How GEDI Informs Ecosystem Studies

In this section:

  • Major themes across selected example applications for considering the capabilities and use of GEDI for environmental monitoring and characterization.
  • Discussion on the advantages and disadvantages of integrating GEDI data into applied science and/or operational workflows.

Professional Using Lidar in Decision-Making

For these professionals, lidar has been a useful tool for urban development, utility management, construction design and planning where precise elevation and surface height modelling are particularly useful for zoning protocols and safety assessments. Lidar elevation, height, and vegetation structure mapping abilities provide insight for disaster risk assessment and mitigation (wild fires, landslides, floods, extreme weather damage). Advances in precision precision agriculture monitoring and large scale agriculture area and type mapping use lidar to monitor the growth and productivity relevant to food security assessments. Additionally, understanding tree crop commodity productivity or disturbances are important economic indicators. Ecosystem studies can plug into such human-use areas and disaster applications where green spaces and sustainable planning are priority. In some regions, ecosystems can play an important role in reducing the impacts of landslides, floods, or extreme weather events, and lidar has the ability to help quantify those ecosystem benefits and even support scenario analyses for future mitigation. Quantifying forest health, biodiversity and ecosystem benefits play significant roles in prioritizing conservation, restoration, and recreational sites (hunting, fish & wildlife) where lidar measurements help to map degradation and disturbances known to prevent how ecosystem priority areas improve quality of life. Policies and pledges to Sustainable Development Goals addressing unsustainable resource management rely on land cover and vegetation measurements that lidar can improve upon.

For GEDI to be fully realized in these applications, users must be able to know how to mitigate its challenges and choose the optimal solution for their purposes. The user will need to understand the negotiation across several key themes of GEDI’s capabilities before deciding on which methods to test and apply for the environmental characterization or monitoring application.

Technical Capabilities

  • What is the ability to characterize elevation, heights, vegetation structure or estimate biomass? Which methods for working the full waveform, footprint level metrics, or gridded and other derived datasets bring the most value to the final products? Which resolutions are most appropriate for the application or chosen methods?
  • Can GEDI’s ability to measure surface elevation (below forest canopies especially) improve upon digital terrain models, which are fundamental datasets for many environmental studies.

Study Area Characteristics

  • Correctly capturing the complexity of Earth’s surface is the crux of remote sensing modelling. Align the choice of data products with known characteristics over particular plant functional types, geographic regions and environmental conditions
    • Consider stratifying models by vegetation type and phenological stages.
    • Investigate data representativeness over geographic regions due to sampling patterns and gaps along with the availability of calibration datasets used to generate GEDI products.
    • Specially prepare the data over regions with high terrain slope, high cloud cover, or particularly challenges vegetated landscape characteristics or annual or seasonal behaviors.

Accuracy Assessment

  • Rigorous accuracy assessment and ground truthing is a key component of any remote sensing derived products or methods. Metrics such as Root Mean Square Error (RMSE), R2, MAE, are commonly used for validating lidar data such as GEDI.
  • Data quality and filtering plays an important role in mitigating errors and noise inherent in the data, or location, and resulting from methods.
  • Quantifying and communicating uncertainties at the sensor and mapping levels are critically important user and future researcher references.

Methodology for Continuous Mapping

  • In addition to data quality checks, several methods are commonly deployed when working with GEDI.
    • Integrating GEDI with multiple data sources, either with other lidar sources, or across different remote sensing platforms such as with radar, or optical, etc. Ancillary data are used to improve the accuracy and comprehensive characterization of the environmental study.
    • Machine learning and statistical modelling are often deployed across many remote sensing studies to predict or infer environmental variables being studied, and assess accuracy as well as correlations amongst remote sensing data and platforms.

A schematic overviewing key decisions with several options at multiple levels of applying GEDI data in your study:

GEDI Data in Practice Across Environmental Conditions and Decision-Making Applications

The best highlight the complexity and diversity of GEDI applications, presented here is a series of example studies showcasing how GEDI data was used for vegetated and non-vegetated classification or monitoring assessments. Each study highlights where specific GEDI datasets were integrated in several environmental thematic areas, and application goals.

