Amelia Holcomb
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Amelia Holcomb has a total of 73 co-authors. Here are the top co-authors by number of shared publications:
- Simon Besnard
GFZ Helmholtz Centre for Geosciences, Potsdam, Germany
4 shared publications - Srinivasan Keshav
University of Cambridge, Cambridge, United Kingdom
4 shared publications David A. Coomes
2 shared publications- David A. Gibbs
World Resources Institute, Washington DC, USA
2 shared publications - Johannes Reiche
Laboratory of Geo-Information Science and Remote Sensing, Wageningen University & Research, Droevendaalsesteeg 3, 6708 PB Wageningen, The Netherlands
2 shared publications Martin Herold
2 shared publications
Latest Publications
- Open Access
- 18 September 2026
- Published by Elsevier in SSRN Electronic Journal
Optimizing GEDI observations pairing for localized forest structure change assessment
Spaceborne LiDAR, particularly the Global Ecosystem Dynamics Investigation (GEDI) mission, provides direct measurements of vegetation vertical structure, which is closely linked to biomass and carbon stocks. However, its sparse sampling combined with no guaranteed repeat cycles limits the assessment of disturbance-related structural change within individual patches. Change estimates based on GEDI pairs, two observations acquired at different locations and times matched to represent the same location, are affected by three main sources of error: pre-existing structural differences between paired footprints, limited pair availability within disturbance patches, and temporal mismatch with reference data or post-disturbance dynamics. To overcome these limitations, we propose a feature-space pairing algorithm that combines spatial proximity with optical Sentinel-2, Sentinel-1 Synthetic Aperture Radar (SAR), and topographic Digital Elevation Model (DEM) covariates to identify structurally coherent GEDI pairs beyond the short-distance search radius used in previous studies (e.g. 40 m). We evaluated the approach in undisturbed forests and five large-scale fire events in Spain using ALS-simulated waveforms and multitemporal ALS surveys as reference data. With our method, we were able to reduce the sampling variance of localized change estimates by up to 46% (and up to 27% for the top-of-canopy) and match the agreement of conventional pairing at a finer aggregation scale (600 m rather than 900 m). These improvements followed from mitigating three limitations of GEDI-pair change assessment. In undisturbed forests, feature-space pairing reduced structural dissimilarity relative to spatial-proximity pairing (greatest at a geographic weight of w = 0.75), keeping pairs as structurally similar as near spatial-proximity pairs even at separations of up to 400 m. By retaining this similarity at larger distances, we could relax the spatial-proximity constraint and enlarge the pool of candidate pairs. In fire-affected areas, this increased the number of available pairs by up to 80%, improved sampling of the full range of post-disturbance structural change, and increased temporal proximity to the ALS reference survey. Together, these results demonstrate that feature-space pairing enables more localized estimates of forest structural change from spaceborne LiDAR and can support improved quantification of biomass loss after disturbance.
- 18 September 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessOptimizing GEDI observations pairing for localized forest structure change assessment
Spaceborne LiDAR, particularly the Global Ecosystem Dynamics Investigation (GEDI) mission, provides direct measurements of vegetation vertical structure, which is closely linked to biomass and carbon stocks. However, its sparse sampling combined with no guaranteed repeat cycles limits the assessment of disturbance-related structural change within individual patches. Change estimates based on GEDI pairs, two observations acquired at different locations and times matched to represent the same location, are affected by three main sources of error: pre-existing structural differences between paired footprints, limited pair availability within disturbance patches, and temporal mismatch with reference data or post-disturbance dynamics. To overcome these limitations, we propose a feature-space pairing algorithm that combines spatial proximity with optical Sentinel-2, Sentinel-1 Synthetic Aperture Radar (SAR), and topographic Digital Elevation Model (DEM) covariates to identify structurally coherent GEDI pairs beyond the short-distance search radius used in previous studies (e.g. 40 m). We evaluated the approach in undisturbed forests and five large-scale fire events in Spain using ALS-simulated waveforms and multitemporal ALS surveys as reference data. With our method, we were able to reduce the sampling variance of localized change estimates by up to 46% (and up to 27% for the top-of-canopy) and match the agreement of conventional pairing at a finer aggregation scale (600 m rather than 900 m). These improvements followed from mitigating three limitations of GEDI-pair change assessment. In undisturbed forests, feature-space pairing reduced structural dissimilarity relative to spatial-proximity pairing (greatest at a geographic weight of w = 0.75), keeping pairs as structurally similar as near spatial-proximity pairs even at separations of up to 400 m. By retaining this similarity at larger distances, we could relax the spatial-proximity constraint and enlarge the pool of candidate pairs. In fire-affected areas, this increased the number of available pairs by up to 80%, improved sampling of the full range of post-disturbance structural change, and increased temporal proximity to the ALS reference survey. Together, these results demonstrate that feature-space pairing enables more localized estimates of forest structural change from spaceborne LiDAR and can support improved quantification of biomass loss after disturbance.
