated GEDI No geolocation correction GEDI real vs GEDI sim GEDI real vs ALS GEDI real vs GEDI sim GEDI real vs ALS Relative height 98 Canopy height (RH98, m) R 2 0.624 0.383 0.519 0.354 RMSE 4.348 7.112 5.253 7.275 RMSE (%) 12.2 30.6 21.8 30.7 Bias −2.316 −5.273 −3.179 −5.338 Footprints (n) 862 862 850 850 A. Cardenas-Martinez et al. Science of Remote Sensing 11 (2025) 100195 7 FHD was weaker compared to canopy height, as expected (Table 4). The difference between collocated and non-collocated distributions of GEDI FHD estimation was small, and both distributions showed R 2 values below 0.6 for GEDI-to-GEDI trends, which doubled the values in the GEDI-to-ALS estimation of FHD (See Figure s3). On-orbit versus simulated GEDI distributions for LAI showed poor predictive capability and an important mismatch between simulated data and real on-orbit measurements (Table 4, Figure s4). This mismatch was evident through R 2 values below 0.25 in all cases, even when compared to ALS-based estimation. The LAI prediction was performed using GEDI footprints in 2019 that outperformed accuracies computed for the year 2020: R 2 = 0.237 and 0.098, respectively, both using ALS-collocated GEDI footprints. GEDI structural trait distributions were also assessed to measure the impact of land-cover and slope on the estimations. Similar to canopy height, collocated waveforms over forested areas consistently showed lower errors for FHD compared to the rest of the footprints (See Figure s5). Furthermore, the impact of the slope on GEDI FHD estimates over forested areas varied significantly depending on the collocation of the waveforms (Figure s6). FHD estimates using the collocated waveforms showed a good performance for flat areas (R 2 =0.657), decreasing moderately for slopes greater than 10◦(R 2 =0.56 for moderate slopes and R 2 =0.6 for steep slopes). Nevertheless, although the RMSE remained similar (~2), FHD distributions using the non-collocated waveforms showed greater variation, ranging from R 2 of 0.577 for flat areas to an R 2 of 0.404 for steep conditions. On the other hand, more than 96% of GEDI footprints used for LAI estimation were considered forested areas, making the land-cover analysis less consistent due to the reduced number of observations in shrublands and grasslands (Figure s7). The results showed that the collocation of GEDI waveforms systematically improved the performance of the LAI estimation in all land-cover classes when comparing both simulated GEDI and ALS to onorbit GEDI. Here, the best performances were observed when comparing collocated on-orbit and simulated GEDI LAI estimates, with R 2 =0.235 in forested areas and R 2 =0.291 in non-forested conditions. In turn, the impact of slope on GEDI LAI estimates was lower than the land-cover and affected by the small number of footprints in flat areas and those with slopes greater than 30◦. Thus, the agreement between on-orbit and simulated GEDI was weaker for flat areas than those with moderate slopes (Figure s8). LAI estimates over moderate slopes experienced also the only improvement in the performance after applying the collocation both for simulated GEDI and ALS. Linear relationships between on-orbit and simulated GEDI footprints were used to predict GEDI structural traits for the entire study area (Figs. 8 and 9). For canopy height, we observed a systematic deviation in the trend line comparing predicted GEDI RH98 to ALS P98. The correlation was remarkably high (above 0.95), but the bias of 2.6 m confirmed what was observed in the footprints (Fig. 8). For the case of LAI or FHD, the alignment of predicted