Global Ecosystem Dynamics Investigation Research Articles

A spatially varying model for small area estimates of biomass density across the contiguous United States

The USDA Forest Service, Forest Inventory and Analysis (FIA) program constructs area estimates of forest attributes through a quasi-systematic sample of field plots distributed across the contiguous US. The program is under increasing pressure to provide estimates over small areas, for which the precision of FIA estimates is limited by small sample sizes. Small area estimation (SAE) models can improve the precision of estimates by leveraging relationships between forest attributes and auxiliary data across multiple small areas of interest. SAE models with large model domains are challenged by requiring auxiliary data encompassing the large domain and by spatially varying relationships between the auxiliary data and forest attributes. We implement a Fay–Herriot SAE model that accounts for spatial variation to make estimates of forest above-ground biomass density (AGBD) within 64,000 hectare hexagons across the contiguous US. Model inference is conducted through a Bayesian paradigm, providing AGBD estimates and quantification of uncertainty through posterior standard deviations and credible intervals. We compare the utility of data from the Global Ecosystem Dynamics Investigation (GEDI) and the National Land Cover Database 2016 Tree Canopy Cover (TCC) product as auxiliary data within the model. Results show that models using GEDI data provide more precise estimates of AGBD than direct estimates using FIA plot data alone while giving accurate quantification of uncertainty. The model using only the TCC product as auxiliary data also reported more precise estimates, but a cross-validation study revealed the reported uncertainty to be overly optimistic for high-biomass areas. We demonstrate that accounting for spatial variation in the model is crucial, and that doing otherwise leads to poor quantification of uncertainty and locally biased estimates. This study not only provides an effective model for small area estimates of AGBD with a large model domain, but also emphasizes the importance of validating models and their reported errors.

Landsat-Scale Regional Forest Canopy Height Mapping Using ICESat-2 Along-Track Heights: Case Study of Eastern Texas

Spaceborne profiling lidar missions such as the Ice, Cloud, and land Elevation Satellite-2 (ICESat-2) are collecting direct elevation measurements, supporting the retrieval of vegetation attributes such as canopy height that are crucial in forest carbon and ecological studies. However, such profiling lidar systems collect observations along predefined ground tracks which limit the spatially complete mapping of forest canopy height. We demonstrate that the fusion of ICESat-2 along-track canopy height estimates and ancillary Landsat and LANDFIRE (Landscape Fire and Resource Management Planning Tools Project) data can enable the generation of spatially complete canopy height data at a regional level in the United States. We developed gradient-boosted regression models relating canopy heights with ancillary data values and used them to predict canopy height in unobserved locations at a 30 m spatial resolution. Model performance varied (R2 = 0.44 − 0.50, MAE = 2.61–2.80 m) when individual (per month) Landsat data and LANDFIRE data were used. Improved performance was observed when combined Landsat and LANDFIRE data were used (R2 = 0.69, MAE = 2.09 m). We produced a gridded canopy height product over our study area in eastern Texas, which agreed moderately (R2 = 0.46, MAE = 4.38 m) with independent airborne lidar-derived canopy heights. Further, we conducted a comparative assessment with the Global Forest Canopy Height product, an existing 30 m spatial resolution canopy height product generated using GEDI (Global Ecosystem Dynamics Investigation) canopy height and multitemporal Landsat data. In general, our product showed better agreement with airborne lidar heights than the global dataset (R2 = 0.19 MAE = 5.83 m). Major differences in canopy height values between the two products are attributed to land cover changes, height metrics used (98th in this study vs 95th percentile), and the inherent differences in lidar sampling and their geolocation uncertainties between ICESat-2 and GEDI. On the whole, our integration of ICESat-2 data with ancillary datasets was effective for spatially complete canopy height mapping. For better modeling performance, we recommend the careful selection of ICESat-2 datasets to remove erroneous data and applying a series of Landsat data to account for phenological changes. The canopy height product provides a valuable spatially detailed and synoptic view of canopy heights over the study area, which would support various forestry and ecological assessments at an enhanced 30 Landsat spatial resolution.

