Fig 9 - uploaded by Chiara Torresan
Content may be subject to copyright.
Stratified models (PFT by geographic region) were equal or better than a single global model applied to each strata in terms of mean residual error (Mg/ha) (b), %RMSE (c) and R 2 (d), with the exception of DBT_Eu.
Source publication
NASAs Global Ecosystem Dynamics Investigation (GEDI) is collecting space-borne full waveform lidar data with a primary science goal of producing accurate estimates of forest aboveground biomass density (AGBD). This paper presents the development of the models used to create GEDIs footprint-level (~25 m) AGBD (GEDI04_A) product, including a descript...
Contexts in source publication
Context 1
... Asia (Fig. 8, Table 3), forcing RH98 into models may have yielded higher sensitivity to biomass in low canopy covers or emergent trees, while removing low (<RH50) predictors yielded models that are theoretically more transferable to on-orbit GEDI data ( Hancock et al., 2019). Constrained sets of predictors rarely impacted the accuracy of models (Fig. 9). In some strata (EBT Asia, GSW Oceania, ENT Oceania), applying both constraints increased the %RMSE more than 5% (Table 4), and in other strata (DBT Europe, DBT North America) there was an increase in mean bias by more than 10 Mg/ha. However, for most strata all four scenarios yielded similar results. The same was true when fitting ...
Context 2
... geographic region level (Table 2) typically performed better than models stratified by geographic region or PFT alone (Table 3) in terms of accuracy assessment (lower mean residual error, lower %RMSE, higher R 2 ). When directly comparing estimates from the most refined PFT by geographic region models with estimates from a single global model fit (Fig. 9), the more stratified models were equal to or better than the global model in a given stratum with respect to %RMSE. The stratified models also had lower MRE values, and the R 2 values were similar between the two sets. Some strata did not benefit from a more stratified model, while others improved ...
Context 3
... stratified by both PFT and geographic region generally performed better than more broadly stratified (only PFT, only geographic region, or global) models, as expected (Fig. 9). However, more detailed strata often used training data outside of their stratum, suggesting either that there were insufficient training data within a stratum to fit a geographically transferable model (e.g., EBT Asia), or that certain strata are structurally similar across broader geographic domains, and thus broader training ...
Context 4
... most noteworthy difference observed between a single global model fit and stratified models is the increased MRE associated with the global model fit (Fig. 9b). This suggests that even when a global model and stratified model had comparable R 2 or %RMSE values (Fig 8), systematic errors were introduced when applying a model that was unrepresentative of the spatial domain to which it was applied. Note that the MRE term selected did not show the direction of error (it was the absolute mean ...
Similar publications
Background
Perinatal depression (PND) describes depression experienced by parents during pregnancy or in the first year after a baby is born. The EQ-5D instrument (a generic measure of health status) is not often collected in perinatal research, however disease-specific measures, such as the Edinburgh Postnatal Depression Scale (EPDS) are widely us...
Citations
... Laser scanning, an active form of remote sensing commonly known as lidar, is suitable to characterize three-dimensional forest structural properties (Lefsky et al., 2002). The Global Ecosystem Dynamics Investigation (GEDI) spaceborne lidar instrument has been providing unprecedented three-dimensional information of tropical and temperate forests worldwide (Dubayah et al., 2020Duncanson et al., 2022) and has made it possible to investigate forest structure over large areas in the Amazon and other forest biomes worldwide. ...
The Brazilian Amazon has been a focus of land development with large swaths of forests converted to agriculture. Forest degradation by selective logging and fires has accompanied the agricultural frontier and has resulted in significant impacts on Amazonian ecosystems. Changes in forest structure resulting from forest disturbances have large impacts on the surface energy balance, including on land surface temperature (LST) and evapotranspiration (ET). This study’s objective is to assess the effects of forest disturbances on water fluxes and forest structure in a transitional forest site in the Southern Amazon. We used ET and LST products from MODIS and Landsat 8 and GEDI‐derived forest structure data to address our research questions. We found that disturbances induced seasonal water stress, more pronounced in croplands/pastures than in forests (differences up to 20% in the dry season), and more pronounced in second‐growth and recently burned areas than in logged and intact forests (differences up to 12% in the dry season). Moreover, ET and LST were negatively related, with more consistent relationships across disturbance classes in the dry season (R ² : 0.41‐0.87) than in the wet season (R ² : 0.18‐0.49). Forest and cropland and pasture classes showed contrasting relationships in the dry season. Finally, we found that forest structure exhibited stronger relationships with ET and LST in the most disturbed forests (R ² : 0.01‐0.43) than in the least disturbed forests (R ² <0.05). Our findings help to elucidate degraded forests functioning under a changing climate and to improve estimates of water and energy fluxes in Amazonian degraded forests.