Elevation and Height Metrics

  • Tree crops are an important resource managed to support local livelihoods and ecosystem services, but are often overlooked or misclassified as its own category in regional or global cropland maps since it can be difficult to distinguish tree crops from other trees with current methods. Adrah et al., 2025 demonstrated the value of GEDI elevation and relative height metric footprints to generalize structural information across areas of different tree types in several agroclimatic zones in the Mediterranean, showing automatic generation of training samples to both improve and expand tree crop mapping without as much manual collection.Mediterranean
  • This research by Guerra Hernandez et al., 2021 provides valuable insights for forest industry operations in Spain, particularly for capturing rapid changes in forest growth rates and updating forest structure information at shorter intervals than traditional Airborne Laser Scanning (ALS) surveys typically allow. The study’s assessments of height growth dynamics and disturbance detection are crucial for managing fast-growing plantations effectively and supporting a paradigm shift in the spatio-temporal evaluation of forest productivity, enabling more responsive and adaptive forest management practices.
  • Research by Islam et al., 2024 directly supports United Nations Sustainable Development Goals (SDGs), particularly SDG 11 (Sustainable cities and communities) by identifying areas prone to heat stress for urban green infrastructure planning, and SDG 15 (Protect terrestrial ecosystems) by determining forest loss or land degradation patterns. The study introduces an innovative dual-phase hybrid classification process using GEDI elevation and relative height footprints to automatically generate training samples for machine learning models, effectively overcoming the limitations of expensive manual data collection while improving classification accuracy at a refined 25-meter spatial resolution for global land-cover-type Local Climate Zones (LCZs) for urban and environmental studies.
  • Research by Lahssini et al., 2022 estimates canopy height (CHM) from GEDI data specifically over tropical forests. Canopy height serves as an important descriptor of forest ecosystems, allowing researchers and managers to quantify biomass and other critical inventory information. The study demonstrates that using regression models built on multiple GEDI relative height metrics, such as Random Forest approaches, significantly improves canopy height estimation accuracy compared to direct estimation from a single metric. The research also provides key inputs and practical recommendations for GEDI users on how to effectively leverage the richness of GEDI information for South American tropical forest studies, highlighting the importance of considerations such as beam type and sensitivity settings.
  • Study by Li et al., 2023 delivers the first baseline assessment of GEDI-RH98 performance specifically in African savannas, enabling the identification of research priorities necessary to expand GEDI canopy height product applicability to non-forest vegetation types. The research offers a direct and rigorous comparison by using simulated GEDI-RH98 derived from reference ALS datasets, which helps in understanding the influence of various factors on GEDI accuracy, including algorithm setting groups, beam type variations, and phenological considerations.
  • GEDI measurements provide detailed information on canopy heights that can effectively distinguish key commodity crops such as maize and sugarcane, which are typically taller than other crops like wheat, rice, and soybean. This height-based discrimination makes GEDI a particularly useful resource for advancing the goal of low-cost, timely, and accurate global mapping of crop types, especially when the data is used to train models with complementary Sentinel-2 optical data for comprehensive agricultural monitoring (Tommaso et al., 2023)
  • Robust wall-to-wall maps of the heights and locations of planted oil palm trees are necessary to better understand the economic and environmental value of managed trees, correct historical estimates of carbon sequestered in nature-based solutions, and understand growth patterns for targeted sustainable development interventions. The study demonstrates that locally trained models using GEDI relative heights combined with optical satellite data can be highly effective even in regions with relatively sparse GEDI coverage like in Nigeria, significantly improving accuracy for planted tree monitoring where global products typically struggle to perform adequately for non-forest tree plantation areas (Tsao et al., 2023)

Non-ecosystem applications

  • While GEDI is a spaceborne lidar optimized to measure “ecosystem dynamics” given its wavelength and waveform processing algorithm development, it still remains a crucial source of altimetry information, if noise removal and interpretation can be improved over non-vegetated regions such as urban areas. This study improved upon methods for removing uncertainties in GEDI’s elevation data over gentle relief urban areas in South America, allowing for more confident integration of elevation data over urban areas. (Barbosa et al., 2022)
  • This study demonstrates the ability of GEDI altimetry (elevation) data to provide accurate water level estimates in European lakes, which has substantial potential to advance our understanding of the hydrological significance of these water bodies. The research is particularly valuable for data-scarce regions where traditional water level monitoring may be limited or unavailable, offering new possibilities for comprehensive hydrological monitoring and management. (Fayad et al., 2020).
  • The study by (Zhongmin, Ma et al., 2024) demonstrates the practical feasibility of combining CYGNSS L1 standard data with DTM data to accurately measure inland water levels, highlighting its particular suitability for monitoring both flash floods and long-term changes in lake and reservoir water levels globally. In this research, GEDI, along with ICESat-2, provides high-consistency satellite-based LiDAR altimetry data that serves as crucial validation for the effectiveness of the proposed methodology.
  • Urban building height represents a fundamental 3D urban structural feature with far-reaching applications across environmental, ecological, and social sciences. This innovative product provides fine spatial resolution, reduced time lags, and ease of updating capabilities, offering novel insights into the recent state of building heights worldwide while revealing significant variations in 3D city forms across different regions. The study demonstrates that GEDI-estimated building height samples prove somewhat effective when compared to reference data, depending on the height of the building, and sample sizes available across urbanization level and policy restrictions (Ma et al., 2024).

