Accurate Simulation of Both Sensitivity and Variability for Amazonian Photosynthesis: Is It Too Much to Ask?

Gallup, Sarah; Baker, Ian; Gallup, John Luke; Restrepo‐Coupe, Natalia; Haynes, Katherine; Geyer, N. M.; Denning, Scott

doi:10.1029/2021ms002555

Cited by 3 publications

(5 citation statements)

References 191 publications

(296 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The best performing models were those that were exposed to training data from all the sites and then "honed" with a limited number of training iterations on the site to be predicted on. This suggests that our trained models (provided freely with the Python package 7 ) could be transferred to a new site with very little manual data or training iterations. By building a broader database of delineated tree crowns it may be possible to further boost the performance.…”

Section: Discussionmentioning

confidence: 99%

“…The R101-FPN backbone consists of a 101 layer deep ResNet (71) module with a Feature Pyramid Network (72) module. The initial model weights were generated from pre-training of the network on the ImageNet dataset 7 . It is possible to "freeze" the backbone to different depths depending on the amount of flexibility the user wants to introduce in moving away from the pre-trained model weights.…”

Section: Supplementary Note 4: Model Architecture Tuning and Trainingmentioning

confidence: 99%

“…However, the strength of the carbon 50 sink is diminishing as a result of global warming (2,3) and there are concerns that forests are reaching a tipping point beyond which they could switch irreversibly to open savanna systems (4). However, forecasting the future of tropical forests is challenging, because little is known about the ways 55 different species will respond to changing climate, or the resilience provided by that diversity (5)(6)(7)(8). To understand the likely responses of forests to further climate change, ecosystem models need to represent growth and mortality processes of individual trees more accurately than is currently the case 60 (9)(10)(11).…”

Section: Introductionmentioning

confidence: 99%

“…Forecasting the future of tropical forests is challenging, because little is known about the ways different species will respond to changing climate, or the resilience provided by that diversity (5)(6)(7)(8). To understand the likely responses of forests to further climate change, ecosystem models need to represent growth and mortality processes of individual trees more accurately than is currently the case (9)(10)(11).…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Accurate delineation of individual tree crowns in tropical forests from aerial RGB imagery using Mask R-CNN

Hickman¹,

Ball

Jackson

et al. 2022

Preprint

View full text Add to dashboard Cite

Tropical forests are a major component of the global carbon cycle and home to two-thirds of terrestrial species. Upper-canopy trees store the majority of forest carbon and can be particularly vulnerable to drought events and storms. Monitoring their growth and mortality is essential to understanding forest resilience to climate change, but large trees are underrepresented in traditional field surveys, so estimates are poorly constrained. Aerial photographs provide spectral and textural information to discriminate between tree crowns in diverse, complex tropical canopies, potentially opening the door to landscape monitoring of large trees. Here we describe a new deep convolutional neural network machine learning method, Detectree2, which builds on the Mask R-CNN computer vision framework to recognise the irregular edges of individual tree crowns from airborne RGB imagery. We trained and evaluated this model with 3,600 manually delineated tree crowns at three sites in Malaysian Borneo and one site in French Guiana. As an example application of this tool, we combined the delineations with repeat lidar surveys of the four sites to estimate the growth and mortality of upper-canopy trees. Detectree2 delineated 65,000 upper-canopy trees across 14 km^2 of aerial images. The skill of the automatic method in delineating the upper-canopy tree crowns was good (F1 = 0.71). As predicted from previous field studies, we found that growth rate declined with tree height and tall trees had higher mortality rates than intermediate-size trees. In addition, tall trees in French Guiana had higher growth and mortality rates than those in Borneo. Our approach demonstrates that machine learning methods can automatically segment trees in abundant, cheap RGB imagery. This tool has many potential applications in forest ecology and conservation, from estimating carbon stocks to monitoring forest restoration. We demonstrate its use in tracking the growth and mortality rates of upper-canopy trees at scales much larger than those achievable with field data.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Supplementary Note 4: Model Architecture Tuning and Trainingmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Accurate delineation of individual tree crowns in tropical forests from aerial RGB imagery using Mask R-CNN

Hickman¹,

Ball

Jackson

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…However, the strength of the carbon sink is diminishing as a result of global warming (Brienen et al, 2015;Hubau et al, 2020) and there are concerns that forests are reaching a tipping point beyond which they could switch irreversibly to open savanna systems (Chai et al, 2021). Forecasting the future of tropical forests is challenging because little is known about the ways different species will respond to changing climate, or the resilience provided by that diversity (Fisher et al, 2018;Gallup et al, 2021;Koven et al, 2020;Restrepo-Coupe et al, 2021). To understand the likely responses of forests to further climate change, ecosystem models need to represent growth and mortality processes of individual trees more accurately than is currently the case (Kellner et al, 2019;Piponiot et al, 2022;Zuidema & van der Sleen, 2022).…”