... Here, we use spaceborne lidar from NASA's Global Ecosystem Dynamics Instrument (GEDI), which provides near-global (between 51.6 • S and 51.6 • N latitude; Fig. 1) estimates of forest structure (see Dubayah et al., 2020 for GEDI details). Although GEDI has been used for applications such as estimating forest canopy heights (Liu et al., 2021;Potapov et al., 2021), estimating biomass and fuel loads across large areas Duncanson et al., 2022;Leite et al., 2022), and coupling the structural information provided by GEDI with additional datasets to predict the biodiversity of trees and birds (Burns et al., 2020;Marselis et al., 2022), to our knowledge, our study is the first effort to examine the relative roles of structural and species diversity in explaining aboveground carbon storage with GEDI data. Building on previous work that found positive associations between structural diversity and net primary production using either lidar (Gough et al., 2019;Hardiman et al., 2011) or forest inventory data (Dȃnescu et al., 2016;LaRue et al., 2023), our study integrates GEDI and forest inventory data to examine diversity-carbon storage relationships across the entire USA. ...
Since biodiversity often increases ecosystem functioning, changes in tree species diversity could substantially influence terrestrial carbon cycling. Yet much less is known about the relationships between forest structural diversity (i.e., the number and physical arrangement of vegetation elements in a forest) and carbon cycling, and the factors that mediate these relationships. We capitalize on spaceborne lidar data from NASA's Global Ecosystem Dynamics Investigation (GEDI) and on-the-ground forest inventory and analysis (FIA) data from 1796 plots across the contiguous United States to assess relationships among the structural and species diversity of live trees and aboveground carbon storage. We found that carbon storage was more strongly correlated with structural diversity than with species diversity, for both forest inventory-based metrics of structural diversity (e.g., height and DBH diversity) and GEDI-based canopy metrics (i.e., foliage height diversity (FHD)). However, the strength of diversity‑carbon storage relationships was mediated by forest origin and forest types. For both plot-based and GEDI-based metrics, the relationship between structural diversity (i.e., height diversity, DBH diversity, and FHD) and carbon storage was positive in natural forests for all forest types (broadleaf, mixed, conifer). For planted forests, structural diversity showed positive relationships in planted conifer forests but not in planted mixed forests. Species diversity did not show strong associations with carbon storage in natural forests but showed a positive relationship in mixed coniferous-broadleaf planted forests. Although plot-based structural diversity metrics refine our understanding of drivers of forest carbon balances at the plot scale, remotely sensed metrics such as those from GEDI can help extend that understanding to regional/national scales in a spatially continuous manner. Carbon storage showed stronger associations with plot-based structural diversity than with stand age, soil variables, or climate variables. Incorporating structural diversity into management and restoration strategies could help guide efforts to increase carbon storage and mitigate climate change as nature-based solutions.
... The GEDI on the International Space Station was orbiting on board and measured canopy height, vertical structure, and elevation between 51.6 • S and 51.6 • N [34]. GEDI offered four different types of products, which included raw waveforms; footprint-level ground and canopy heights; grid-form heights; and biomass [34,36]. The raw GEDI waveform data were a Level-1 product [36]. ...
... GEDI offered four different types of products, which included raw waveforms; footprint-level ground and canopy heights; grid-form heights; and biomass [34,36]. The raw GEDI waveform data were a Level-1 product [36]. Recently released on the Google Earth Engine (GEE), the GEDI Level-2A data offered canopy-relative height (RH) metrics, RH0-RH100 [37]. ...