GEDIand ALS-based data distributions was still high (R 2 >0.7) although the bias in LAI was Fig. 5. Quantile distributions showing the relationship between on-orbit GEDI estimates of relative height 98, ALS-collocated GEDI estimates of enhanced geolocation and the ALS 98th height percentile. Results are presented for GEDI collocation using ALS to correct geolocation error (collocated) and no geolocation correction, keeping on-orbit positions (non-collocated). The mean 2D offset between pairs of data was 8 m. A. Cardenas-Martinez et al. Science of Remote Sensing 11 (2025) 100195 8 particularly strong, and estimations were substantially displaced from the 1:1 line. Distributions showed a cut-off value in the low end due to an overestimation of GEDI, imposing a tight constraint to describe FHD and especially LAI domains (Fig. 9). Despite this artefact, most of the observations for FHD followed the 1:1 trendline and the relative RMSE is below 20%. We used model predictions for GEDI and structural traits retrieved from ALS at 25 m to map differences. For the case of canopy height, the difference between RH98 and P98 was systematic across the study area (Fig. 8), with an overestimation of RH98 compared to ALS P98. For FHD, we observed transition areas where few patches did not show the overall underestimation of GEDI compared to ALS-based estimates of FHD (Fig. 9). The spatial layout for LAI estimates showed the highest spatial variability in the estimates: systematically high values for GEDI nonforested areas and with a canopy height value of 0 in the ALS-derived CHM product explain the largest differences, while GEDI predicted low LAI compared to ALS over dense forest areas. 3.4. Assessing the dominance of structural traits The integrated representation of forest structure at 25-m resolution using canopy height, FHD and LAI was useful in detecting transition areas showing abrupt changes in the dominance of each trait (Fig. 10). Red tones show high canopy height and low values for both FHD and LAI, corresponding to P. halepensis reaching up to 34 m tall. These pinespecific stands planted for restoration decades ago show less structural variability, especially in the upper canopies, as confirmed by low GEDI estimation values for FHD and LAI. Similarly, high values for RH98 and FHD can be observed in the northern part (mixed stands of P. halepensis and P. Pinaster). Areas represented in blue and green tones in the southern part (A. pinsapo stands) show high values of LAI and FHD. In this case, A. pinsapo forests are characterized by a high canopy density and layering, often associated with a dense understory of J. communis or R. ulmifolius. This phenomenon could be explained by the restoration efforts focussed on creating heterogeneous conditions between oldgrowth stands and younger patches. Further, observed high values for LAI in central sections of the study area correspond to P. halepensis and A. Pinsapo mixed stands. 4. Discussion Structural trait maps obtained from GEDI and ALS can provide rapid baselines for supporting forest monitoring and conservation. Actually, the careful selection of GEDI footprints for training fused products between optical data, radar data, airborne LiDAR and GEDI is a timely research topic (Francini et al., 2022; Potapov et al., 2021; Qi et al., 2025; Zhao et al., 2024). For example, Francini et al. (2022) and Potapov et al. (2021) used GEDI and Landsat to map changes in forest biomass due to forest disturbances. Other studies as Qi et al. (2019) employed simulated GEDI and TanDEM-X InSAR data to improve forest structural mapping