First validation of GEDI canopy heights in African savannas

Savannas have complex, discontinuous woody vegetation structures that vary greatly in vertical and spatial arrangement and change due to climatic, ecological and management impacts. While airborne laser scanning (ALS) data have provided detailed information on vertical vegetation structure and is widely used in ecological studies, it is lacking in availability and repeat frequency. Although the Global Ecosystem Dynamics Investigation (GEDI) waveform Light Detection and Ranging (LiDAR) sensor and algorithms were optimized for measuring dense forests, it was anticipated that GEDI metrics could provide useful characterization of lower stature, sparse savannas structures. This study provided the first baseline validation of Version 2 GEDI (L2A) relative height 98 (RH98) by comparing the on-orbit GEDI-RH98orb to the simulated GEDI-RH98sim derived from ALS data across diverse savanna vegetation. It furthermore determined the influence of various factors on error, e.g. algorithm setting group (SGs), beam type, day vs. night, beam sensitivity, and vegetation phenology. After applying quality flags, 22,813 GEDI footprints were analyzed across 11 sites. SGs 4–6 that are aimed at dense forests had much larger errors than SGs 1–3. The phenological conditions at the time of GEDI data acquisition had a very large influence on the error of RH98orb. During leaf-on conditions for savanna vegetation with RH98sim < 15 m, RH98orb was very accurate with R2 = 0.61, mean bias = −0.55 m, %bias = −11.1%, RMSE = 1.64 m and %RMSE = 29.8%. In leaf-off conditions where RH98sim < 15 m, RH98orb was less accurate with R2 = 0.43, mean bias = −1.47 m, %bias = −26.5%, RMSE = 2.03 m and %RMSE = 40.9%. During leaf-off conditions, the GEDI LiDAR signal at the start of the waveform may be weaker as it interacts with denuded branches and may be truncated as noise, leading to a large negative height bias. Therefore, assessments of deciduous vegetation structures should be conducted during leaf-on periods. In leaf-on conditions, GEDI's RH98orb was very accurate between canopy heights of 3 and 7 m, with a mean bias of −0.79 m (−10%). The bias of RH98orb was not influenced by canopy cover. Due to the GEDI LiDAR pulse width of 15.6 ns, the GEDI-RH98 data product cannot reliably estimate canopy heights of shrubs below 2.34 m and will require more complex deconvolution of the waveform. GEDI's RH98 accurately estimates the canopy height of trees between 3 and 15 m allowing assessment of canopy heights over vast savanna areas.

Mapping fine-scale building heights in urban agglomeration with spaceborne lidar

The increasing availability of 3-D urban data yields new insights into urban developments and their implications for population density, energy consumption, and the carbon budget. However, existing products of urban building heights at a regional or global scale are mostly subject to coarse grid size or long-time lags. Fine-resolution building height maps remain unavailable for most regions to present recent vertical information. Also, the lack of sampled building heights with spatial and temporal consistency at larger scales makes it difficult to update building height maps. The newly available spaceborne lidar data from the Global Ecosystem Dynamics Investigation (GEDI) are expected to overcome these critical barriers. Here we propose a generalizable approach to mapping large-scale distributions of building heights by fusing GEDI-derived relative height metrics, optical data (i.e., Landsat-8 and Sentinel-2), and radar data (Sentinel-1). We applied our approach to the Yangtze River Delta region (YRD), one of China's largest urban agglomerations. We investigated the availability of building height samples generated from GEDI-based relative height metrics and the optimal grid size of building height mapping into 30-m (about footprint level) and 150-m (regional level) grid sizes. We finally developed a random forest model using active and passive sensors to extrapolate GEDI-derived samples discretely to the continuous building height map at the 150-m grid size in 2019. We revealed the spatial distribution patterns of building heights and the urbanization effect on the city's mean building heights in the YRD region. We found that: (1) GEDI-derived building height samples were highly consistent with the reference building height data at the footprint (i.e., Pearson's r = 0.86, RMSE = 9.25 m, n = 237) and regional (i.e., Pearson's r = 0.76, RMSE = 6.36 m, n = 69) levels. (2) The Pearson's correlation coefficient (r) and root mean square error (RMSE) of the final spatially continuous and up-to-date building height map were 0.85 and 5.03 m, respectively. (3) The resulting map showed a strong positive correlation between the city's mean building height and urbanization level. Our work makes it possible to use remotely sensed data to make global maps of building heights at a 150-m scale.

Fusing GEDI with earth observation data for large area aboveground biomass mapping