... The GEDI on the International Spac Station was orbiting on board and measured canopy height, vertical structure, an elevation between 51.6°S and 51.6°N [34]. GEDI offered four different types of product which included raw waveforms; footprint-level ground and canopy heights; grid-form heights; and biomass [34,36]. The raw GEDI waveform data were a Level-1 product [36 The GEDI Level-2A data contained 2 versions (Version 1 and Version 2). ...
Stand age is a significant factor when investigating forest resource management. How to obtain age data at a sub-compartment level on a large regional scale conveniently and in real time has become an urgent scientific challenge in forestry research. In this study, we established two strategies for stand-age estimation at sub-compartment and pixel levels, specifically object-based and pixel-based approaches. First, the relationship between canopy height and stand age was established based on field measurement data, which was achieved at the Mao’er Mountain Experimental Forest Farm in 2020 and 2021. The stand age was estimated using the relationship between the canopy height, the stand age, and the canopy-height map, which was generated from multi-resource remote sensing data. The results showed that the validation accuracy of the object-based estimation results of the stand age and the canopy height was better than that of the pixel-based estimation results, with a root mean squared error (RMSE) increase of 40.17% and 33.47%, respectively. Then, the estimated stand age was divided into different age classes and compared with the forest inventory data (FID). As a comparison, the object-based estimation results had better consistency with the FID in the region of the broad-leaved forests and the coniferous forests. In addition, the pixel-based estimation results had better accuracy in the mixed forest regions. This study provided a reference for estimating stand age and met the requirements for stand-age data at the pixel and sub-compartment levels for studies involving different forestry applications.
... Credible, accurate and reliable monitoring of stocks and changes in forest structure are critical for achieving international goals and national commitments to forest conservation, management, climate change and sustainable development [1,2]. Achieving this requires accurate measurements of tree height, which is used as a predictor of biomass [3] and informs the subsequent retrieval of carbon stocks, structure, function and biodiversity [4]. Estimating forest structure, however, is problematic in tidally and seasonally flooded forests where tree height estimates vary relative to sub-canopy tidal/flooding conditions. ...
Flooding controls wetland carbon cycling and hinders accurate measurements of ecosystem structure from remotely sensed data. Single measurements over a changeable surface can yield large uncertainties, exacerbated by insufficient repeat sampling of inundation dynamics across globally distributed locations. In forested wetlands, flood frequency and duration is critical to controlling carbon cycling, but high canopy cover obscures the view of inundation stage where large fluctuations in flooding directly impact measurements of ecosystem structure.
... While the first and third assumptions were addressed by Patterson et al. (2019) during GEDI's pre-launch phase, the extent to which the L4A models are unbiased everywhere is not wellestablished (Dubayah et al., 2022b;Duncanson L. et al., 2022). This possibility is acknowledged by Duncanson L. et al. (2022), and represents a substantial challenge in developing globally representative prediction models due to geographic gaps in training data used to calibrate the L4A models that must be transferable to entire continents. ...
... While the first and third assumptions were addressed by Patterson et al. (2019) during GEDI's pre-launch phase, the extent to which the L4A models are unbiased everywhere is not wellestablished (Dubayah et al., 2022b;Duncanson L. et al., 2022). This possibility is acknowledged by Duncanson L. et al. (2022), and represents a substantial challenge in developing globally representative prediction models due to geographic gaps in training data used to calibrate the L4A models that must be transferable to entire continents. Such an assessment requires validation of GEDI's AGBD estimates with independent reference data, however standardized reference data does not exist to validate a global biomass map at 1 km resolution . ...