across several mountainous and non-mountainous forested areas in the Americas. Although GEDI lasers were not designed to operate in sparse mountain forests, GEDI data have been consistently used in savannas and discontinuous vegetation before (Dorado-Roda et al., 2021; Hoffr´ en et al., 2023; Li et al., 2023). In this study, we assess the accuracy of enhanced geolocated on-orbit GEDI over a Mediterranean mountainous forest, being to our knowledge the first attempt to validate GEDI canopy height measurements in these sparse, mountainous Mediterranean environments. Our results disentangle key factors affecting the accuracy of GEDI measurements in these ecosystems. Therefore, we contribute to the GEDI state-of-the-art, adding new insights from Mediterranean ecosystems: sparse mountain forests with abrupt transitions in terms of species dominance, forest cover and terrain steepness. 4.1. Retrieval of forest structure from GEDI Numerous studies have compared GEDI metrics to ALS without addressing important issues on geolocation accuracy (Puletti et al., 2020; Rishmawi et al., 2021; Zhu et al., 2022). Directly comparing distributions of ALS height percentiles to GEDI energy-based percentiles might not be the most appropriate method to assess the GEDI Fig. 6. Scatterplots showing the accuracy of GEDI canopy height estimates using relative height 98 (RH98). On-orbit collocated and non-collocated GEDI measurements and simulated data using ALS are compared. The distribution of ALS height percentiles is also presented. Results are presented for land-cover class “forests” and combining classes “shrublands” and “grasslands” (ESA V200 10-m map). A. Cardenas-Martinez et al. Science of Remote Sensing 11 (2025) 100195 9 Goetz, S., Dubayah, R., Duncanson, L., 2022. Revisiting the status of forest carbon stock changes in the context of the measurement and monitoring needs, capabilities and potential for addressing reduced emissions from deforestation and forest degradation. Environ. Res. Lett. 17 (11), 111003. https://doi.org/10.1088/17489326/ac9c1d. Gonz´ alez-´ Avila, S., Ortega, E., Martín, B., 2023. Multiscale fragmentation of forest types in Spain. For. Ecol. Manag. 546, 121317. https://doi.org/10.1016/j. foreco.2023.121317. Grassi, G., House, J., Dentener, F., Federici, S., den Elzen, M., Penman, J., 2017. The key role of forests in meeting climate targets requires science for credible mitigation. Nat. Clim. Change 7 (3), 220–226. https://doi.org/10.1038/nclimate3227. Grünig, M., Seidl, R., Senf, C., 2023. Increasing aridity causes larger and more severe forest fires across Europe. Global Change Biol. 29 (6), 1648–1659. https://doi.org/ 10.1111/gcb.16547. Guerra-Hern´ andez, J., Pascual, A., 2021. Using GEDI lidar data and airborne laser scanning to assess height growth dynamics in fast-growing species: a showcase in Spain. Forest Ecosystems 8 (1), 14. https://doi.org/10.1186/s40663-021-00291-2. Guerra-Hern´ andez, J., Pereira, J.M.C., Stovall, A., Pascual, A., 2024. Impact of fire severity on forest structure and biomass stocks using NASA GEDI data. Insights from the 2020 and 2021 wildfire season in Spain and Portugal. Sci. Rem. Sens. 9, 100134. https://doi.org/10.1016/j.srs.2024.100134. Hakkenberg, C.R., Tang, H., Burns, P., Goetz, S.J., 2023. Canopy structure from space using GEDI lidar. Front. Ecol. Environ. 