• A method for biomass mapping using GEDI and Earth Observation data is proposed. • Biomass maps were produced for Australia and the United States in 2020. • The saturation effect is reduced in comparison to the previous product. • Data fusion led to improved accuracies driven by Sentinel-2 and land cover data. • Predictions were more reliable in forests as compared to grasslands and shrublands. An accurate and spatially explicit estimation of biomass is required for sustainable forest management, prevention of biodiversity loss, and carbon accounting for climate change mitigation. This study offers a methodology to generate wall-to-wall aboveground biomass density (AGBD) maps that exclusively relies on open access earth observation (EO) data. Specifically, spaceborne Global Ecosystem Dynamics Investigation (GEDI) LiDAR data were fused with Sentinel-1 synthetic-aperture radar, Sentinel-2 multispectral, elevation, and land cover data to produce biomass maps of Australia and the United States for 2020. The gradient boosting machine learning framework was applied to predict AGBD and its uncertainty at the resolutions of 100 m and 200 m. The performance of models based on (1) Sentinel-2 imagery and land cover and (2) a combination of Sentinel-2 and Sentinel-1 imagery with elevation and land cover data were compared. The most accurate gradient boosting model was identified using a Bayesian hyperparameter optimization with a 5-fold cross-validation. The Sentinel-2 imagery and land cover data analysis resulted in AGBD estimated with the coefficient of determination ( R 2 ) of 0.61 – 0.71, root-mean-square error ( RMSE ) of 59 – 86 Mg/ha, and relative root-mean-square error ( RMSE% ) of 45 – 80%. The accuracy of the models improved with the addition of Sentinel-1 and elevation data: AGBD estimation with R 2 of 0.66 – 0.74, RMSE of 55 – 81 Mg/ha, and RMSE% of 41 – 77%. It was found that Sentinel-2 and land cover-derived predictors were the most important in estimating annual AGBD. The proposed method also reduced the saturation effect, which is common in high biomass areas when predicting AGBD using satellite imagery. Prediction maps produced in this study could serve as a baseline for current AGB stocks of forested lands equal to 9.8 Pg and 37.1 Pg in Australia and the United States, respectively. Overall, this research highlights methodological opportunities for combining open access EO data to yield more accurate and globally applicable AGB maps through data fusion.

Nationwide native forest structure maps for Argentina based on forest inventory data, SAR Sentinel-1 and vegetation metrics from Sentinel-2 imagery

Detailed maps of forest structure attributes are crucial for sustainable forest management, conservation, and forest ecosystem science at the landscape level. Mapping the structure of broad heterogeneous forests is challenging, but the integration of extensive field inventory plots with wall-to-wall metrics derived from synthetic aperture radar (SAR) and optical remote sensing offers a potential solution. Our goal was to map forest structure attributes (diameter at breast height, basal area, mean height, dominant height, wood volume and canopy cover) at 30-m resolution across the diverse 463,000 km2 of native forests of Argentina based on SAR Sentinel-1, vegetation metrics from Sentinel-2 and geographic coordinates. We modelled the forest structure attributes based on the latest national forest inventory, generated uncertainty maps, quantified the contribution of the predictors, and compared our height predictions with those from GEDI (Global Ecosystem Dynamics Investigation) and GFCH (Global Forest Canopy Height). We analyzed 3788 forest inventory plots (1000 m2 each) from Argentina's Second Native Forest Inventory (2015–2020) to develop predictive random forest regression models. From Sentinel-1, we included both VV (vertical transmitted and received) and VH (vertical transmitted and horizontal received) polarizations and calculated 1st and 2nd order textures within 3 × 3 pixels to match the size of the inventory plots. For Sentinel-2, we derived EVI (enhanced vegetation index), calculated DHIs (dynamic habitat indices (annual cumulative, minimum and variation) and the EVI median, then generated 1st and 2nd order textures within 3 × 3 pixels of these variables. Our models including metrics from Sentinel-1 and 2, plus latitude and longitude predicted forest structure attributes well with root mean square errors (RMSE) ranging from 23.8% to 70.3%. Mean and dominant height models had notably good performance presenting relatively low RMSE (24.5% and 23.8%, respectively). Metrics from VH polarization and longitude were overall the most important predictors, but optimal predictors differed among the different forest structure attributes. Height predictions (r = 0.89 and 0.85) outperformed those from GEDI (r = 0.81) and the GFCH (r = 0.66), suggesting that SAR Sentinel-1, DHIs from Sentinel-2 plus geographic coordinates provide great opportunities to map multiple forest structure attributes for large areas. Based on our models, we generated spatially-explicit maps of multiple forest structure attributes as well as uncertainty maps at 30-m spatial resolution for all Argentina's native forest areas in support of forest management and conservation planning across the country.

Assessing protected area’s carbon stocks and ecological structure at regional-scale using GEDI lidar

Protected areas (PAs) serve as a critical strategy for protecting natural resources, conserving biodiversity, and mitigating climate change. While there is a critical need to guide area-based conservation efforts, a systematic assessment of PA effectiveness for storing carbon stocks has not been possible due to the lack of globally consistent forest biomass data. In this study, we present a new methodology utilizing forest structural information and aboveground biomass density (AGBD) obtained from the Global Ecosystem Dynamics Investigation (GEDI) mission. We compare PAs with similar, unprotected forests obtained through statistical matching to assess differences in carbon storage and forest structure. We also assess matching outcomes for a robust and minimally biased way to quantify PA efficacy. We find that all analyzed PAs in Tanzania possess higher biomass densities than their unprotected counterfactuals (24.4% higher on average). This is also true for other forest structure metrics, including tree height, canopy cover, and plant area index (PAI). We also find that community-governed PAs are the most effective category of PAs at preserving forest structure and AGBD – often outperforming those managed by international or national entities. In addition, PAs designated under more than one entity perform better than the PAs with a single designation, especially those with multiple international designations. Finally, our findings suggest that smaller PAs may be more effective for conservation, depending on levels of connectivity. Taken together, these findings support the designation of PAs as an effective means for forest management with considerable potential to protect forest ecosystems and achieve long-term climate goals.