Atmospheric CO 2 concentrations are dependent on land-atmosphere carbon fluxes resultant from forest dynamics and land-use changes. These fluxes are not well-constrained, in part because reliable baseline estimates of forest carbon stocks and the associated uncertainties are lacking. NASA's Global Ecosystem Dynamics Investigation (GEDI) produces estimates of aboveground biomass density (AGBD) that are unique because GEDI's hybrid estimation framework enables formal uncertainty calculations that accompany the biomass estimates. However, GEDI's estimates are not without issue; a recent validation using design-based AGBD estimates from the US Forest Inventory and Analysis (FIA) program revealed systematic differences between GEDI and FIA estimates within a hexagon tessellation of the continental United States. Here, we explored these differences and identified two issues impacting GEDI's estimation process: incomplete filtering of low quality GEDI observations and regional biases in GEDI's footprint-level biomass models. We developed a solution to each, in the form of improved data filtering and GEDI-FIA fusion AGBD models, developed in a scale-invariant small area estimation framework, that were compatible with hybrid estimation. We calibrated 10 regional Fay-Herriot models at the hexagon scale for application at the unit scale of GEDI footprints, for which we provide a mathematical justification and empirical testing of the models' scale-invariance. These models predicted realistic distributions of unit level AGBD, with equal or improved performance relative to GEDI's L4A models for all regions. We then produced GEDI-FIA fusion estimates that were more precise than the FIA estimates and resulted in a bias reduction of 86.7% relative to the original GEDI estimates: 19.3% due to improved data filtering and 67.5% due to the new AGBD models. Our findings indicate that (1) small area estimation models trained in a scale-invariant framework can produce realistic predictions of AGBD, and (2) there is substantial spatial variability in the relationship between GEDI forest structure metrics and AGBD. This work is a step toward achieving reliable baseline forest carbon stocks, provides a viable methodology for training remote sensing biomass models, and may serve as a reference for other investigations of GEDI AGBD estimates.
... Finally, we compared the GEDI RH98 values to TanDEM-X height values to determine how comparable height estimates from each sensor are in the Niger Delta. For this comparison, we chose to compare to the RH98 since other work has found this to be a more stable parameter for estimating canopy height (Duncanson et al., 2022). ...
Invasive species are a leading threat to biodiversity worldwide. Nypa palm ( Nypa fruticans ) has emerged as the predominant invasive species in the Niger Delta region of Nigeria. While endemic mangroves have high rates of carbon sequestration, stabilize coastlines and protect biodiversity, Nypa does not provide these services outside its native region of Southeast Asia. Oil exploration and urbanization in this region also exacerbate mangrove loss and Nypa spread. As Nypa is difficult to distinguish from endemic mangrove species in remotely sensed data, estimates of mangrove and ecosystem services losses in Nigeria are highly uncertain. Here, we analyse multisensor satellite data with machine learning to quantify the rapid expansion of Nypa from 2015 to 2020 in Nigeria. Using Landsat imagery and random forest classification, we quantify total potential Nypa extent in Nigeria in 2019. We then produced a Nypa extent map using iterative combinations of Sentinel‐1 SAR, Sentinel‐2 MSI and ALOS PALSAR. Random forest classifications using SAR data from ALOS and Sentinel‐1 were best suited for mapping Nypa extent with similar accuracies (78% and 75%, respectively). Based on data availability and accuracy, we focussed our change analysis on Sentinel‐1 SAR. Our results show ~28 000 ha of mangroves were converted to Nypa in Nigeria by 2020 and covered a larger extent than endemic mangroves, compounding the effect of the existing degradation and deforestation in the region. We also compared forest height and complexity estimates from Global Ecosystem Dynamics Investigation LiDAR to further distinguish between endemic mangroves and Nypa in three dimensions. Nypa structural variability, measured by top‐of‐canopy height, vegetation cover, plant area index, and foliage height diversity, was lower than that of mangroves. At current rates of Nypa expansion, the entire area of study would be invaded by Nypa by 2028, with potentially detrimental consequences to the ecosystem services provided by mangroves.
... Within our model assessments, we found moderate to high predictive performance for a set of eight GEDI structure metrics across our diverse study region. Much of the existing GEDI literature has focused on accuracy assessments of the GEDI waveform geolocations and structure measures (Adam et al., 2020;Li et al., 2023), developing footprint level biomass models (Duncanson et al., 2022), or leveraged simulated GEDI data in lieu of actual footprint samples (Burns et al., 2020;Silva et al., 2021). Studies have begun to investigate GEDI footprint samples as the basis for scaling up forest structure measures by leveraging fusions of passive and active satellite earth observations, although the majority of those studies have been focused on elevation, canopy height, or biomass Potapov et al., 2021;Shendryk, 2022). ...