- Open Access
- 3 July 2026
- Published by American Association for the Advancement of Science (AAAS) in Science Advances
- Vol. 12 (27) , eadz1923
A meta-analysis of carbon losses and gains from tropical moist forest degradation and regeneration
Aboveground carbon (AGC) fluxes from deforestation and subsequent regrowth in tropical moist forest (TMF) are increasingly well characterized, but carbon losses and gains following partial disturbance are uncertain. We synthesized 146 studies quantifying postdisturbance AGC changes relative to undisturbed forests across TMF. Immediate AGC losses (mean ± 1 SD; 2.5 ± 2.3 years after disturbance) following partial anthropogenic disturbances were greatest for forest fires (49 ± 26%), selective logging (34 ± 20%), and edge effects (31 ± 19%). Higher-frequency and -intensity disturbances significantly increased carbon loss. After 20 years of regeneration, AGC stock was higher in recovering degraded forests (41 to 117%) compared to secondary regrowth forests after complete deforestation (1 to 74%), indicating greater regeneration potential when forest structure is preserved. Our compiled database and associated meta-analysis improve accuracy and completeness for carbon inventory reporting and modeling. Substantial AGC losses and gains from distinct degradation and recovery processes are now better characterized, serving as an evidence base for policies to halt degradation and foster recovery for climate mitigation.
- 3 July 2026
- Published by American Association for the Advancement of Science (AAAS) in Science Advances
- Vol. 12 (27) , eadz1923
Open AccessA meta-analysis of carbon losses and gains from tropical moist forest degradation and regeneration
Aboveground carbon (AGC) fluxes from deforestation and subsequent regrowth in tropical moist forest (TMF) are increasingly well characterized, but carbon losses and gains following partial disturbance are uncertain. We synthesized 146 studies quantifying postdisturbance AGC changes relative to undisturbed forests across TMF. Immediate AGC losses (mean ± 1 SD; 2.5 ± 2.3 years after disturbance) following partial anthropogenic disturbances were greatest for forest fires (49 ± 26%), selective logging (34 ± 20%), and edge effects (31 ± 19%). Higher-frequency and -intensity disturbances significantly increased carbon loss. After 20 years of regeneration, AGC stock was higher in recovering degraded forests (41 to 117%) compared to secondary regrowth forests after complete deforestation (1 to 74%), indicating greater regeneration potential when forest structure is preserved. Our compiled database and associated meta-analysis improve accuracy and completeness for carbon inventory reporting and modeling. Substantial AGC losses and gains from distinct degradation and recovery processes are now better characterized, serving as an evidence base for policies to halt degradation and foster recovery for climate mitigation.