21 (1), 55–56. https://doi.org/10.1002/ fee.2585. Hancock, S., Armston, J., Hofton, M., Sun, X., Tang, H., Duncanson, L.I., Kellner, J.R., Dubayah, R., 2019. The GEDI simulator: a large-footprint waveform lidar simulator for calibration and validation of spaceborne missions. Earth Space Sci. 6 (2), 294–310. https://doi.org/10.1029/2018EA000506. Harris, N.L., Gibbs, D.A., Baccini, A., Birdsey, R.A., de Bruin, S., Farina, M., Fatoyinbo, L., Hansen, M.C., Herold, M., Houghton, R.A., Potapov, P.V., Suarez, D.R., RomanCuesta, R.M., Saatchi, S.S., Slay, C.M., Turubanova, S.A., Tyukavina, A., 2021. Global maps of twenty-first century forest carbon fluxes. Nat. Clim. Change 11 (3), 234–240. https://doi.org/10.1038/s41558-020-00976-6. Hartmann, H., Moura, C.F., Anderegg, W.R.L., Ruehr, N.K., Salmon, Y., Allen, C.D., Arndt, S.K., Breshears, D.D., Davi, H., Galbraith, D., Ruthrof, K.X., Wunder, J., Adams, H.D., Bloemen, J., Cailleret, M., Cobb, R., Gessler, A., Grams, T.E.E., Jansen, S., et al., 2018. Research frontiers for improving our understanding of drought-induced tree and forest mortality. New Phytol. 218 (1), 15–28. https://doi. org/10.1111/nph.15048. Heiskanen, J., Rautiainen, M., Stenberg, P., M˜ ottus, M., Vesanto, V.-H., 2013. Sensitivity of narrowband vegetation indices to boreal forest LAI, reflectance seasonality and species composition. ISPRS J. Photogrammetry Remote Sens. 78, 1–14. https://doi. org/10.1016/j.isprsjprs.2013.01.001. Hirschmugl, M., Lippl, F., Sobe, C., 2023. Assessing the vertical structure of forests using airborne and spaceborne LiDAR data in the Austrian alps. Rem. Sens. 15 (3), 664. https://doi.org/10.3390/rs15030664. Hoffr´ en, R., Lamelas, M.T., de la Riva, J., Domingo, D., Montealegre, A.L., GarcíaMartín, A., Revilla, S., 2023. Assessing GEDI-NASA system for forest fuels classification using machine learning techniques. Int. J. Appl. Earth Obs. Geoinf. 116, 103175. https://doi.org/10.1016/j.jag.2022.103175. Huettermann, S., Jones, S., Soto-Berelov, M., Hislop, S., 2022. Intercomparison of real and simulated GEDI observations across sclerophyll forests. Rem. Sens. 14 (9), 2096. https://doi.org/10.3390/rs14092096. Huettermann, S., Jones, S., Soto-Berelov, M., Hislop, S., 2023. Using Landsat time series and bi-temporal GEDI to compare spectral and structural vegetation responses after fire. Int. J. Appl. Earth Obs. Geoinf. 122, 103403. https://doi.org/10.1016/j. jag.2023.103403. Ilangakoon, N., Glenn, N.F., Schneider, F.D., Dashti, H., Hancock, S., Spaete, L., Goulden, T., 2021. Airborne and spaceborne lidar reveal trends and patterns of functional diversity in a semi-arid ecosystem. Frontiers in Remote Sensing 2. https:// doi.org/10.3389/frsen.2021.743320. Kamoske, A.G., Dahlin, K.M., Stark, S.C., Serbin, S.P., 2019. Leaf area density from airborne LiDAR: comparing sensors and resolutions in a temperate broadleaf forest ecosystem. For. Ecol. Manag. 433, 364–375. https://doi.org/10.1016/j. foreco.2018.11.017. Kellner, J.R., Armston, J., Duncanson, L., 2023. Algorithm theoretical basis document for GEDI footprint aboveground biomass density. Earth Space Sci. 10 (4). https://doi. org/10.1029/2022EA002516. Kucharik, C.J., Norman, J.M., Gower, S.T., 1998. Measurements of branch area and adjusting leaf area index indirect measurements. Agric. For. Meteorol. 91 (1–2), 69–88. https://doi.org/10.1016/S0168-1923(98)00064-1. Li, X., Wessels, K., Armston, J., Duncanson, L., Urbazaev, M., Naidoo, L., Mathieu, R., Main, R., 2024. Evaluation of GEDI footprint level biomass models in Southern African Savannas using airborne LiDAR and field measurements. Science of Remote Sensing 10, 100161. https://doi.org/10.1016/j.srs.2024.100161. Li, X., Wessels, K., Armston, J., Hancock, S., Mathieu, R., Main, R., Naidoo, L., Erasmus, B., Scholes, R., 2023. First validation of GEDI canopy heights in African