Estimating Forest Heights and Wood Volume using a Deep Learning Approach from Gedi Waveform Data
Autor(a) principal: | |
---|---|
Data de Publicação: | 2022 |
Outros Autores: | , , , , , , |
Tipo de documento: | Artigo de conferência |
Idioma: | eng |
Título da fonte: | Repositório Institucional da UNESP |
Texto Completo: | http://dx.doi.org/10.1109/IGARSS46834.2022.9883973 http://hdl.handle.net/11449/246138 |
Resumo: | The Global Ecosystem Dynamics Investigation (GEDI) instrument, as all FW systems, relies on very sophisticated pre-processing steps to generate a priori metrics in order to accurately estimate forest characteristics, such as forest heights and wood volume. The ever-expanding volume of acquired GEDI data, which to September 2020 comprised more than 25 billion shots, and requiring more than 90 TB of storage space, raises new challenges in terms of adapted preprocessing methods for the suitable exploitation of such a huge and complex amount of LiDAR data. Therefore, to avoid metric computation, we leveraged deep learning techniques in order to estimate canopy dominant heights (Hdom) and wood volume (V) of Eucalyptus plantations over five different regions in Brazil. Performance comparisons were conducted between a convolutional neural network based model that uses GEDI waveform data and a previously used, metric based, Random Forest regressor (RF). Cross-validated results showed that the CNN based model compared well against the RF counterpart for both Hdom and V. Indeed, the RMSE on the estimation of Hdom from the CNN based model was 1.61 m with a coefficient of determination R2 of 0.90, while the RF model produced an accuracy on Hdom estimates of 1.45 m(R2=0.92). For V, CNN based estimates was 27.35 m3.ha-1(R2 of 0.88), while for RF, the RMSE was 27.60 m3.ha-1 (R2=0.88). |
id |
UNSP_345cb736e95f8f1290702b54a5771a51 |
---|---|
oai_identifier_str |
oai:repositorio.unesp.br:11449/246138 |
network_acronym_str |
UNSP |
network_name_str |
Repositório Institucional da UNESP |
repository_id_str |
2946 |
spelling |
Estimating Forest Heights and Wood Volume using a Deep Learning Approach from Gedi Waveform DataBrazilCNNDominant heightEucalyptusGEDILidarWood volumeThe Global Ecosystem Dynamics Investigation (GEDI) instrument, as all FW systems, relies on very sophisticated pre-processing steps to generate a priori metrics in order to accurately estimate forest characteristics, such as forest heights and wood volume. The ever-expanding volume of acquired GEDI data, which to September 2020 comprised more than 25 billion shots, and requiring more than 90 TB of storage space, raises new challenges in terms of adapted preprocessing methods for the suitable exploitation of such a huge and complex amount of LiDAR data. Therefore, to avoid metric computation, we leveraged deep learning techniques in order to estimate canopy dominant heights (Hdom) and wood volume (V) of Eucalyptus plantations over five different regions in Brazil. Performance comparisons were conducted between a convolutional neural network based model that uses GEDI waveform data and a previously used, metric based, Random Forest regressor (RF). Cross-validated results showed that the CNN based model compared well against the RF counterpart for both Hdom and V. Indeed, the RMSE on the estimation of Hdom from the CNN based model was 1.61 m with a coefficient of determination R2 of 0.90, while the RF model produced an accuracy on Hdom estimates of 1.45 m(R2=0.92). For V, CNN based estimates was 27.35 m3.ha-1(R2 of 0.88), while for RF, the RMSE was 27.60 m3.ha-1 (R2=0.88).Cirad Cnrs Inrae Tetis Univ Montpellier AgroParisTechUnesp Faculdade de Ciências Agronômicas, SPSuzano SA 13465-970, Estrada Limeira, 391, SPCirad Umr Eco&SolsEco&Sols Univ Montpellier Cirad Inra Ird Montpellier SupAgroUnesp Faculdade de Ciências Agronômicas, SPAgroParisTechUniversidade Estadual Paulista (UNESP)13465-970Umr Eco&SolsMontpellier SupAgroFayad, IbrahimIenco, DinoBaghdadi, NicolasGaetano, RaffaeleAlvares, Clayton Alcarde [UNESP]Stape, Jose Luiz [UNESP]Scolforo, Henrique FerracoLe Maire, Guerric2023-07-29T12:32:40Z2023-07-29T12:32:40Z2022-01-01info:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/conferenceObject7301-7304http://dx.doi.org/10.1109/IGARSS46834.2022.9883973International Geoscience and Remote Sensing Symposium (IGARSS), v. 2022-July, p. 7301-7304.http://hdl.handle.net/11449/24613810.1109/IGARSS46834.2022.98839732-s2.0-85140371370Scopusreponame:Repositório Institucional da UNESPinstname:Universidade Estadual Paulista (UNESP)instacron:UNESPengInternational Geoscience and Remote Sensing Symposium (IGARSS)info:eu-repo/semantics/openAccess2023-07-29T12:32:40Zoai:repositorio.unesp.br:11449/246138Repositório InstitucionalPUBhttp://repositorio.unesp.br/oai/requestopendoar:29462023-07-29T12:32:40Repositório Institucional da UNESP - Universidade Estadual Paulista (UNESP)false |
dc.title.none.fl_str_mv |