- Open Access
- 14 March 2026
- Published by Elsevier in Forest Ecology and Management
- Vol. 611
The phenology of European forests as seen by MODIS Leaf Area Index and GEDI Plant Area Index: Toward an integrated approach
- 14 March 2026
- Published by Elsevier in Forest Ecology and Management
- Vol. 611
Open AccessThe phenology of European forests as seen by MODIS Leaf Area Index and GEDI Plant Area Index: Toward an integrated approach
- Open Access
- 26 September 2025
- Published by The Open Journal in Journal of Open Source Software
- Vol. 10 (113) , 8593
gediDB: A toolbox for processing and providing Global Ecosystem Dynamics Investigation (GEDI) L2A-B and L4A-C data
The Global Ecosystem Dynamics Investigation (GEDI) mission provides spaceborne LiDAR observations that are essential for characterising Earth's forest structure and carbon dynamics. However, GEDI datasets are distributed as complex HDF5 granules, which pose significant challenges for efficient, large-scale data processing and analysis. To overcome these hurdles, we developed gediDB, an open-source Python standardised Application Programming Interface (API) that streamlines both the processing and querying of GEDI Level 2A-B and Level 4A-C datasets. Built on the optimised multidimensional array database TileDB, gediDB enables operational-scale processing, rapid spatial and temporal queries, and reproducible LiDAR-based analyses of forest biomass, carbon stocks, and structural change. gediDB centres on two modules:
- 26 September 2025
- Published by The Open Journal in Journal of Open Source Software
- Vol. 10 (113) , 8593
Open AccessgediDB: A toolbox for processing and providing Global Ecosystem Dynamics Investigation (GEDI) L2A-B and L4A-C data
The Global Ecosystem Dynamics Investigation (GEDI) mission provides spaceborne LiDAR observations that are essential for characterising Earth's forest structure and carbon dynamics. However, GEDI datasets are distributed as complex HDF5 granules, which pose significant challenges for efficient, large-scale data processing and analysis. To overcome these hurdles, we developed gediDB, an open-source Python standardised Application Programming Interface (API) that streamlines both the processing and querying of GEDI Level 2A-B and Level 4A-C datasets. Built on the optimised multidimensional array database TileDB, gediDB enables operational-scale processing, rapid spatial and temporal queries, and reproducible LiDAR-based analyses of forest biomass, carbon stocks, and structural change. gediDB centres on two modules:
- Open Access
- 18 March 2025
- Published by Copernicus GmbH
Understorey density modulates CO2 anomaly during climate extremes.
Understanding how forest ecosystems respond to climate extremes, such as droughts and heatwaves, is critical for predicting carbon cycle dynamics across spatio-temporal scales. This study tests the hypothesis that understorey vegetation density significantly modulates the magnitude and direction of CO2 flux anomalies during extreme climatic events. Using data from the GEDI spaceborne LiDAR mission, we derived structural metrics such as the vertical distribution of vegetation (e.g., proportion of lower vs. upper canopy vegetation) and inter-canopy spacing. These structural features were integrated with carbon flux observations from FLUXNET towers and globally using the FLUXCOM-X data-driven modelling framework.We aim to investigate whether forests with well-developed understorey layers exhibit amplified CO2 flux anomalies, potentially leading to decreased carbon sink capacity during climate extremes. We hypothesize that increased understorey density could make ecosystems less stable during these events, potentially amplifying anomalies due to feedback such as enhanced evapotranspiration or altered microclimatic conditions. By linking pre-event CO2 flux entropy—a measure of baseline variability and predictability in carbon flux dynamics—to flux anomalies during climate extremes, this study will assess how vertical forest structure and understorey characteristics influence these responses, with the potential to upscale findings to continental scales.This research bridges multi-scale datasets and innovative analytical methods to address key challenges in understanding biosphere-atmosphere interactions during climate extremes. The findings could provide novel insights into how vertical vegetation profiles shape ecosystem resilience to a rapidly changing climate, potentially improving the representation of forest structure and carbon dynamics in Earth System Models across scales.
- 18 March 2025
- Published by Copernicus GmbH
Open AccessUnderstorey density modulates CO2 anomaly during climate extremes.
Understanding how forest ecosystems respond to climate extremes, such as droughts and heatwaves, is critical for predicting carbon cycle dynamics across spatio-temporal scales. This study tests the hypothesis that understorey vegetation density significantly modulates the magnitude and direction of CO2 flux anomalies during extreme climatic events. Using data from the GEDI spaceborne LiDAR mission, we derived structural metrics such as the vertical distribution of vegetation (e.g., proportion of lower vs. upper canopy vegetation) and inter-canopy spacing. These structural features were integrated with carbon flux observations from FLUXNET towers and globally using the FLUXCOM-X data-driven modelling framework.We aim to investigate whether forests with well-developed understorey layers exhibit amplified CO2 flux anomalies, potentially leading to decreased carbon sink capacity during climate extremes. We hypothesize that increased understorey density could make ecosystems less stable during these events, potentially amplifying anomalies due to feedback such as enhanced evapotranspiration or altered microclimatic conditions. By linking pre-event CO2 flux entropy—a measure of baseline variability and predictability in carbon flux dynamics—to flux anomalies during climate extremes, this study will assess how vertical forest structure and understorey characteristics influence these responses, with the potential to upscale findings to continental scales.This research bridges multi-scale datasets and innovative analytical methods to address key challenges in understanding biosphere-atmosphere interactions during climate extremes. The findings could provide novel insights into how vertical vegetation profiles shape ecosystem resilience to a rapidly changing climate, potentially improving the representation of forest structure and carbon dynamics in Earth System Models across scales.