savannas. Remote Sensing of Environment 285, 113402. https://doi.org/10.1016/j. rse.2022.113402. Linares, J.C., Delgado-Huertas, A., Carreira, J.A., 2011. Climatic trends and different drought adaptive capacity and vulnerability in a mixed Abies pinsapo–Pinus halepensis forest. Climatic Change 105 (1–2), 67–90. https://doi.org/10.1007/ s10584-010-9878-6. Liu, A., Cheng, X., Chen, Z., 2021. Performance evaluation of GEDI and ICESat-2 laser altimeter data for terrain and canopy height retrievals. Remote Sensing of Environment 264, 112571. https://doi.org/10.1016/j.rse.2021.112571. Liu, J., Skidmore, A.K., Jones, S., Wang, T., Heurich, M., Zhu, X., Shi, Y., 2018. Large offnadir scan angle of airborne LiDAR can severely affect the estimates of forest structure metrics. ISPRS J. Photogrammetry Remote Sens. 136, 13–25. https://doi. org/10.1016/j.isprsjprs.2017.12.004. Ma, L., Hurtt, G., Tang, H., Lamb, R., Lister, A., Chini, L., Dubayah, R., Armston, J., Campbell, E., Duncanson, L., Healey, S., O’Neil-Dunne, J., Ott, L., Poulter, B., Shen, Q., 2023. Spatial heterogeneity of global forest aboveground carbon stocks and fluxes constrained by spaceborne lidar data and mechanistic modeling. Global Change Biol. 29 (12), 3378–3394. https://doi.org/10.1111/gcb.16682. MacArthur, R.H., Horn, H.S., 1969. Foliage profile by vertical measurements. Ecology 50 (5), 802–804. https://doi.org/10.2307/1933693. MacArthur, R.H., MacArthur, J.W., 1961. On bird species diversity. Ecology 42 (3), 594–598. https://doi.org/10.2307/1932254. Mandl, L., Stritih, A., Seidl, R., Ginzler, C., Senf, C., 2023. Spaceborne LiDAR for characterizing forest structure across scales in the European Alps. Remote Sensing in Ecology and Conservation. https://doi.org/10.1002/rse2.330. Marselis, S.M., Keil, P., Chase, J.M., Dubayah, R., 2022. The use of GEDI canopy structure for explaining variation in tree species richness in natural forests. Environ. Res. Lett. 17 (4), 045003. https://doi.org/10.1088/1748-9326/ac583f. Marvin, D.C., Koh, L.P., Lynam, A.J., Wich, S., Davies, A.B., Krishnamurthy, R., Stokes, E., Starkey, R., Asner, G.P., 2016. Integrating technologies for scalable ecology and conservation. Global Ecology and Conservation 7, 262–275. https://doi. org/10.1016/j.gecco.2016.07.002. M´ endez-Cea, B., García-García, I., Linares, J.C., Gallego, F.J., 2023. Warming appears as the main risk of non-adaptedness for western Mediterranean relict fir forests under expected climate change scenarios. Front. Plant Sci. 14. https://doi.org/10.3389/ fpls.2023.1155441. Moreira, F., Viedma, O., Arianoutsou, M., Curt, T., Koutsias, N., Rigolot, E., Barbati, A., Corona, P., Vaz, P., Xanthopoulos, G., Mouillot, F., Bilgili, E., 2011. Landscape – wildfire interactions in southern Europe: implications for landscape management. J. Environ. Manag. 92 (10), 2389–2402. https://doi.org/10.1016/j. jenvman.2011.06.028. Næsset, E., 1997. Determination of mean tree height of forest stands using airborne laser scanner data. ISPRS J. Photogrammetry Remote Sens. 52 (2), 49–56. https://doi. org/10.1016/S0924-2716(97)83000-6. Navarro-Cerrillo, R.M., Gonz´ alez-Moreno, P., Ruiz-G´ omez, F.J., S´ anchez-Cuesta, R., Gazol, A., Camarero, J.J., 2022. Drought stress and pests increase defoliation and mortality rates in vulnerable Abies pinsapo forests. For. Ecol. Manag. 