Unsuccessful Applications

  • Understanding TMF successional stages is crucial for improving forest conservation, monitoring forest health, comprehending spatial vegetation patterns, and effective conservation strategies, as these dynamics directly impact forest productivity and carbon storage capacity. Altarez et al., 2025 hoped to map the successional stages of Tropical Montane Forests to provide critical information for biodiversity conservation strategies, carbon management, and habitat protection, particularly in areas of high endemism and ecological importance like these forests in the Philippines. Unfortunately, the application of GEDI’s relative height metrics was unsuccessful over such a highly dense and mountainous forest with heterogeneous vegetation structure. Monitoring successional stage dynamics remains an important goal, however future research is needed to continue testing the applicability of GEDI for this purpose.
  • Characterizing canopy structure of successional forest dynamics and supporting land stewardship planning, particularly identifying at-risk areas for conservation prioritization and zoning agricultural expansion such with oil palm in the Peruvian Amazon. GEDI elevation and relative height observations have the potential to characterize distinct forest types and/or successional stages of regeneration by filling the gap in understanding the structural patterns with these types and stages at landscape scales. While some complexities of the forest were captured, the ability to differentiate between forest types and stages was limited due to the heterogeneity and density of the areas, geolocation errors and nonrandom spatial distribution and data collection all impacted classification of certain forest or regrowth types. (Cooley et al., 2024).
  • AGB and tree mortality is crucial for understanding forest health and carbon dynamics, as well as for informing strategic interventions aimed at preserving forest health and mitigating climate change impacts. GEDI simulated waveforms are valuable for broader-scale forest monitoring due to its capacity to capture vertical canopy structure and ground returns across large areas, in theory could help improve tracking forest carbon stocks and identifying widespread mortality or biomass loss. A study by Bueno et al., 2025 attempted to simulate GEDI waveforms over the USA to interpret individual tree biomass and mortality. While relevant height, structure, and biomass variables were able to be extracted, their application towards individual tree biomass and mortality was too granular for the data compared to airborne methods.

Waveform and GEDI simulation use cases

  • Estimating canopy fuel load (CFL) and canopy bulk density (CBD) for wild fire behavior modeling in Europe at a continental scale is key to assessing fire spread risk and vegetation regeneration potential. The generated wall-to-wall maps based on GEDI waveforms contribute to more effective crown fire risk reduction strategies in Europe and improve coordinated understanding of forests for wildfire management. This is the first dataset of spatially explicit information on these parameters at the European continental scale. (Aragonese et al., 2025)

L4A Biomass

  • Aboveground biomass serves as a crucial indicator for carbon monitoring and is essential for sustainable forest management, prevention of biodiversity loss, and accurate carbon accounting. This study advances AGBD mapping in Alabama, USA by demonstrating that ecoregion-specific models utilizing GEDI biomass footprints achieve significantly higher accuracy than broader statewide models. The research also highlights the critical importance of combining a cloud-based workflow using optical, SAR, and lidar data with ancillary information to effectively capture the complexity of forest ecosystems for accurate AGBD estimation. (Sandamali et al., 2025)

L4B Gridded Biomass

  • Representing carbon stocks in tropical and subtropical areas where data gaps exist and enhances national efforts to obtain AGB reference values to meet global tracking and economic trend valuation. GEDI when fused with other datasets helped fill gaps in aboveground biomass and aboveground carbon estimations across vegetation types in submontane dense humid forests of Brazil. (Dionizio et al., 2025)

Applying Multiple GEDI Metrics

  • Forest structure is key to understanding the resilience of tropical forests and predicting their response to environmental and climatic fluctuations. GEDI provides detailed information on canopy structure, which is vital for forest recovery, resilience, and ecosystem services. A study by Doyle et al., 2025 strengthened confidence in GEDI’s combined L2A, L2B, and L4A relative height, structure, and biomass footprint ability to measure a gradient of degraded forest states to support post-degradation monitoring and management strategies for conservation and restoration in Amazonia given the opportunity to derive a continuous forest structural state metric.
  • Research by Dwiputra et al., 2023 addresses the critical need for comprehensive vegetation mapping that goes beyond simple forest/non-forest classifications, which is essential for tackling the global climate crisis and biodiversity loss. The study demonstrates that GEDI waveform and L2A height metrics data, particularly when combined with Sentinel-1 and Sentinel-2 satellite imagery, can produce accurate wall-to-wall heterogeneous vegetation type maps in Southeast Asia, and proves valuable for reference data collection, especially in data-scarce environments where traditional mapping methods may be limited. This comprehensive mapping approach is particularly useful in designing and monitoring nature-based solutions (NbS) projects.
  • Understanding functional traits and diversity in semi-arid ecosystems provides crucial indicators of resource availability and responses to environmental perturbations such as fires, carrying important global implications for ecosystem management. The research by Ilangakoon et al., 2021 demonstrates GEDI’s waveform, relative height, and vegetation structure measurements exciting potential to identify critical biophysical and ecological shifts in North America and helps understand its capacity to monitor changes in carbon-cycle dynamics, habitats, and biodiversity across the globe in the climatically important semi-arid ecosystems.
  • Estimating large-scale multi-layer fuel loads, including ground, surface, shrubs, trees, and total fuel loads in tropical savanna ecosystems can improve fire management planning and decision-making processes at both regional and global scales. The study marks the first research demonstrating putting L2A relative heights and L2B structural metrics practical usefulness in estimating fuel loads at such a large geographic scale, effectively expanding spaceborne lidar applications for integrated fire management and carbon monitoring initiatives in South American savanna ecosystems. (Leite et al., 2022).
  • A study by McClure et al, 2024 classifies forest overstory and understory layers in Tanzania while assessing species diversity using GEDI L1B data and GEDI-derived structural metrics. This research demonstrates that GEDI provides valuable insights into the complex multi-layer structure of forests and significantly helps in biodiversity assessment across different biomes for comprehensive ecological monitoring, particularly in distinguishing between different forest layers and their associated biodiversity patterns.
  • While GEDI footprint-level products offer valuable point samples, integrating GEDI sample data with wall-to-wall global optical data time-series enables comprehensive long-term monitoring of vegetation structure across vast land areas, providing consistent temporal coverage for global multidecadal analysis. The spatiotemporal extrapolation of GEDI information does offer consistent multidecadal monitoring of vegetation structure. (Potapov et al., 2021)