- Open Access
- 18 March 2025
- Published by Copernicus GmbH
Too few, too many, or just the right number of estimates? Goldilock’s problem on post-disturbance carbon emissions and removal factors in tropical forests.
Tropical forests are dynamic ecosystems shaped by deforestation, degradation, and recovery processes, with consequences for the carbon cycle. While emissions from deforestation have been well understood and quantified, information on emissions from degradation such as fire, logging, windrow and drought remain relatively poorly quantified, reflecting the complexity of these processes in space and time. Similarly, the carbon recovery potential of degraded forests is understudied compared to secondary forests regrowing after deforestation. Closing these knowledge gaps is crucial to reduce uncertainties in estimates of the tropical carbon budget and for addressing the priorities of international climate policies, which increasingly emphasize the value of protecting and restoring forests, without which we cannot constrain global warming to critical limits.In recent years, research on carbon emissions and removals in tropical forests has surged, driven in part by advancements in Earth Observation. Here we synthesize these approaches with the aim to bring clarity and advance our understanding on aboveground carbon (AGC) emission and removal factors applicable for tropical moist forests. We contextualise the current studies, highlighting where there are sufficient data estimates to quantify emissions and removals post-disturbance, and where specific kinds of estimates are lacking.Our synthesis of 66 studies of AGC loss due to disturbance shows emission estimates vary widely across disturbance types: average AGC losses are 3% (range 1–4%) for extreme drought, 27% (range 3–75%) for selective logging, and 52% (range 9–83%) for fire, relative to nearby und previously undisturbed forest. Our analysis underscores the need to account for disturbance severity, frequency and the cumulative effects of interacting disturbances to reduce variability between emissions estimates.For AGC recovery, our synthesis of 68 studies indicates that degraded forests regained 41–117% of AGC within 20 years relative to undisturbed forests; significantly higher than forests regrowing from deforestation, which regained between 1% and 74% of undisturbed forest AGC. Younger recovering forests ( 20 years). In the Amazon region, where we have the greatest number of field site and region-specific remote sensing data, we see good agreement between field- and satellite- derived regrowth rate estimates. Remote sensing data therefore has the potential to fill the gaps in our spatial knowledge where field data is limited.Our results also highlight some of the major gaps that still exist to provide long-lasting and relevant information into the policy and wider carbon budget science domain. Key research needs include: (i) reducing the variability of emission factors within disturbance types by further stratifying according to disturbance severity, frequency and co-occuring disturbances, (ii) addressing the research bias towards the Americas, particularly the Amazon, by expanding studies to areas where there are currently fewer estimates. Finally, we call for a more integrated approach between research focusing on deforestation, degradation and regrowth, recovery and consider these processes as interconnected, co-occurring and influencing each other in space and time.
- 18 March 2025
- Published by Copernicus GmbH
Open AccessToo few, too many, or just the right number of estimates? Goldilock’s problem on post-disturbance carbon emissions and removal factors in tropical forests.