504, 119824. https://doi.org/10.1016/j.foreco.2021.119824. Oliveira, P.V.C., Zhang, X., Peterson, B., Ometto, J.P., 2023. Using simulated GEDI waveforms to evaluate the effects of beam sensitivity and terrain slope on GEDI L2A relative height metrics over the Brazilian Amazon Forest. Science of Remote Sensing 7, 100083. https://doi.org/10.1016/j.srs.2023.100083. Pascual, 2019. Using tree detection based on airborne laser scanning to improve forest inventory considering edge effects and the Co-registration factor. Rem. Sens. 11 (22), 2675. https://doi.org/10.3390/rs11222675. Pascual, A., Guerra-Hernandez, J., 2023. Correction of phenology-induced effects in forest canopy height models based on airborne laser scanning data. Insights from the deciduous mountain forests in Picos de Europa National Park in Spain. Ecol. Inf. 75, 102092. https://doi.org/10.1016/j.ecoinf.2023.102092. Pascual, A., Guerra-Hern´ andez, J., Armston, J., Minor, D.M., Duncanson, L., May, P.B., Kellner, J.R., Dubayah, R., 2023. Assessing the performance of NASA’s GEDI L4A footprint aboveground biomass density models using National Forest Inventory and airborne laser scanning data in Mediterranean forest ecosystems. For. Ecol. Manag. 538, 120975. https://doi.org/10.1016/j.foreco.2023.120975. Pascual, A., Guerra-Hern´ andez, J., Cosenza, D.N., Sandoval, V., 2020. The role of improved ground positioning and forest structural complexity when performing forest inventory using airborne laser scanning. Rem. Sens. 12 (3). https://doi.org/ 10.3390/rs12030413. Pimmasarn, S., Tripathi, N.K., Ninsawat, S., Sasaki, N., 2020. Applying LiDAR to quantify the plant area index along a successional gradient in a tropical forest of Thailand. Forests 11 (5), 520. https://doi.org/10.3390/f11050520. Potapov, P., Li, X., Hernandez-Serna, A., Tyukavina, A., Hansen, M.C., Kommareddy, A., Pickens, A., Turubanova, S., Tang, H., Silva, C.E., Armston, J., Dubayah, R., Blair, J. B., Hofton, M., 2021. Mapping global forest canopy height through integration of GEDI and Landsat data. Remote Sensing of Environment 253, 112165. https://doi. org/10.1016/j.rse.2020.112165. Puletti, N., Grotti, M., Ferrara, C., Chianucci, F., 2020. Lidar-based estimates of aboveground biomass through ground, aerial, and satellite observation: a case study in a Mediterranean forest. J. Appl. Remote Sens. 14 (4). https://doi.org/10.1117/1. JRS.14.044501. Qi, W., Armston, J., Choi, C., Stovall, A., Saarela, S., Pardini, M., Fatoyinbo, L., Papathanassiou, K., Pascual, A., Dubayah, R., 2025. Mapping large-scale pantropical forest canopy height by integrating GEDI lidar and TanDEM-X InSAR data. Rem. Sens. Environ. 318, 114534. https://doi.org/10.1016/j.rse.2024.114534. Qi, W., Lee, S.-K., Hancock, S., Luthcke, S., Tang, H., Armston, J., Dubayah, R., 2019. Improved forest height estimation by fusion of simulated GEDI Lidar data and TanDEM-X InSAR data. Remote Sensing of Environment 221, 621–634. https://doi. org/10.1016/j.rse.2018.11.035. Quiros, E., Polo, M.-E., Fragoso-Campon, L., 2021. GEDI elevation accuracy assessment: a case study of southwest Spain. IEEE J. Sel. Top. Appl. Earth Obs. Rem. Sens. 14, 5285–5299. https://doi.org/10.1109/JSTARS.2021.3080711. R Core Team, 2022. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing. A. Cardenas-Martinez et al. Science of Remote Sensing 11 (2025) 100195 16 Rishmawi, K., Huang, C., Zhan, X., 2021. Monitoring key forest structure attributes across the conterminous United States by integrating GEDI LiDAR measurements and VIIRS data. Rem. Sens. 13 (3), 442. https://doi.org/10.3390/rs13030442. Roussel, J.-R., Auty, D., Coops, N.C., Tompalski, P., Goodbody, T.R.H., Meador, A.S., Bourdon, J.