Vegetation Structure

  • A study by Elliott et al., 2024 explores the assessment of GEDI height, canopy cover and foliage height diversity data fusions to map North American woodpecker distributions and identify biodiversity hotspots with the understanding that woodpeckers are closely linked with songbird diversity. making their habitat identification crucial for conservation planning efforts. The findings demonstrate that GEDI-fusion metrics significantly enhance the modeling of biodiversity hotspots leveraging broad-scale relationships among environmental variables, forest structure, and woodpecker diversity patterns. Habitat identification is crucial for management and conservation efforts, and the results of this study suggest the potential of woodpeckers as effective indicator species for ecosystem health.
  • Mapping ecosystem extent is essential for monitoring ecosystems and their dynamic changes over time given extent is a critical biodiversity variable. By combining GEDI lidar’s structural metrics with spectral and textural features from other sensors like Landsat and PALSAR-2, a study by Geremew et al., 2023 enhances accuracy substantially improves the utility of land cover maps for various applications including climate monitoring, food security assessment, and conservation planning in East Africa.
  • The research addresses the crucial need for information about the composition and distribution of dominant tree species for sustainable forest management practices in temperate and subtropical forests. The study demonstrates that GEDI-derived forest vertical structure characterization complements traditional spectral features. This finding addresses common issues of spectral similarity and saturation while showing good stability and applicability across different seasons and climatic regions in East Asia. (Han et al., 2024)
  • The PAI profile provides essential information on leaves, branches, and stems that is crucial for comprehensive forest structure mapping. Understanding how various GEDI sensor parameters, such as signal-to-noise ratio, beam type, and acquisition time, affect PAI accuracy is crucial for effectively leveraging GEDI data and avoiding high-error footprints in wall-to-wall forest structure mapping applications. This evaluation work by Jia et al., 2025 is vital for the effective application and interpretation of GEDI’s PAI, canopy cover, and understory PAI products across different forest types in North America, allowing future researchers to select highest quality data for future structural studies.
  • A study by Marselis et al., 2022 attempts to explain variation in tree species richness in natural forests using GEDI-derived forest structure data. Understanding the role of forest structure in driving patterns of tree species richness is critically important for biodiversity conservation efforts and global ecological understanding. GEDI forest structure data was able to support identification of local tree species diversity, but was not attributed to variation of species diversity. Such variables may require multiple predictors to be investigated to explain variation, and GEDI may not be particularly useful as one of them.
  • Biophysical parameters represent highly relevant information for decision-makers in forestry practice for species and biodiversity mapping. The study recommends the synergistic use of Sentinel-2 and GEDI products for comprehensive forest monitoring in Germany, including detecting deadwood structures, monitoring succession processes, and assessing biodiversity and fire risk in post-disturbance landscapes in Central Germany (Putzenlechner et al., 2024)
  • Mapping spatial patterns of forest diversity in temperate mixed forests is important for assessing both anthropogenic and natural disturbances and their effects on biodiversity when working with limited resources. The study shows that GEDI-derived Foliage Height Diversity (FHD) metrics, when combined with optical images, demonstrate promising potential for accurately estimating forest species diversity over large areas in China, greatly improving conservation and management capabilities for forest resources. (Ren et al., 2023)
  • Schneider et al., 2020 maps the diversity of canopy structure from space, including comprehensive functional richness and beta diversity assessments. Structural diversity plays a crucial role in the carbon cycle, ecosystem services provision, plant and animal diversity maintenance, and habitat characterization with potential links to animal diversity and habitat characteristics, which could prove transformative for global ecology. The technology also helps monitor biodiversity and policy targets in North America while improving the representation of plant canopies in dynamic vegetation and land surface models.