Continuous characterizations of forest structure are critical for modeling wildlife habitat as well as for assessing trade-offs with additional ecosystem services. To overcome the spatial and temporal limitations of airborne lidar data for studying wide-ranging animals and for monitoring wildlife habitat through time, novel sampling data sources, including the space-borne Global Ecosystem Dynamics Investigation (GEDI) lidar instrument, may be incorporated within data fusion frameworks to scale up satellite-based estimates of forest structure across continuous spatial extents. The objectives of this study were to: 1) investigate the value and limitations of satellite data sources for generating GEDI-fusion models and 30 m resolution predictive maps of eight forest structure measures across six western U.S. states (Colorado, Wyoming, Idaho, Oregon, Washington, and Montana); 2) evaluate the suitability of GEDI as a reference data source and assess any spatiotemporal biases of GEDI-fusion maps using samples of airborne lidar data; and 3) examine differences in GEDI-fusion products for inclusion within wildlife habitat models for three keystone woodpecker species with varying forest structure needs. We focused on two fusion models, one that combined Landsat, Sentinel-1 Synthetic Aperture Radar, disturbance, topographic, and bioclimatic predictor information (combined model), and one that was restricted to Landsat, topographic, and bioclimatic predictors (Landsat/topo/bio model). Model performance varied across the eight GEDI structure measures although all representing moderate to high predictive performance (model testing R ² values ranging from 0.36 to 0.76). Results were similar between fusion models, as well as for map validations for years of model creation (2019–2020) and hindcasted years (2016–2018). Within our wildlife case studies, modeling encounter rates of the three woodpecker species using GEDI-fusion inputs yielded AUC values ranging from 0.76–0.87 with observed relationships that followed our ecological understanding of the species. While our results show promise for the use of remote sensing data fusions for scaling up GEDI structure metrics of value for habitat modeling and other applications across broad continuous extents, further assessments are needed to test their performance within habitat modeling for additional species of conservation interest as well as biodiversity assessments.
... For instance, the GEDI Level 4 Biomass (L4B) product provides 1-km aggregated estimations of above-ground biomass density (AGBD) that come from allometric equations based on waveform metrics calibrated on the biomass measured across forest plots (Dubayah et al., 2022;Duncanson et al., 2022). However, to properly monitor forests at a local scale, especially in Europe, where forests are divided into small stands of a few hectares, a typical 10 to 50 m spatial resolution is needed. ...
... For instance, the GEDI Level 4 Biomass (L4B) product provides 1-km aggregated estimations of above-ground biomass density (AGBD) that come from allometric equations based on waveform metrics calibrated on the biomass measured across forest plots (Dubayah et al., 2022;Duncanson et al., 2022). However, to properly monitor forests at a local scale, especially in Europe, where forests are divided into small stands of a few hectares, a typical 10 to 50 m spatial resolution is needed. ...
The contribution of forests to carbon storage and biodiversity conservation highlights the need for accurate forest height and biomass mapping and monitoring. In France, forests are managed mainly by private owners and divided into small stands, requiring 10 to 50 m spatial resolution data to be correctly separated. Further, 35 % of the French forest territory is covered by mountains and Mediterranean forests which are managed very extensively. In this work, we used a deep-learning model based on multi-stream remote sensing measurements (NASA’s GEDI LiDAR mission and ESA’s Copernicus Sentinel 1 & 2 satellites) to create a 10 m resolution canopy height map of France for 2020 (FORMS-H). In a second step, with allometric equations fitted to the French National Forest Inventory (NFI) plot data, we created a 30 m resolution above-ground biomass density (AGBD) map (Mg ha-1) of France (FORMS-B). Extensive validation was conducted. First, independent datasets from Airborne Laser Scanning (ALS) and NFI data from thousands of plots reveal a mean absolute error (MAE) of 2.94 m for FORMS-H, which outperforms existing canopy height models. Second, FORMS-B was validated using two independent forest inventory datasets from the Renecofor permanent forest plot network and from the GLORIE forest inventory with MAE of 59.6 Mg ha-1 and 19.6 Mg.ha-1 respectively, providing greater performance than other AGBD products sampled over France. These results highlight the importance of coupling remote sensing technologies with recent advances in computer science to bring material insights to climate-efficient forest management policies. Additionally, our approach is based on open-access data having global coverage and a high spatial and temporal resolution, making the maps reproducible and easily scalable. FORMS products can be accessed from https://doi.org/10.5281/zenodo.7840108 (Schwartz et al., 2023).