Section: Introductionmentioning

confidence: 99%

Accurate delineation of individual tree crowns in tropical forests from aerial RGB imagery using Mask R‐CNN

Ball

Hickman

Jackson

et al. 2023

Remote Sens Ecol Conserv

View full text Add to dashboard Cite

Tropical forests are a major component of the global carbon cycle and home to two‐thirds of terrestrial species. Upper‐canopy trees store the majority of forest carbon and can be vulnerable to drought events and storms. Monitoring their growth and mortality is essential to understanding forest resilience to climate change, but in the context of forest carbon storage, large trees are underrepresented in traditional field surveys, so estimates are poorly constrained. Aerial photographs provide spectral and textural information to discriminate between tree crowns in diverse, complex tropical canopies, potentially opening the door to landscape monitoring of large trees. Here we describe a new deep convolutional neural network method, Detectree2, which builds on the Mask R‐CNN computer vision framework to recognize the irregular edges of individual tree crowns from airborne RGB imagery. We trained and evaluated this model with 3797 manually delineated tree crowns at three sites in Malaysian Borneo and one site in French Guiana. As an example application, we combined the delineations with repeat lidar surveys (taken between 3 and 6 years apart) of the four sites to estimate the growth and mortality of upper‐canopy trees. Detectree2 delineated 65 000 upper‐canopy trees across 14 km2 of aerial images. The skill of the automatic method in delineating unseen test trees was good (F1 score = 0.64) and for the tallest category of trees was excellent (F1 score = 0.74). As predicted from previous field studies, we found that growth rate declined with tree height and tall trees had higher mortality rates than intermediate‐size trees. Our approach demonstrates that deep learning methods can automatically segment trees in widely accessible RGB imagery. This tool (provided as an open‐source Python package) has many potential applications in forest ecology and conservation, from estimating carbon stocks to monitoring forest phenology and restoration.Python package available to install at https://github.com/PatBall1/Detectree2.

show abstract

Understory plant communities show resistance to drought, hurricanes, and experimental warming in a wet tropical forest

Alonso‐Rodríguez

Wood

Torres‐Díaz

et al. 2022

Front. For. Glob. Change

View full text Add to dashboard Cite

Global climate change has led to rising temperatures and to more frequent and intense climatic events, such as storms and droughts. Changes in climate and disturbance regimes can have non-additive effects on plant communities and result in complicated legacies we have yet to understand. This is especially true for tropical forests, which play a significant role in regulating global climate. We used understory vegetation data from the Tropical Responses to Altered Climate Experiment (TRACE) in Puerto Rico to evaluate how plant communities responded to climate warming and disturbance. The TRACE understory vegetation was exposed to a severe drought (2015), 2 years of experimental warming (4°C above ambient in half of the plots, 2016–2017 and 2018–2019), and two major hurricanes (Irma and María, September 2017). Woody seedlings and saplings were censused yearly from 2015 to 2019, with an additional census in 2015 after the drought ended. We evaluated disturbance-driven changes in species richness, diversity, and composition across ontogeny. We then used Bayesian predictive trait modeling to assess how species responded to disturbance and how this might influence the functional structure of the plant community. Our results show decreased seedling richness after hurricane disturbance, as well as increased sapling richness and diversity after warming. We found a shift in species composition through time for both seedlings and saplings, yet the individual effects of each disturbance were not significant. At both ontogenetic stages, we observed about twice as many species responding to experimental warming as those responding to drought and hurricanes. Predicted changes in functional structure point to disturbance-driven functional shifts toward a mixture of fast-growing and drought-tolerant species. Our findings demonstrate that the tropical forest understory community is more resistant to climatic stressors than expected, especially at the sapling stage. However, early signs of changes in species composition suggest that, in a warming climate with frequent droughts and hurricanes, plant communities might shift over time toward fast-growing or drought-tolerant species.

show abstract

Accurate Simulation of Both Sensitivity and Variability for Amazonian Photosynthesis: Is It Too Much to Ask?

Cited by 3 publications

References 191 publications

Accurate delineation of individual tree crowns in tropical forests from aerial RGB imagery using Mask R-CNN

Accurate delineation of individual tree crowns in tropical forests from aerial RGB imagery using Mask R-CNN

Accurate delineation of individual tree crowns in tropical forests from aerial RGB imagery using Mask R‐CNN

Understory plant communities show resistance to drought, hurricanes, and experimental warming in a wet tropical forest

Contact Info

Product

Resources

About