Estimating Forest Heights and Wood Volume using a Deep Learning Approach from Gedi Waveform Data |
title |
Estimating Forest Heights and Wood Volume using a Deep Learning Approach from Gedi Waveform Data |
spellingShingle |
Estimating Forest Heights and Wood Volume using a Deep Learning Approach from Gedi Waveform Data Fayad, Ibrahim Brazil CNN Dominant height Eucalyptus GEDI Lidar Wood volume |
title_short |
Estimating Forest Heights and Wood Volume using a Deep Learning Approach from Gedi Waveform Data |
title_full |
Estimating Forest Heights and Wood Volume using a Deep Learning Approach from Gedi Waveform Data |
title_fullStr |
Estimating Forest Heights and Wood Volume using a Deep Learning Approach from Gedi Waveform Data |
title_full_unstemmed |
Estimating Forest Heights and Wood Volume using a Deep Learning Approach from Gedi Waveform Data |
title_sort |
Estimating Forest Heights and Wood Volume using a Deep Learning Approach from Gedi Waveform Data |
author |
Fayad, Ibrahim |
author_facet |
Fayad, Ibrahim Ienco, Dino Baghdadi, Nicolas Gaetano, Raffaele Alvares, Clayton Alcarde [UNESP] Stape, Jose Luiz [UNESP] Scolforo, Henrique Ferraco Le Maire, Guerric |
author_role |
author |
author2 |
Ienco, Dino Baghdadi, Nicolas Gaetano, Raffaele Alvares, Clayton Alcarde [UNESP] Stape, Jose Luiz [UNESP] Scolforo, Henrique Ferraco Le Maire, Guerric |
author2_role |
author author author author author author author |
dc.contributor.none.fl_str_mv |
AgroParisTech Universidade Estadual Paulista (UNESP) 13465-970 Umr Eco&Sols Montpellier SupAgro |
dc.contributor.author.fl_str_mv |
Fayad, Ibrahim Ienco, Dino Baghdadi, Nicolas Gaetano, Raffaele Alvares, Clayton Alcarde [UNESP] Stape, Jose Luiz [UNESP] Scolforo, Henrique Ferraco Le Maire, Guerric |
dc.subject.por.fl_str_mv |
Brazil CNN Dominant height Eucalyptus GEDI Lidar Wood volume |
topic |
Brazil CNN Dominant height Eucalyptus GEDI Lidar Wood volume |
description |
The Global Ecosystem Dynamics Investigation (GEDI) instrument, as all FW systems, relies on very sophisticated pre-processing steps to generate a priori metrics in order to accurately estimate forest characteristics, such as forest heights and wood volume. The ever-expanding volume of acquired GEDI data, which to September 2020 comprised more than 25 billion shots, and requiring more than 90 TB of storage space, raises new challenges in terms of adapted preprocessing methods for the suitable exploitation of such a huge and complex amount of LiDAR data. Therefore, to avoid metric computation, we leveraged deep learning techniques in order to estimate canopy dominant heights (Hdom) and wood volume (V) of Eucalyptus plantations over five different regions in Brazil. Performance comparisons were conducted between a convolutional neural network based model that uses GEDI waveform data and a previously used, metric based, Random Forest regressor (RF). Cross-validated results showed that the CNN based model compared well against the RF counterpart for both Hdom and V. Indeed, the RMSE on the estimation of Hdom from the CNN based model was 1.61 m with a coefficient of determination R2 of 0.90, while the RF model produced an accuracy on Hdom estimates of 1.45 m(R2=0.92). For V, CNN based estimates was 27.35 m3.ha-1(R2 of 0.88), while for RF, the RMSE was 27.60 m3.ha-1 (R2=0.88). |
publishDate |
2022 |
dc.date.none.fl_str_mv |
2022-01-01 2023-07-29T12:32:40Z 2023-07-29T12:32:40Z |
dc.type.status.fl_str_mv |
info:eu-repo/semantics/publishedVersion |
dc.type.driver.fl_str_mv |
info:eu-repo/semantics/conferenceObject |
format |
conferenceObject |
status_str |
publishedVersion |
dc.identifier.uri.fl_str_mv |
http://dx.doi.org/10.1109/IGARSS46834.2022.9883973 International Geoscience and Remote Sensing Symposium (IGARSS), v. 2022-July, p. 7301-7304. http://hdl.handle.net/11449/246138 10.1109/IGARSS46834.2022.9883973 2-s2.0-85140371370 |
url |
http://dx.doi.org/10.1109/IGARSS46834.2022.9883973 http://hdl.handle.net/11449/246138 |
identifier_str_mv |
International Geoscience and Remote Sensing Symposium (IGARSS), v. 2022-July, p. 7301-7304. 10.1109/IGARSS46834.2022.9883973 2-s2.0-85140371370 |
dc.language.iso.fl_str_mv |
eng |
language |
eng |
dc.relation.none.fl_str_mv |
International Geoscience and Remote Sensing Symposium (IGARSS) |
dc.rights.driver.fl_str_mv |
info:eu-repo/semantics/openAccess |
eu_rights_str_mv |
openAccess |
dc.format.none.fl_str_mv |
7301-7304 |
dc.source.none.fl_str_mv |
Scopus reponame:Repositório Institucional da UNESP instname:Universidade Estadual Paulista (UNESP) instacron:UNESP |
instname_str |
Universidade Estadual Paulista (UNESP) |
instacron_str |
UNESP |
institution |
UNESP |
reponame_str |
Repositório Institucional da UNESP |
collection |
Repositório Institucional da UNESP |
repository.name.fl_str_mv |
Repositório Institucional da UNESP - Universidade Estadual Paulista (UNESP) |
repository.mail.fl_str_mv |
|
_version_ |
1799965665674657792 |