Tailoring GEDI Data for Context

GEDI has opened new opportunities for measuring forest structure, elevation, biomass, and vegetation dynamics from spaceborne lidar. Yet, preparing GEDI data for meaningful application requires context-specific choices that vary by landscape type, data product, and end-user goal. This section synthesizes insights from around fifty published studies applying GEDI data across diverse vegetated and non-vegetated regions. By reviewing how researchers have approached data preparation, integration, and validation, the section provides a comparative view of strategies that have successfully, or unsuccessfully, translated GEDI observations into improved decision-making products. The emphasis is on sharing practical techniques, highlighting study-area constraints, and categorizing preparation pathways that users may consider when tailoring GEDI to their own applications.

The examples reviewed spanned across 2020-2025, with a greater proportion of examples from 2024-2025 to capture assessments and recommendations from the most recent GEDI data product versions. The selected examples were determined to have decision-making relevance if the studies assessed performance of GEDI products in newly applied environmental regions or locations. Additional qualifications include whether the integration of GEDI resulted in information readily input into further evaluations relating to the application purpose, such as the next step of an analysis, design of a responsive strategy, or recommended actions (could be in the form of a dataset, map, statistic, recommendation/conclusion).

Fig 1. Count of the number of publications per year between 2020-2025. These 47 publications underwent data extraction to build the following Sankey diagram figures. to summarize GEDI data processing techniques deployed for various data products, downstream analyses, application areas, and locations.

The review has a clear objective of reviewing what has been tested with regards to quality filtering techniques suggested for specific GEDI elevation and vegetation metrics or estimations, as they were reviewed over a particular land cover type or proposed downstream purpose. The focus on qualifying application categories (as information generation towards vegetation related decision-making support) was particularly important for evaluating the “outcome” of the particular combination of data processing applied over an area or interest, given the publication-specific goals of selecting the highest quality GEDI data for downstream analyses which may include wall-to-wall mapping, classification, Deep Learning methods, allometric modelling, etc. Given the scope of this training module, details on methodological approaches for the screening procedures and documentation, data extraction standards, risk and bias assessment will not be included in this module.

Results revealed patterns across the included studies regarding study designs, participant characteristics, and geographic distribution. The synthesis of findings addressed both primary and secondary outcomes, with statistical analysis including assessment of heterogeneity and, where appropriate, meta-analysis to quantify effect sizes. Risk of bias evaluation highlighted the overall quality of evidence and identified common methodological limitations that could impact the interpretation of results. Subgroup analyses were performed to explore potential sources of variation in outcomes across different populations or intervention characteristics.

Of particular interest is understanding which GEDI data variables are used across applications that are most relevant to real-world applications. The co-development team selected case studies focusing on generating new or improved products ready for direct or future land monitoring and/or management purposes.

Which GEDI data products did each of these studies use? Were the same processing methods used for each? Were there any differences or patterns of improved results when certain datasets or processing techniques were used over certain areas of the world? Or for a given ultimate purpose of the study?

The ongoing research becomes several-fold, a) continuously improving the ability of the technology and engineering feat itself (launching new and improved lidar missions, b) Calibrating, validating, versioning improvements on existing data and resulting data products (research and analysis field of interest), c) testing where, when, and how current GEDI data and its resulting data products can produce equivalent, novel, or improved products relevant for actionable use by those who already do, or could benefit from such ecosystem information. Under c) lies a slew of opportunities for questioning the capabilities of and limitations to GEDI data under specific contexts, whether those limitations be due to the impact of spatial constraints, whether the proper data processing was completed appropriately for that purpose or particular dataset. Elements of this training, such as this page, review which, how, and where GEDI data was used to offer an initial understanding of where you as a participant can start exploring GEDI as a means to improve data-driven decision making. Do you begin at playing around with a singular or multiple GEDI products? Do you explore how the metrics change over time and certain locations when different quality processing techniques are applied? Do you focus on fusing ecosystem parameters into a model? Or go back to the drawing board on whether GEDI is applicable for whichever mapping, monitoring, or management data points you are looking to acquire over an area?