Predicting the Forest Canopy Height from LiDAR and Multi-Sensor Data Using Machine Learning over India

Forest canopy height estimates, at a regional scale, help understand the forest carbon storage, ecosystem processes, the development of forest management and the restoration policies to mitigate global climate change, etc. The recent availability of the NASA’s Global Ecosystem Dynamics Investigation (GEDI) LiDAR data has opened up new avenues to assess the plant canopy height at a footprint level. Here, we present a novel approach using the random forest (RF) for the wall-to-wall canopy height estimation over India’s forests (i.e., evergreen forest, deciduous forest, mixed forest, plantation, and shrubland) by employing the high-resolution top-of-the-atmosphere (TOA) reflectance and vegetation indices, the synthetic aperture radar (SAR) backscatters, the topography and tree canopy density, as the proxy variables. The variable importance plot indicated that the SAR backscatters, tree canopy density and the topography are the most influential height predictors. 33.15% of India’s forest cover demonstrated the canopy height &lt;10 m, while 44.51% accounted for 10–20 m and 22.34% of forests demonstrated a higher canopy height (&gt;20 m). This study advocates the importance and use of GEDI data for estimating the canopy height, preferably in data-deficit mountainous regions, where most of India’s natural forest vegetation exists.

Modeling returned signal of a satellite laser altimeter over an urban area.

Modeling the returned signal of a satellite laser altimeter is a significant theoretical foundation for understanding the topography of the Earth. According to the lidar equation and working principle of laser altimeter, we establish the mathematical expression of the urban target response waveform (TRW) by using a definite integral formation. The characteristic parameters of the TRW, including signal energy, time delay, pulse width, and skewness are deduced in detail. Based on the specifications of the global ecosystem dynamics investigation laser altimeter and the assumption that the target is considered a rectangular plane, we simulate the influences of the geometrical parameters of the target, including the region area, the center geolocation, and the surface slope, on the TRW patterns and the characteristic parameters. The simulation results show that the TRW would present left-skewed and right-skewed patterns, with the exception that the center geolocation is nearly coincident with the laser footprint center. Such skewed distributions make the time delay derived from the TRW time centroid inaccurate for representing the range between the laser altimeter and the target center. In addition, the TRW energy is positively proportional to the region area and the proximity of the center geolocation to the laser footprint center and attenuated by the cosine of the surface slope. The root-mean-square pulse width is principally determined by the tangent of the surface slope. The simulation results contribute to a better understanding of the relationship between the TRW pattern and the geometrical parameters of the illuminated target. In particular, the characteristic parameters of the TRW can provide the detailed references for developing the waveform processing algorithm and the inversion of the surface elevation, the slope, and the reflectance. The returned signal in this paper may facilitate the application of the satellite laser altimeter over an urban area.

Comparison of forest stand height interpolation of GEDI and ICESat-2 LiDAR measurements over tropical and sub-tropical forests in India.

Forests absorb atmospheric carbon and hence play a vital role in carbon sequestration and climate regulation. Recent research emphasizes developing technology and methods to understand the carbon sequestration potential in various forest ecosystems. Forest stand height estimation is one of the crucial parameters in allometry that estimates forest biomass. An attempt is made in this study to map forest stand height in tropical and sub-tropical forests in India using recently launched spaceborne LiDAR platforms Ice Cloud and Elevation Satellite (ICESat-2) and Global Ecosystem Dynamics Investigation (GEDI). A geostatistical kriging approach is used to interpolate the forest stand height, and the generated stand height surface is validated using ground truth samples. The results showed that GEDI data performed better with an RMSE of 3.99 m and 2.62 m in tropical forests than the ICESat-2 data, which showed an RMSE of 5.71 m and 5.08 m, respectively. A similar pattern was observed in sub-tropical forests where GEDI modelled stand height outperformed ICESat-2 modelled stand height. This analysis demonstrates the potential of existing spaceborne LiDAR platforms in interpolating forest stand height at different forest types. Also, the research emphasizes the necessity of a high density of LiDAR footprints spread in both across- and along-track directions for accurate interpolation of forest stand height.