-F., de Boissieu, F., Achim, A., 2020. lidR: an R package for analysis of Airborne Laser Scanning (ALS) data. Remote Sensing of Environment 251, 112061. https://doi.org/10.1016/j.rse.2020.112061. Roy, D.P., Kashongwe, H.B., Armston, J., 2021. The impact of geolocation uncertainty on GEDI tropical forest canopy height estimation and change monitoring. Science of Remote Sensing 4, 100024. https://doi.org/10.1016/j.srs.2021.100024. Ruffault, J., Curt, T., Moron, V., Trigo, R.M., Mouillot, F., Koutsias, N., Pimont, F., Martin-StPaul, N., Barbero, R., Dupuy, J.-L., Russo, A., Belhadj-Khedher, C., 2020. Increased likelihood of heat-induced large wildfires in the Mediterranean Basin. Sci. Rep. 10 (1), 13790. https://doi.org/10.1038/s41598-020-70069-z. Schimel, D., Pavlick, R., Fisher, J.B., Asner, G.P., Saatchi, S., Townsend, P., Miller, C., Frankenberg, C., Hibbard, K., Cox, P., 2015. Observing terrestrial ecosystems and the carbon cycle from space. Global Change Biol. 21 (5), 1762–1776. https://doi.org/ 10.1111/gcb.12822. Schneider, F.D., Ferraz, A., Hancock, S., Duncanson, L.I., Dubayah, R.O., Pavlick, R.P., Schimel, D.S., 2020. Towards mapping the diversity of canopy structure from space with GEDI. Environ. Res. Lett. 15 (11), 115006. https://doi.org/10.1088/17489326/ab9e99. Schneider, F.D., Morsdorf, F., Schmid, B., Petchey, O.L., Hueni, A., Schimel, D.S., Schaepman, M.E., 2017. Mapping functional diversity from remotely sensed morphological and physiological forest traits. Nat. Commun. 8 (1), 1441. https:// doi.org/10.1038/s41467-017-01530-3. Stritih, A., Seidl, R., Senf, C., 2023. Alternative states in the structure of mountain forests across the {Alps} and the role of disturbance and recovery. Landsc. Ecol. https://doi. org/10.1007/s10980-023-01597-y. Tang, H., Armston, J., 2019. Algorithm Theoretical Basis Document (ATBD) for GEDI L2B Footprint Canopy Cover and Vertical Profile Metrics. Tang, H., Stoker, J., Luthcke, S., Armston, J., Lee, K., Blair, B., Hofton, M., 2023. Evaluating and mitigating the impact of systematic geolocation error on canopy height measurement performance of GEDI. Remote Sensing of Environment 291, 113571. https://doi.org/10.1016/j.rse.2023.113571. Terrasolid, 2022. Terrascan (021.005). Torresani, M., Rocchini, D., Alberti, A., Moudrý, V., Heym, M., Thouverai, E., Kacic, P., Tomelleri, E., 2023. LiDAR GEDI derived tree canopy height heterogeneity reveals patterns of biodiversity in forest ecosystems. Ecol. Inf. 76, 102082. https://doi.org/ 10.1016/j.ecoinf.2023.102082. Urbazaev, M., Hess, L.L., Hancock, S., Sato, L.Y., Ometto, J.P., Thiel, C., Dubois, C., Heckel, K., Urban, M., Adam, M., Schmullius, C., 2022. Assessment of terrain elevation estimates from ICESat-2 and GEDI spaceborne LiDAR missions across different land cover and forest types. Science of Remote Sensing 6, 100067. https:// doi.org/10.1016/j.srs.2022.100067. Valbuena, R., O’Connor, B., Zellweger, F., Simonson, W., Vihervaara, P., Maltamo, M., Silva, C.A., Almeida, D.R.A., Danks, F., Morsdorf, F., Chirici, G., Lucas, R., Coomes, D.A., Coops, N.C., 2020. Standardizing ecosystem morphological traits from 3D information sources. Trends Ecol. Evol. 35 (8), 656–667. https://doi.org/ 10.1016/j.tree.2020.03.006. Valbuena, R., Packalen, P., Martın-Fernandez, S., Maltamo, M., 2012. Diversity and equitability ordering profiles applied to study forest structure. For. Ecol. Manag. 