Considerations for Diverse Study Areas

  • Tropical Forests: Present challenges due to dense canopies, making ground detection difficult. Algorithm setting group 5, with its lower waveform signal end threshold, is often more adapted for detecting weak ground returns in these environments (Lahssini et al, 2022).
    • dense forests additional sensitivity filters are usually essential, while in sparse canopies, relaxing some filters may help retain sample size.
  • Savannas: GEDI can accurately characterize tree heights (>3m) but struggles with shorter, sparse woody vegetation (<2.34m) due to pulse width limitations. Leaf-on conditions are critical for accurate assessment (Li et al., 2023).
  • Mountainous/Steep Terrain: These areas pose significant challenges due to increased waveform complexity and geolocation errors. Stricter slope filters or slope-adaptive methods may be necessary (McClure et al., 2024)(Ilangakoon et al., 2021).
  • Urban Areas: GEDI can be used for building height estimation when combined with time-series imagery and topographic information. (Ma et al., 2024) Specialized filtering may be needed to distinguish built-up areas.
  • Croplands: GEDI signals are informative. RF models trained on multiple RHs can classify tall and short crops effectively. View angle filtering is important.(Tomasso et al., 2023)

Vegetation Type and Phenology

  • Dense tropical/temperate forests:
    • Full-power beams + high sensitivity (≥0.95–0.98).
    • Consider ASG 5 for weak ground returns.
    • Filtering may reduce density; multi-sensor fusion with SAR or ALS helps.
  • Sparse vegetation / savannas / shrublands:
    • GEDI struggles below ~3 m vegetation height (due to pulse width).
    • Mitigation: integrate GEDI with optical or SAR (Sentinel-2, Sentinel-1).
      Possible to exclude very low-stature vegetation (<2–3 m) if focusing on taller vegetation.
  • Agricultural crops:
    • GEDI can distinguish tall vs. short crops but underestimates tall crops in low biomass conditions.
    • Integrate with vegetation indices (e.g., GCVI) to avoid misclassification.
  • Deciduous/phenological effects:
    • Use leaf-on acquisitions for canopy structure.
    • Leaf-off conditions introduce underestimation, especially in savannas/woodlands.
  • Phenology / Seasonality:
    • For canopy structure: use leaf-on data.
    • For deciduous forests: exclude leaf_off_flag = 1, unless models are robust to RH98-only predictors.

Derived Metrics & Product-Specific Filtering

  • Canopy Height (CHM)
    • RH98 or RH95 is preferred preferred metric (RH100 is typically noisy)
      • Remove unrealistic heights (80-100m), depending on the application.
    • For mapping CHM in tall or dense forests, apply sensitivity >0.98 and use power beams only.
  • Aboveground Biomass (AGBD, L4A/L4B)
    • Parametric OLS models link RH metrics to biomass.
    • Apply stratification by PFT and geographic region
    • Filter by relative standard error: (agbd_se / agbd * 100), removing > 50%.
    • Consider where the total number of shots used could influence results. For L4B, hybrid inference is used where criteria of greater than 2 shots per 1 km cell are not met.

  • LAI / PAI

    • Filters: `l2a_quality_flag = 1`, `l2b_quality_flag = 1`, `degrade_flag = 0`, no water/urban/snow, leaf_off_flag = 0.
    • ρv/ρg ratio (canopy-to-ground reflectance): assumed =1.5 in L2B, but biome-specific calibration improves accuracy (esp. needleleaf, wetlands).
    • Clumping index (Ω): GEDI assumes Ω=1 → future correction needed for true LAI.
    • Seasonal bias: leaf-off underestimation due to static ρv/ρg.

Forest Change, Recovery & Understory

  • Temporal Consistency: Ensure acquisition and reference data (e.g., ALS) are close in time.
  • Edges & Heterogeneity:
    • Errors increase in small patches, gaps, and mixed forests.
    • Option: discard shots not fully contained in forest polygons, or exclude canopy cover <70%.
  • Understory Mapping:
    • Use phenology contrasts (leaf-on vs. leaf-off) to separate overstory/understory.
    • Filters: full-power beams, sensitivity >0.90.
    • Limitations: unreliable below ~5 m height, struggles with <3 m woody cover.
    • Use derivative RH metrics (e.g., mRH, vmRH) to capture sub-canopy structure.
    • Consider L1B waveform processing for vertical profile detail.

Scenario-Based Filtering Examples

  1. Dense Tropical Forests (e.g., Amazonia):
    • quality_flag=1 + degrade_flag=0
    • Sensitivity ≥0.98
    • Power beams only
    • Slope <15°–20°
      DEM check (>30 m exclusion)
    • Possible ASG 5 selection
    • Fusion with SAR to reduce canopy saturation effects
  2. Savannas / Sparse Woodlands:
    • quality_flag=1 + degrade_flag=0
    • Sensitivity ≥0.90 (relaxed)
    • All beams retained
    • Exclude vegetation <2–3 m if irrelevant
    • Use leaf-on only
    • Fusion with optical/SAR for low-biomass areas
  3. Temperate Deciduous Forests:
    • quality_flag=1 + degrade_flag=0
    • Sensitivity ≥0.95
    • Prefer nighttime acquisitions
    • Use leaf-on periods
    • DEM checks >50 m exclusion
    • Validate with ALS for calibration
  4. Steep Terrain (mountainous forests or croplands):
    • quality_flag=1 + degrade_flag=0
    • Sensitivity ≥0.95 (higher if dense canopy)
    • Exclude slopes >20–30° (>6° in croplands)
    • DEM difference >30–50 m exclusion
    • Aggregate to 1 km for biomass estimates