Modelling forest volume with small area estimation of forest inventory using GEDI footprints as auxiliary information

• A suitable GEDI RH metric was selected to represent NFI dominant height at plot level. • A method that pairs up NFI plots with nearby GEDI footprints up to a critical distance suggested by the semi-variance examination. • Small area estimation of forest volume was improved with the help of GEDI auxiliary data. The French National Forest Inventory provides detailed forest information up to large national and regional scales. Forest inventory for small areas of interest within a large population is equally important for decision making, such as for local forest planning and management purposes. However, sampling these small areas with sufficient ground plots is often not cost efficient. In response, small area estimation has gained increasing popularity in forest inventory. It consists of a set of techniques that enables predictions of forest attributes of subpopulation with the help of auxiliary information that compensates for the small field samples. Common sources of auxiliary information usually come from remote sensing technology, such as airborne laser scanning and satellite imagery. The newly launched NASA’s Global Ecosystem Dynamics Investigation (GEDI), a full waveform Lidar instrument, provides an unprecedented opportunity of collecting large-scale and dense forest sample plots given its sampling frequency and spatial coverage. However, the geolocation uncertainty associated with GEDI footprints create important challenges for their use for small area estimations. In this study, we designed a process that provides NFI measurements at plot level with GEDI auxiliary information from nearby footprints. We demonstrated that GEDI RH 98 is equivalent to NFI dominant height at plot level. We stressed the importance of pairing NFI plots with nearby GEDI footprints, based on not only the distance in between but also their similarities, i.e., forest heights and forest types. Subsequently, these NFI-GEDI pairs were used for small area estimations following a two-phase sampling scheme. We showcased that, with an adequate sample size, small area estimation with GEDI auxiliary data can improve the accuracy of forest volume estimates.

Spatially Continuous Mapping of Forest Canopy Height in Canada by Combining GEDI and ICESat-2 with PALSAR and Sentinel

Continuous large-scale mapping of forest canopy height is crucial for estimating and reporting forest carbon content, analyzing forest degradation and restoration, or to model ecosystem variables such as aboveground biomass. Over the last years, the spaceborne Light Detection and Ranging (LiDAR) sensor specifically designed to acquire forest structure information, Global Ecosystem Dynamics Investigation (GEDI), has been used to extract forest canopy height information over large areas. Yet, GEDI has no spatial coverage for most forested areas in Canada and other high latitude regions. On the other hand, the spaceborne LiDAR called Ice, Cloud, and Land Elevation Satellite-2 (ICESat-2) provides a global coverage but was not specially developed to study forested ecosystems. Nonetheless, both spaceborne LiDAR sensors obtain point-based information, making spatially continuous forest canopy height estimation very challenging. This study compared the performance of both spaceborne LiDAR, GEDI and ICESat-2, combined with ALOS-2/PALSAR-2 and Sentinel-1 and -2 data to produce continuous canopy height maps in Canada for the year 2020. A set-aside dataset and airborne LiDAR (ALS) from a national LiDAR campaign were used for accuracy assessment. Both maps overestimated canopy height in relation to ALS data, but GEDI had a better performance than ICESat-2 with a mean difference (MD) of 0.9 m and 2.9 m, and a root mean square error (RMSE) of 4.2 m and 5.2 m, respectively. However, as both GEDI and ALS have no coverage in most of the hemi-boreal forests, ICESat-2 captures the tall canopy heights expected for these forests better than GEDI. PALSAR-2 HV polarization was the most important covariate to predict canopy height, showing the great potential of L-band in comparison to C-band from Sentinel-1 or optical data from Sentinel-2. The approach proposed here can be used operationally to produce annual canopy height maps for areas that lack GEDI and ICESat-2 coverage.

Evaluation of NASA’s GEDI Lidar Observations for Estimating Biomass in Temperate and Tropical Forests

Accurate estimation of forest aboveground biomass (AGB) is vital for informing ecosystem and carbon management. The Global Ecosystem Dynamics Investigation (GEDI) instrument—a new-generation spaceborne lidar system from NASA—provides the first global coverage of high-resolution 3D altimetry data aimed specifically for mapping Earth’s forests, but its performance is yet to be tested for large parts of the world. Here, our goal is to evaluate the accuracies of GEDI in measuring terrain, forest vertical structures, and AGB in reference to independent airborne lidar data over temperate and tropical forests in North America. We compared GEDI-derived elevations and canopy heights (e.g., relative height percentiles such as RH50 and RH100) with those from the Shuttle Radar Topography Mission (SRTM) or from two airborne lidar systems: the Laser Vegetation Imaging Sensor (LVIS) and Goddard’s Lidar, Hyperspectral and Thermal system (G-LiHT). We also estimated GEDI’s geolocation errors by matching GEDI waveforms and G-LiHT pseudo-waveforms. We assessed the predictive power of GEDI metrics in estimating AGB using Random Forests regression. Results showed that GEDI-derived ground elevations correlated strongly those from LVIS, G-LiHT, and LVIS (R2 &gt; 0.91), but with nonnegligible RMSEs of 5.7 m (G-LiHT), 3.1 m (LVIS), and 10.9 m (SRTM). GEDI canopy heights had poorer correlation with LVIS (e.g., R2 = 0.44 for RH100) than with G-LiHT (e.g., R2 = 0.60 for RH100). The estimated horizontal geolocation errors of GEDI footprints averaged 6.5 meters, comparable to the nominal accuracy of 9 m. Correction for the locational errors improved the correlation of GEDI vs G-LiHT canopy heights significantly, on average by 53% (e.g., R2 from 0.57 to 0.82 for RH50). GEDI canopy metrics were useful for predicting AGB (R2 = 0.82 and RMSE = 19.1 Mg/Ha), with the maximum canopy height RH100 being the most useful predictor. Our results highlight the importance of accommodating or correcting for GEDI geolocation errors for estimating forest characteristics and provide empirical evidence on the utility of GEDI for monitoring global biomass dynamics from space.