276, 185–195. https://doi.org/10.1016/j.foreco.2012.03.036. Vogeler, J.C., Fekety, P.A., Elliott, L., Swayze, N.C., Filippelli, S.K., Barry, B., Holbrook, J.D., Vierling, K.T., 2023. Evaluating GEDI data fusions for continuous characterizations of forest wildlife habitat. Frontiers in Remote Sensing 4. https:// doi.org/10.3389/frsen.2023.1196554. Wang, C., Elmore, A.J., Numata, I., Cochrane, M.A., Shaogang, L., Huang, J., Zhao, Y., Li, Y., 2022. Factors affecting relative height and ground elevation estimations of GEDI among forest types across the conterminous USA. GIScience \& Remote Sensing 59 (1), 975–999. https://doi.org/10.1080/15481603.2022.2085354. Weiss, M., Baret, F., Smith, G.J., Jonckheere, I., Coppin, P., 2004. Review of methods for in situ leaf area index (LAI) determination. Agricultural and Forest Meteorology 121 (1–2), 37–53. https://doi.org/10.1016/j.agrformet.2003.08.001. Wu, J., Kobayashi, H., Stark, S.C., Meng, R., Guan, K., Tran, N.N., Gao, S., Yang, W., Restrepo-Coupe, N., Miura, T., Oliviera, R.C., Rogers, A., Dye, D.G., Nelson, B.W., Serbin, S.P., Huete, A.R., Saleska, S.R., 2018. Biological processes dominate seasonality of remotely sensed canopy greenness in an Amazon evergreen forest. New Phytologist 217 (4), 1507–1520. https://doi.org/10.1111/nph.14939. Xu, L., Saatchi, S.S., Yang, Y., Yu, Y., Pongratz, J., Bloom, A.A., Bowman, K., Worden, J., Liu, J., Yin, Y., Domke, G., McRoberts, R.E., Woodall, C., Nabuurs, G.-J., DeMiguel, S., Keller, M., Harris, N., Maxwell, S., Schimel, D., 2021. Changes in global terrestrial live biomass over the 21st century. Science Advances 7 (27), eabe9829. https://doi.org/10.1126/sciadv.abe9829. Zanaga, D., Van De Kerchove, R., Daems, D., De Keersmaecker, W., Brockmann, C., Kirches, G., Wevers, J., Cartus, O., Santoro, M., Fritz, S., Lesiv, M., Herold, M., Tsendbazar, N.-E., Xu, P., Ramoino, F., Arino, O., 2022. ESA WorldCover 10 M 2021 V200. Zenodo. https://doi.org/10.5281/zenodo.7254221. Zhang, W., Qi, J., Wan, P., Wang, H., Xie, D., Wang, X., Yan, G., 2016. An easy-to-use airborne LiDAR data filtering method based on cloth simulation. Remote Sensing 8 (6), 501. https://doi.org/10.3390/rs8060501. Zhao, X., Chen, J.M., Zhang, Y., Jiao, Z., Liu, L., Qiu, F., Zang, J., Cao, R., 2024. Global mapping of forest clumping index based on GEDI canopy height and complementary data. ISPRS Journal of Photogrammetry and Remote Sensing 209, 1–16. https://doi. org/10.1016/j.isprsjprs.2024.01.014. Zheng, Z., Zeng, Y., Schneider, F.D., Zhao, Y., Zhao, D., Schmid, B., Schaepman, M.E., Morsdorf, F., 2021. Mapping functional diversity using individual tree-based morphological and physiological traits in a subtropical forest. Remote Sensing of Environment 252, 112170. https://doi.org/10.1016/j.rse.2020.112170. Zheng, Z., Zeng, Y., Schuman, M.C., Jiang, H., Schmid, B., Schaepman, M.E., Morsdorf, F., 2022. Individual tree-based vs pixel-based approaches to mapping forest functional traits and diversity by remote sensing. International Journal of Applied Earth Observation and Geoinformation 114, 103074. https://doi.org/ 10.1016/j.jag.2022.103074. Zhu, X., Nie, S., Wang, C., Xi, X., Lao, J., Li, D., 2022. Consistency analysis of forest height retrievals between GEDI and ICESat-2. Remote Sensing of Environment 281, 113244. https://doi.org/10.1016/j.rse.2022.113244. Ziegler, A., Heisig, J., Ludwig, M., Reudenbach, C., Meyer, H., Nauss, T., 2023. Using GEDI as training data for an ongoing mapping of landscape-scale dynamics of the plant area index. Environmental Research Letters 18 (7), 075003. https://doi.org/ 10.1088/1748-9326/acde8f. Glossary AGBD: Aboveground Biomass Density ALS: Airborne Laser Scanning CHM: Canopy Height Model DTM: Digital Terrain Model FHD: Foliage Height Diversity GEDI: Global Ecosystem Dynamics Investigation LAI: Leaf Area Index NFI: National Forest Inventory P: Percentile PAI: Plant Area Index RH: Relative Height A. Cardenas-Martinez et al. Science of Remote Sensing 11 (2025) 100195 17