Integration with Other Data Sources

  • ALS (airborne lidar):
    • Validate and calibrate GEDI-derived metrics.
    • Use simulators to estimate GEDI-like footprints for bias correction.
  • Optical + SAR fusion:
    • Sentinel-2, Landsat, PlanetScope (optical) + Sentinel-1, PALSAR (SAR).
    • Machine learning (RF, XGBoost, CNNs, etc.) for wall-to-wall mapping.
  • Interdependency: GEDI’s sparse sampling requires fusion for landscape-scale mapping; complementary sensors also mitigate GEDI’s weaknesses in low biomass, sparse vegetation, or cloudy regions.

Machine Learning Aided Filtering

  • Incorporate confidence layers from ML/DL models to flag/remove low-certainty measurements e.g., ≥ 0.8.

The review’s findings have shown that the current state of applications examples for GEDI are quite diverse. Recommended processing techniques and expected high level performance geographic areas actively study the variability in GEDI’s datasets and provide solid solutions and workflows for how to work with and evaluate such challenges. The concluding recommendations, while specific and evidence-based over particular study areas, suggest steps for evaluating the true impact of several confounding and independent variables within the constraints of an real-world application and desired product development. However, several limitations were identified across the literature, where there remains inconclusive, or infrequently studies and evaluations of GEDI over non-tropical forests and ecosystems. Where there were more diverse examples with concrete solutions, the concreteness of the findings were so closely related to the application context, making it difficult to tease out more generalizable or transferrable methods to other regions and/or application areas, whether within or external to ecosystem related projects. , including potential publication bias, heterogeneity across studies, and constraints on generalizability to broader populations. Given the areas of active research are in quality improvements and generating new products based on data fusion and extensive modelling and uncertainty outputs, applied researchers should continue to deploy specified evaluation of the representativeness of the GEDI datasets and test across the most common recommendations to ensure the best use of the data, but also contribute to the mission by way of novel use and/or detailed review of its performance. It is expected that with continued examples of applications enabled by a reference guide such as this, the continued areas of active research can potentially ingest such global and localized takeaways into their improvements of the data products and/or design of future spaceborne lidar missions. This can only be made possible, if more specific examples and application gaps are filled by users with decision-making purposes in mind when creating outputs with GEDI, and contributing to the scientific community or communicating the use of its technology with its source. Facilitating access, exploration workflows, and application workflow reviews, such as found in this training, are aimed to aid in such uptake and feedback loops.

The results and recommendations should remain highly contextual, and be treated as comparative recommendations to help you make sense of, and strategically plan GEDI data exploration and analyses over your own area of interest.

The review of study area specific GEDI applications demonstrates both the growing potential and the current limits of GEDI integration. While researchers have developed a wide array of preparation strategies, no universal method has emerged, emphasizing that each application requires context-sensitive adjustments. The examples summarized provide valuable starting points but also illustrate the need for additional testing, innovation, and validation across new environments and land-cover types. Users are encouraged to view GEDI as a flexible yet evolving resource: one where thoughtful experimentation and the publication of novel approaches can contribute to a stronger, shared foundation for future applications in science, management, and policy.

Summary of What GEDI Has to Offer

GEDI is the first spaceborne, full-waveform lidar system optimized for estimating vegetation structure. It provides billions of laser shots collected annually gathering critical three-dimensional information on canopy height, foliage and woody structure, and AGBD. For decision-making in climate policy, conservation, and land-use management, GEDI offers unique advantages that complement traditional field surveys and other remote sensing methods. Field surveys, while highly accurate and validate remote sensing products, are considered costly, labor-intensive, and due to this, have limited spatial and temporal coverage. With proper quality filtering and/or calibration, GEDI can complement or build upon existing field surveys by providing globally consistent structural samples. GEDI integration increases sample size and land cover representativeness, enabling large-scale inference. GEDI is especially effective at detecting vegetation height and structural changes not easily captured by other sensors, and has been shown to improve degradation mapping, forest health and growth patterns that relate to disturbance monitoring, carbon quantification, or biodiversity assessments, for example.

Compared to other remote sensing technologies, GEDI fills critical observational gaps. Optical systems such as Landsat and Sentinel-2 provide wall-to-wall coverage but lack the ability to capture vertical structure due to occlusion and saturation in dense canopies. SAR sensors do penetrate vegetation, but are sensitive to moisture and roughness, and cannot reliably capture detailed canopy structure. Airborne and UAV lidar provide highly accurate, fine-scale structural maps, but their use is limited by high cost and restricted geographic coverage. GEDI, by contrast, has the capacity to penetrate canopies and record vertical structure reduces spectral saturation effects seen in optical data, while data fusion with optical or radar observations provides a more complete picture of ecosystem structure, health, and function.