Tropical forests are dynamic ecosystems shaped by deforestation, degradation, and recovery processes, with consequences for the carbon cycle. While emissions from deforestation have been well understood and quantified, information on emissions from degradation such as fire, logging, windrow and drought remain relatively poorly quantified, reflecting the complexity of these processes in space and time. Similarly, the carbon recovery potential of degraded forests is understudied compared to secondary forests regrowing after deforestation. Closing these knowledge gaps is crucial to reduce uncertainties in estimates of the tropical carbon budget and for addressing the priorities of international climate policies, which increasingly emphasize the value of protecting and restoring forests, without which we cannot constrain global warming to critical limits.In recent years, research on carbon emissions and removals in tropical forests has surged, driven in part by advancements in Earth Observation. Here we synthesize these approaches with the aim to bring clarity and advance our understanding on aboveground carbon (AGC) emission and removal factors applicable for tropical moist forests. We contextualise the current studies, highlighting where there are sufficient data estimates to quantify emissions and removals post-disturbance, and where specific kinds of estimates are lacking.Our synthesis of 66 studies of AGC loss due to disturbance shows emission estimates vary widely across disturbance types: average AGC losses are 3% (range 1–4%) for extreme drought, 27% (range 3–75%) for selective logging, and 52% (range 9–83%) for fire, relative to nearby und previously undisturbed forest. Our analysis underscores the need to account for disturbance severity, frequency and the cumulative effects of interacting disturbances to reduce variability between emissions estimates.For AGC recovery, our synthesis of 68 studies indicates that degraded forests regained 41–117% of AGC within 20 years relative to undisturbed forests; significantly higher than forests regrowing from deforestation, which regained between 1% and 74% of undisturbed forest AGC. Younger recovering forests ( 20 years). In the Amazon region, where we have the greatest number of field site and region-specific remote sensing data, we see good agreement between field- and satellite- derived regrowth rate estimates. Remote sensing data therefore has the potential to fill the gaps in our spatial knowledge where field data is limited.Our results also highlight some of the major gaps that still exist to provide long-lasting and relevant information into the policy and wider carbon budget science domain. Key research needs include: (i) reducing the variability of emission factors within disturbance types by further stratifying according to disturbance severity, frequency and co-occuring disturbances, (ii) addressing the research bias towards the Americas, particularly the Amazon, by expanding studies to areas where there are currently fewer estimates. Finally, we call for a more integrated approach between research focusing on deforestation, degradation and regrowth, recovery and consider these processes as interconnected, co-occurring and influencing each other in space and time.
- Open Access
- 23 August 2024
- Published by Elsevier in Ecological Informatics
- Vol. 82
An app for tree trunk diameter estimation from coarse optical depth maps
- 23 August 2024
- Published by Elsevier in Ecological Informatics
- Vol. 82
Open AccessAn app for tree trunk diameter estimation from coarse optical depth maps
- Open Access
- 3 April 2024
- Published by Cambridge University Press (CUP) in Cambridge Open Engage
An App for Tree Trunk Diameter Estimation from Coarse Optical Depth Maps
Trunk diameter is related to the overall health and level of carbon sequestration in a tree. Trunk diameter measurement, therefore, is a key task in both forest plot and urban settings. Unlike the traditional approach of manual measurement with a measuring tape or calipers, several recent approaches rely on sophisticated technologies such as Terrestrial Laser Scanning (TLS), LiDAR, and time-of-flight sensors that provide fine-grain depth maps, which are used for depth-assisted image segmentation in downstream processing. These technologies are supported only on specialized devices or high-end smartphones. We present a mobile application that uses coarse-grain depth maps derived from an optical sensor, and so can be run on most common Android devices. Moreover, we use a state-of-the-art deep neural network to estimate trunk diameter from an image and its corresponding coarse depth map (RGB-D). We tested our app using a dataset collected from four countries and under challenging conditions including occlusion, leaning trees, and irregular shapes and found that our algorithm has a MAE of 2.58 cm and an RMSE of 3.57 cm, which is comparable to accuracy from fine-grain depth maps. Moreover, diameter measurement using our app is more than 5 times faster than traditional manual surveying.
- 3 April 2024
- Published by Cambridge University Press (CUP) in Cambridge Open Engage
Open AccessAn App for Tree Trunk Diameter Estimation from Coarse Optical Depth Maps
Trunk diameter is related to the overall health and level of carbon sequestration in a tree. Trunk diameter measurement, therefore, is a key task in both forest plot and urban settings. Unlike the traditional approach of manual measurement with a measuring tape or calipers, several recent approaches rely on sophisticated technologies such as Terrestrial Laser Scanning (TLS), LiDAR, and time-of-flight sensors that provide fine-grain depth maps, which are used for depth-assisted image segmentation in downstream processing. These technologies are supported only on specialized devices or high-end smartphones. We present a mobile application that uses coarse-grain depth maps derived from an optical sensor, and so can be run on most common Android devices. Moreover, we use a state-of-the-art deep neural network to estimate trunk diameter from an image and its corresponding coarse depth map (RGB-D). We tested our app using a dataset collected from four countries and under challenging conditions including occlusion, leaning trees, and irregular shapes and found that our algorithm has a MAE of 2.58 cm and an RMSE of 3.57 cm, which is comparable to accuracy from fine-grain depth maps. Moreover, diameter measurement using our app is more than 5 times faster than traditional manual surveying.