Consistency analysis of forest height retrievals between GEDI and ICESat-2

Two space-borne LiDAR missions (Global Ecosystem Dynamics Investigation, GEDI; Ice, Cloud, and land Elevation Satellite-2, ICESat-2) have unique advantages in retrieving forest heights. Fusing these two missions will greatly increase the number of forest height samples and realize their geographical complementarity, providing unprecedented opportunities for global forest height mapping. However, ICESat-2 and GEDI use different LiDAR technologies, which may lead to inconsistencies in the forest height estimation between the two missions. This study aims to explore the potential of obtaining consistent forest heights from ICESat-2 and GEDI data in support of global forest height mapping. First, the accuracies of GEDI and ICESat-2 forest heights in four different scenarios (nigh/daytime and strong/weak beam) were validated and compared utilizing airborne LiDAR data. Second, we quantitatively evaluated the differences in the forest heights derived from the GEDI and ICESat-2 data within their overlapping footprints and analyzed the effects of the terrain slope, forest coverage, forest type and study site on these differences. Third, the forest height consistency models were built based on GEDI-derived forest heights and ICESat-2 feature parameters using stepwise regression (SR) and random forest (RF) algorithms as well as a combination of both SR and RF algorithms. Finally, we evaluated the transferability of forest height consistency models to different forest types and study sites, and further analyzed the potential of building a universally consistent model. The results showed that the accuracies of both GEDI and ICESat-2 derived forest heights differ among four different scenarios, and the accuracy of forest heights extracted from GEDI data (R2 = 0.93, RMSE = 2.99 m for power beams at night) is higher than that extracted from ICESat-2 data (R2 = 0.78, RMSE = 4.62 m for strong beams at night) regardless of scenarios. The forest height differences exist between these two missions, and show an apparent change trend with increasing forest coverage. By establishing the consistency models, the forest height difference between GEDI and ICESat-2 can be reduced. Compared to the forest height consistency models built by SR or RF algorithms, the consistency models established by combining the RF and SR algorithms have higher accuracy (average R2 = 0.86, RMSE = 2.56 m for ATL08). Additionally, the consistency models developed specifically for one forest type or study site are less transferable, while a universal consistency model is applicable for various vegetation types and study sites. After building a universal consistency model based on both SR and RF algorithms, the consistent forest heights were obtained from GEDI and ICESat-2 data. Overall, this study demonstrates the possibility of combining different modes of space-borne LiDAR data to obtain consistent forest height datasets.

Assessment of terrain elevation estimates from ICESat-2 and GEDI spaceborne LiDAR missions across different land cover and forest types