... The assessment was based on a forest height estimation. The level of uncertainty varied between regions and vegetation types [26,27]; however, saturation at the high levels of biomass typical for passive optical data [28] was not clear using GEDI observations. The GEDI was an experimental mission onboard the International Space Station. ...
Our objective was to develop a method for the assessment of forest area and structural variables for cases in which the availability of representative ground reference data is poor and these data are not collected from the whole area of interest. We implemented two independent approaches to the estimation of the forest variables of a European boreal forest: (i) the computation of wall-to-wall estimates using moderate- to low-resolution VIIRS imagery from the Suomi NPP mission; and (ii) the visual interpretation of plots of samples from very high resolution (VHR) satellite data obtained via a two-stage design. Our focus was on the statistical comparison of forest resources at a country or larger level. The study area was boreal forest ranging from Norway to the Ural Mountains in Russia. We computed a seamless mosaic from 111 VIIRS images. From the mosaic, we computed predictions for the forest area, growing stock volume, height of the dominating tree layer, proportion of conifers and broadleaved trees, site fertility class, and leaf area index. The reference data for the VIIRS imagery were national forest inventory (NFI)-based raster maps from Finland. The first stage sample of VHR data included 42 images; of these, a second stage sample of 2690 plots was visually interpreted for the same variables. The forest area prediction from VIIRS for the whole study area was 1.2% higher than the VHR-based result. All other structural variable predictions using VIIRS fitted within the 95% confidence intervals computed from the VHR sample except for estimates of the main tree species groups, which were outside the limits. A comparison of VIIRS-based forest area estimates using Finnish and Swedish NFI data indicated overestimations of 10.0% points and 4.6% points, whereas the total growing stock volumes were overestimated by 8% and underestimated by 3.4%, respectively. The correlation coefficients between the VIIRS and VHR image predictions at the 42 VHR image locations varied from 0.70 to 0.85. The VIIRS maps strongly averaged the local predictions due to their coarse spatial resolutions. Based on our findings, the approach using two independent estimations yielded similar figures for the central forest variables for the European boreal forest. A model computed using reference data from a small part of the area of interest can provide satisfactory predictions for a much larger area with a similar biome. Therefore, our concept is applicable to the estimation and overall mapping of a forest area and central structural variables at regional to national levels.
... GEDI (launched December 5, 2018) is the first satellite-based LiDAR mission designed to study forests [167] and, in conjunction with ATLAS on board ICESat-2 [168], could facilitate large-scale forest AGB estimates. Indeed, there have been studies that have used GEDI data to estimate biomass at regional and global scales [169,170]. ...
Quantifying forest aboveground biomass (AGB) is essential for elucidating the global carbon cycle and the response of forest ecosystems to climate change. Over the past five decades, remote-sensing techniques have played a vital role in forest AGB estimation at different scales. Here, we present an overview of the progress in remote sensing-based forest AGB estimation. More in detail, we first describe the principles of remote sensing techniques in forest AGB estimation: that is, the construction and use of parameters associated with AGB (rather than the direct measurement of AGB values). Second, we review forest AGB remotely sensed data sources (including passive optical, microwave, and LiDAR) and methods (e.g., empirical, physical, mechanistic, and comprehensive models) alongside their limitations and advantages. Third, we discuss possible sources of uncertainty in resultant forest AGB estimates, including those associated with remote sensing imagery, sample plot survey data, stand structure, and statistical models. Finally, we offer forward-looking perspectives and insights on prospective research directions for remote sensing-based forest AGB estimation. Remote sensing is anticipated to play an increasingly important role in future forest AGB estimation and carbon cycle studies. Overall, this comprehensive review may (1) benefit the research communities focused on carbon cycle, remote sensing, and climate change elucidation, (2) provide a theoretical basis for the study of the carbon cycle and global climate change, (3) inform forest ecosystems and carbon management, and (4) aid in the elucidation of forest feedbacks to climate change.