Summary of Applications Challenges to Mitigate

GEDI also faces notable challenges and limitations. As a sampling mission, GEDI collects 25 m diameter footprints rather than providing wall-to-wall maps. While gridded products exist, their standard 1 km resolution is too coarse for many localized applications, particularly in smallholder agricultural systems or fragmented landscapes. Temporal sampling is irregular, with gaps caused by various conditions which may limit its ability to track disturbances or recovery consistently over the same location. Its geolocation uncertainty can be particularly impactful over heterogeneous landscapes or along forest edges. GEDI also struggles with specific conditions: low-stature vegetation (<3–5 m), very dense tropical forests where ground returns are weak, and steep slopes that mix canopy and ground signals. Beam sensitivity, waveform algorithm settings, and globally applied algorithmic assumptions in product generation introduce additional uncertainties for specific ecosystems. These limitations underscore the importance of careful data filtering, algorithm refinement, and integration with complementary datasets.

Active areas of research continue to focus on calibration, validation, and model development to extend GEDI’s utility. Validation studies against airborne lidar and field plots do show strong performance in dense forests, but other studies have found reduced accuracy in sparse or complex vegetation. Environmental conditions such as phenology, cloud cover, and acquisition time also affect performance, many studies evaluate several methods for improved results. Data fusion approaches, especially when combining GEDI samples with optical and SAR, are being advanced through machine learning models. This has direct applications in carbon accounting, biodiversity modeling, wildfire risk assessment, flood monitoring, agricultural crop mapping, and urban climate resilience planning.

Despite its sampling nature and operational constraints, GEDI provides a transformative resource for global ecosystem monitoring. Its freely available data, global reach, and synergy with customizable coding and cloud computing platforms make it cost-effective and accessible for diverse users, from policymakers to local land managers. While GEDI cannot independently provide continuous maps or fully resolve fine-scale heterogeneity alone, when fused with other sensors and combined with field data the capacity to track ecosystem dynamics, estimate carbon stocks, and inform evidence-based environmental decision-making enhances.

  Advantages Disadvantages or Challenges to Overcome
Measurement type Direct 3D canopy structure and derived biomass estimations. Limited in low-stature or very dense forests (where ground returns weak).
Accuracy Validation studies provide insight into the regions, time periods, and strategies for best results. Errors increase on steep slopes, forest edges, or heterogeneous landscapes.
Coverage Global sampling at 25 m footprints with billions of shots annually. Sample-based (not wall-to-wall); gridded products typically 1 km resolution, too coarse for local use. Quality filtering may greatly reduce sample size.
Comparison to optical Overcomes spectral saturation typically occurring with optical imagery, penetrates canopy to capture vertical structure. Optical is still needed for species, phenology, and cover type mapping.
Comparison to SAR Provides structural detail that SAR cannot reliably capture. Cannot penetrate clouds, and SAR is still needed for ecosystem applications.
Comparison to airborne lidar Global reach at low cost and offers consistent methodology. Lower spatial resolution and accuracy than airborne lidar and is less flexible when prioritizing temporal overlaps.
Integration with field data Expands sample size, increases representativeness, enables calibration of biomass models. Ground plots unevenly distributed; mismatch between plot size and GEDI footprint
Applications Carbon accounting, biodiversity modeling, forest degradation detection, habitat mapping, wildfire/flood risk Sampling gaps, geolocation uncertainty, application specific sensitivity, and computational demands.
Operational considerations Freely available, processed into accessible products. Finite mission lifespan on the ISS, and no guaranteed long-term continuity aside from future planned missions like EDGE.

So what?

GEDI has transformed global ecosystem monitoring by providing the first consistent, spaceborne lidar measurements of forest structure and biomass. For policymakers, the key value lies in reducing uncertainty in global carbon accounting, improving transparency for climate agreements (e.g. REDD+), and detecting degradation processes that other satellites miss. However, GEDI is a sampling mission, not a wall-to-wall mapper, and its lifespan is tied to the International Space Station.

  • GEDI data can strengthen national greenhouse gas inventories and reporting under the Paris Agreement by improving estimates of forest carbon stocks and changes.
  • Integration of GEDI with optical, radar, and field survey data is essential for reliable, high-resolution products that support land-use planning, biodiversity conservation, and climate resilience.
  • Investment in data continuity and successor missions is needed to ensure long-term monitoring capacity, since GEDI’s lifespan is limited.
  • Capacity-building and data-sharing platforms can democratize GEDI use, supporting decision-making by developing countries with limited technical resources.

In short, GEDI is game-changing, but is an interim step. Its biggest policy relevance is not as a standalone solution, but as a catalyst for integrated monitoring systems that enable robust, evidence-based ecosystem management and policy.

Curtosy of EarthRISE at NASA Marshall Space Flight Center and the Lab for Applied Scienecs at the University of Alabama in Huntsville