Accurate measurements of terrain elevation are crucial for many ecological applications. In this study, we sought to assess new global three-dimensional Earth observation data acquired by the spaceborne Light Detection and Ranging (LiDAR) missions Ice, Cloud, and Land Elevation Satellite-2 (ICESat-2) and Global Ecosystem Dynamics Investigation (GEDI). For this, we examined the “ATLAS/ICESat-2 L3A Land and Vegetation Height”, version 5 (20 × 14 m and 100 × 14 m segments) and the “GEDI Level 2A Footprint Elevation and Height Metrics”, version 2 (25 m circle). We conducted our analysis across four land cover classes (bare soil, herbaceous, forest, savanna), and six forest types (temperate broad-leaved, temperate needle-leaved, temperate mixed, tropical upland, tropical floodplain, and tropical secondary forest). For assessment of terrain elevation estimates from spaceborne LiDAR data we used high resolution airborne data. Our results indicate that both LiDAR missions provide accurate terrain elevation estimates across different land cover classes and forest types with mean error less than 1 m, except in tropical forests. However, using a GEDI algorithm with a lower signal end threshold (e.g., algorithm 5) can improve the accuracy of terrain elevation estimates for tropical upland forests. Specific environmental parameters (terrain slope, canopy height and canopy cover) and sensor parameters (GEDI degrade flags, terrain estimation algorithm; ICESat-2 number of terrain photons, terrain uncertainty) can be applied to improve the accuracy of ICESat-2 and GEDI-based terrain estimates. Although the goodness-of-fit statistics from the two spaceborne LiDARs are not directly comparable since they possess different footprint sizes (100 × 14 m segment or 20 × 14 m segment vs. 25 m circle), we observed similar trends on the impact of terrain slope, canopy cover and canopy height for both sensors. Terrain slope strongly impacts the accuracy of both ICESat-2 and GEDI terrain elevation estimates for both forested and non-forested areas. In the case of GEDI the impact of slope is, however, partly caused by horizontal geolocation error. Moreover, dense canopies (i.e., canopy cover higher than 90%) affect the accuracy of spaceborne LiDAR terrain estimates, while canopy height does not, when considering samples over flat terrains. Our analysis of the accuracy and precision of current versions of spaceborne LiDAR products for different vegetation types and environmental conditions provides insights on parameter selection and estimated uncertainty to inform users of these key global datasets.

GEDI launches a new era of biomass inference from space

Accurate estimation of aboveground forest biomass stocks is required to assess the impacts of land use changes such as deforestation and subsequent regrowth on concentrations of atmospheric CO2. The Global Ecosystem Dynamics Investigation (GEDI) is a lidar mission launched by NASA to the International Space Station in 2018. GEDI was specifically designed to retrieve vegetation structure within a novel, theoretical sampling design that explicitly quantifies biomass and its uncertainty across a variety of spatial scales. In this paper we provide the estimates of pan-tropical and temperate biomass derived from two years of GEDI observations. We present estimates of mean biomass densities at 1 km resolution, as well as estimates aggregated to the national level for every country GEDI observes, and at the sub-national level for the United States. For all estimates we provide the standard error of the mean biomass. These data serve as a baseline for current biomass stocks and their future changes, and the mission’s integrated use of formal statistical inference points the way towards the possibility of a new generation of powerful monitoring tools from space.

Mixed tropical forests canopy height mapping from spaceborne LiDAR GEDI and multisensor imagery using machine learning models

Spatial mapping of forests canopy height (Hcanopy) provides an opportunity to assess above-ground biomass, net primary productivity, carbon dioxide (CO2) sequestration, biodiversity conservation and forest fire risks. This study incorporated a continuous coverage of multi-spectral optical and synthetic aperture radar (SAR) along with sparsely global ecosystem dynamics investigation (GEDI) spaceborne Light Detection and Ranging (LiDAR) data in the machine learning (ML) models for mapping Hcanopy in the mixed tropical forests of Shoolpaneshwar wildlife sanctuary (SWLS), Gujarat, India. We trained seven ML models, including quantile random forest (QRF), support vector machine (SVM), Bayesian regularization for feed-forward neural networks (BRNN), conditional inference random forest (Cforest), Extreme gradient boosting (Xgbtree), multivariate adaptive regression splines (MARS), and k-nearest neighbors (KNN) using GEDI_02A extracted Hcanopy as training data. We used predictors which were extracted from LiDAR (GEDI metrics), multispectral optical (Landsat -8, Sentinel-2), and SAR (ALOS-2/PALSAR-2, Sentinel-1). A 10-fold cross-validation (CV) resampling was used to avoid overfitting or underfitting. The comparison of the models performances shows that the BRNN model has the highest satisfactory accuracy metrics, such as root mean square error (RMSE) of 4.686 m, R-squared (R2) of 0.49 and mean absolute error (MAE) of 3.66 m. Low training samples of tall canopies (>25 m), presence of mixed vegetation, geometric and structural variability and sloppy terrain of SWLS possibly restricted models from performing well. Field validation shows an R2 of 0.55, satisfactory for mixed tropical forests using spaceborne LiDAR. The present work provides insights into using spaceborne LiDAR GEDI data with optical and SAR data for Hcanopy mapping through ML models, which help to manage SWLS and further implications of forest Hcanopy mapping over large spatial scales.