- Open Access
- 6 November 2023
- Published by Elsevier in Science of Remote Sensing
- Vol. 8
Computational tools for assessing forest recovery with GEDI shots and forest change maps
- 6 November 2023
- Published by Elsevier in Science of Remote Sensing
- Vol. 8
Open AccessComputational tools for assessing forest recovery with GEDI shots and forest change maps
- Open Access
- 29 January 2023
- Published by MDPI AG in Remote Sensing
- Vol. 15 (3) , 772
Robust Single-Image Tree Diameter Estimation with Mobile Phones
Ground-based forest inventories are reliable methods for forest carbon monitoring, reporting, and verification schemes and the cornerstone of forest ecology research. Recent work using LiDAR-equipped mobile phones to automate parts of the forest inventory process assumes that tree trunks are well-spaced and visually unoccluded, or else require manual intervention or offline processing to identify and measure tree trunks. In this paper, we designed an algorithm that exploits a low-cost smartphone LiDAR sensor to estimate the trunk diameter automatically from a single image in complex and realistic field conditions. We implemented our design and built it into an app on a Huawei P30 Pro smartphone, demonstrating that the algorithm has low enough computational costs to run on this commodity platform in near real-time. We evaluated our app in 3 different forests across 3 seasons and found that in a corpus of 97 sample tree images, our app estimated the trunk diameter with a RMSE of 3.7 cm ( = 0.97; 8.0% mean absolute error) compared to manual DBH measurement. It achieved a 100% tree detection rate while reducing the surveyor time by up to a factor of 4.6. Our work contributes to the search for a low-cost, low-expertise alternative to terrestrial laser scanning that is nonetheless robust and efficient enough to compete with manual methods. We highlight the challenges that low-end mobile depth scanners face in occluded conditions and offer a lightweight, fully automatic approach for segmenting depth images and estimating the trunk diameter despite these challenges. Our approach lowers the barriers to in situ forest measurements outside of an urban or plantation context, maintaining a tree detection and accuracy rate comparable to previous mobile phone methods even in complex forest conditions.- 29 January 2023
- Published by MDPI AG in Remote Sensing
- Vol. 15 (3) , 772
Open AccessRobust Single-Image Tree Diameter Estimation with Mobile Phones
Ground-based forest inventories are reliable methods for forest carbon monitoring, reporting, and verification schemes and the cornerstone of forest ecology research. Recent work using LiDAR-equipped mobile phones to automate parts of the forest inventory process assumes that tree trunks are well-spaced and visually unoccluded, or else require manual intervention or offline processing to identify and measure tree trunks. In this paper, we designed an algorithm that exploits a low-cost smartphone LiDAR sensor to estimate the trunk diameter automatically from a single image in complex and realistic field conditions. We implemented our design and built it into an app on a Huawei P30 Pro smartphone, demonstrating that the algorithm has low enough computational costs to run on this commodity platform in near real-time. We evaluated our app in 3 different forests across 3 seasons and found that in a corpus of 97 sample tree images, our app estimated the trunk diameter with a RMSE of 3.7 cm ( = 0.97; 8.0% mean absolute error) compared to manual DBH measurement. It achieved a 100% tree detection rate while reducing the surveyor time by up to a factor of 4.6. Our work contributes to the search for a low-cost, low-expertise alternative to terrestrial laser scanning that is nonetheless robust and efficient enough to compete with manual methods. We highlight the challenges that low-end mobile depth scanners face in occluded conditions and offer a lightweight, fully automatic approach for segmenting depth images and estimating the trunk diameter despite these challenges. Our approach lowers the barriers to in situ forest measurements outside of an urban or plantation context, maintaining a tree detection and accuracy rate comparable to previous mobile phone methods even in complex forest conditions.