Detecting Mountain Forest Dynamics in the Eastern Himalayas

Forest dynamics is critical to forested ecosystems, and considerable efforts have been devoted to monitoring long-term forest dynamics with the goals of sustainable management and conservation of forests. However, little attention has been given to mountain forests, which are more challenging to monitor due to complex topography, weather, and their distribution. We developed a 30-m resolution tree-canopy cover (TCC) and forest change dataset for the Eastern Himalayas from 1986 to 2021. The tree-canopy cover estimation was validated against estimates from the space-borne Global Ecosystem Dynamics Investigation (GEDI), demonstrating strong consistency (R-square greater than 0.81). A comprehensive assessment for the forest change dataset was performed using 448 visually interpreted points and reported high accuracy of the dataset, i.e., 97.7% and 95.9% for forest loss and gain, respectively. Higher producer and user accuracies were reported for forest loss (PA = 78.0%, UA = 60.9%) than these for forest gain (PA = 61.7%, UA = 56.7%). The results indicated that (1) the mean tree-canopy cover in the region increased by 2.76% over the past three decades, from 40.67% in 1990 to 43.43% in 2020, suggesting the forests have improved during the period; (2) forest loss was identified for a total area of 6990 km2 across the study area, which is less than the 10,700 km2 identified as forest gain; (3) stronger forest gains were found at elevations greater than 3000 m asl, indicating faster forest growth in high elevations likely influenced by the warming temperatures in the Eastern Himalayas.

Performance of GEDI Space-Borne LiDAR for Quantifying Structural Variation in the Temperate Forests of South-Eastern Australia

Monitoring forest structural properties is critical for a range of applications because structure is key to understanding and quantifying forest biophysical functioning, including stand dynamics, evapotranspiration, habitat, and recovery from disturbances. Monitoring of forest structural properties at desirable frequencies and cost globally is enabled by space-borne LiDAR missions such as the global ecosystem dynamics investigation (GEDI) mission. This study assessed the accuracy of GEDI estimates for canopy height, total plant area index (PAI), and vertical profile of plant area volume density (PAVD) and elevation over a gradient of canopy height and terrain slope, compared to estimates derived from airborne laser scanning (ALS) across two forest age-classes in the Central Highlands region of south-eastern Australia. ALS was used as a reference dataset for validation of GEDI (Version 2) dataset. Canopy height and total PAI analyses were carried out at the landscape level to understand the influence of beam-type, height of the canopy, and terrain slope. An assessment of GEDI’s terrain elevation accuracy was also carried out at the landscape level. The PAVD profile evaluation was carried out using footprints grouped into two forest age-classes, based on the areas of mountain ash (Eucalyptus regnans) forest burnt in the Central Highlands during the 1939 and 2009 wildfires. The results indicate that although GEDI is found to significantly under-estimate the total PAI and slightly over-estimate the canopy height, the GEDI estimates of canopy height and the vertical PAVD profile (above 25 m) show a good level of accuracy. Both beam-types had comparable accuracies, with increasing slope having a slightly detrimental effect on accuracy. The elevation accuracy of GEDI found the RMSE to be 10.58 m and bias to be 1.28 m, with an R2 of 1.00. The results showed GEDI is suitable for canopy densities and height in complex forests of south-eastern Australia.

A Framework for Improving Wall-to-Wall Canopy Height Mapping by Integrating GEDI LiDAR

Spatially continuous canopy height is a vital input for modeling forest structures and functioning. The global ecosystem dynamics investigation (GEDI) waveform can penetrate a canopy to precisely find the ground and measure canopy height, but it is spatially discontinuous over the earth’s surface. A common method to achieve wall-to-wall canopy height mapping is to integrate a set of field-measured canopy heights and spectral bands from optical and/or microwave remote sensing data as ancillary information. However, due partly to the saturation of spectral reflectance to canopy height, the product of this method may misrepresent canopy height. As a result, neither GEDI footprints nor interpolated maps using the common method can accurately produce spatially continuous canopy height maps alone. To address this issue, this study proposes a framework of point-surface fusion for canopy height mapping (FPSF-CH) that uses GEDI data to calibrate the initial wall-to-wall canopy height map derived from a sub-model of FPSF-CH. The effectiveness of the proposed FPSF-CH was validated by comparison to canopy heights derived from (1) a high-resolution canopy height model derived from airborne discrete point cloud lidar across three test sites, (2) a global canopy height product (GDAL RH95), and (3) the results of the FPSF-CH sub-model without fusing with the GEDI canopy height. The results showed that the RMSE and rRMSE of FPSF-CH were 3.82, 4.05, and 3.48 m, and 18.77, 16.24, and 13.81% across the three test sites, respectively. The FPSF-CH achieved improvement over GDAL RH95, with reductions in RMSE values of 1.28, 2.25, and 2.23 m, and reductions in rRMSE values of 6.29, 9.01, and 8.90% across the three test sites, respectively. Additionally, the better performance of the FPSF-CH compared with its sub-model further confirmed the effectiveness of integrating GEDI data for calibrating wall-to-wall canopy height mapping. The proposed FPSF-CH integrates GEDI LiDAR data to provide a new avenue for accurate wall-to-wall canopy height mapping critical to applications, such as estimations of biomass, biodiversity, and carbon stocks.