Modelos preditivos de biomassa em Floresta Amazônica a partir de dados LiDAR
Autor(a) principal: | |
---|---|
Data de Publicação: | 2019 |
Tipo de documento: | Dissertação |
Idioma: | por |
Título da fonte: | Manancial - Repositório Digital da UFSM |
Texto Completo: | http://repositorio.ufsm.br/handle/1/16754 |
Resumo: | LiDAR (Light Detection and Ranging) remote sensing data combined with machine learning techniques has presented great potential for modeling large-scale forest atributes. In this context, the work aims to evaluate the application of machine learning techniques in the construction of models that relate LiDAR metrics and forest inventory data in the prediction of biomass in tropical forest. Initially, above-ground biomass was computed by an adjusted allometric equation, using biometric variables inventoried in 85 sample units at Fazenda Cauaxi, in the municipality of Paragominas / PA. The biomass of the plots (variable of interest) was related to 87 LiDAR metrics (explanatory variables), obtained by processing the LiDAR points clouds. This database was randomly divided into 70% for model adjustment and 30% for validation. The predictive performance of three different machine learning techniques (Random Forest - RF, Support Vector Machine - SVM and Artificial Neural Networks - ANN) was compared to a Generalized Linear Model (GLM) technique, traditionally used in nonparametric estimations. The results indicated that the information derived from the airborne LiDAR survey proved to be efficient and perfectly applicable to the modeling process of biomass in a tropical environment. With the exception of the RF model, with R² of 0.60, the machine learning models obtained better performance in the training stage. The value of 0.99 for the R² and the superior performance in the other adjustment quality indicators (RMSE, Syx, BIAS and DM), gave the ANN model the condition of better adaptation to the training data. In the validation stage, the GLM and RF models that presented the worst indicators in relation to the adjustment, showed superior performance, while the ANN estimates showed the greatest distortion. In general, the Spearman correlation between the estimated and observed values presented a behavior inversely proportional to the degree of adjustment of the models in the training stage, varying from 0.57 to 0.87 for the ANN and GLM models respectively. In spite of the lower adjustment of the RF model and the lower generalization capacity of the ANN and SVM models, the Wilcoxon Rank Sum Test did not detect a significant difference between the biomass values observed and predicted by the different models. In this way, it was possible to observe that the machine learning algorithms were able to detect and reproduce well the nonparametric data structure and to cope with generalized regression, without the need for data dimensionality reduction techniques, which gave more agility to the modeling process. |
id |
UFSM_308500ad27ad19ed6d5bcdf5523116d3 |
---|---|
oai_identifier_str |
oai:repositorio.ufsm.br:1/16754 |
network_acronym_str |
UFSM |
network_name_str |
Manancial - Repositório Digital da UFSM |
repository_id_str |
|
spelling |
Modelos preditivos de biomassa em Floresta Amazônica a partir de dados LiDARPredictive biomass models in Amazon Rainforest by LiDAR dataMensuração florestalInteligência artificialAmazôniaSensoriamento remotoForest measurementArtificial intelligenceAmazonRemote sensingCNPQ::CIENCIAS AGRARIAS::RECURSOS FLORESTAIS E ENGENHARIA FLORESTALLiDAR (Light Detection and Ranging) remote sensing data combined with machine learning techniques has presented great potential for modeling large-scale forest atributes. In this context, the work aims to evaluate the application of machine learning techniques in the construction of models that relate LiDAR metrics and forest inventory data in the prediction of biomass in tropical forest. Initially, above-ground biomass was computed by an adjusted allometric equation, using biometric variables inventoried in 85 sample units at Fazenda Cauaxi, in the municipality of Paragominas / PA. The biomass of the plots (variable of interest) was related to 87 LiDAR metrics (explanatory variables), obtained by processing the LiDAR points clouds. This database was randomly divided into 70% for model adjustment and 30% for validation. The predictive performance of three different machine learning techniques (Random Forest - RF, Support Vector Machine - SVM and Artificial Neural Networks - ANN) was compared to a Generalized Linear Model (GLM) technique, traditionally used in nonparametric estimations. The results indicated that the information derived from the airborne LiDAR survey proved to be efficient and perfectly applicable to the modeling process of biomass in a tropical environment. With the exception of the RF model, with R² of 0.60, the machine learning models obtained better performance in the training stage. The value of 0.99 for the R² and the superior performance in the other adjustment quality indicators (RMSE, Syx, BIAS and DM), gave the ANN model the condition of better adaptation to the training data. In the validation stage, the GLM and RF models that presented the worst indicators in relation to the adjustment, showed superior performance, while the ANN estimates showed the greatest distortion. In general, the Spearman correlation between the estimated and observed values presented a behavior inversely proportional to the degree of adjustment of the models in the training stage, varying from 0.57 to 0.87 for the ANN and GLM models respectively. In spite of the lower adjustment of the RF model and the lower generalization capacity of the ANN and SVM models, the Wilcoxon Rank Sum Test did not detect a significant difference between the biomass values observed and predicted by the different models. In this way, it was possible to observe that the machine learning algorithms were able to detect and reproduce well the nonparametric data structure and to cope with generalized regression, without the need for data dimensionality reduction techniques, which gave more agility to the modeling process.Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPESDados de sensores remotos LiDAR (Light Detection and Ranging), combinados com técnicas de aprendizado de máquina tem apresentado grande potencial para a modelagem de atributos florestais em larga escala. Nesse contexto, o trabalho tem como objetivo avaliar a aplicação de técnicas de aprendizado de máquina na construção de modelos que relacionam métricas LiDAR e dados de inventário florestal na predição de biomassa em floresta tropical. Inicialmente, foi computada a biomassa acima do solo por meio de uma equação alométrica ajustada, fazendo uso de variáveis biométricas inventariadas em 85 unidades amostrais na Fazenda Cauaxi, município de Paragominas/PA. A biomassa das parcelas (variável de interesse) foi relacionada com 87 métricas LiDAR (variáveis explicativas), obtidas via processamento das nuvens de pontos LiDAR. Essa base de dados foi dividida aleatoriamente em 70% para ajuste dos modelos e 30% destinados à validação. Comparou-se o desempenho preditivo de três diferentes técnicas de aprendizado de máquina (Random Forest - RF, Support Vector Machine - SVM e Artificial Neural Network - ANN) frente a técnica Generalized Linear Model - GLM, tradicionalmente empregada em estimativas não paramétricas. Os resultados indicaram que as informações derivadas do levantamento LiDAR aerotransportado mostraram-se eficientes e perfeitamente aplicáveis ao processo de modelagem da biomassa em ambiente tropical. À exceção do modelo RF, com R² de 0,60, os modelos de aprendizado de máquina obtiveram melhor desempenho na etapa de treinamento. O valor de 0,99 para o R² e o desempenho superior nos demais indicadores da qualidade de ajuste (RMSE, Syx, BIAS e DM), conferiram ao modelo ANN a condição de melhor adequação aos dados de treino. Já na etapa de validação, os modelos GLM e RF que haviam apresentado os piores indicadores em relação ao ajuste, mostraram desempenho superior, enquanto que as estimativas ANN apresentaram a maior distorção. De modo geral, a correlação de Spearman entre os valores estimados e observados apresentou comportamento inversamente proporcional ao grau de ajuste dos modelos na etapa de treinamento, variando de 0,57 a 0,87 para os modelos ANN e GLM respectivamente. A despeito do ajuste inferior do modelo RF e da menor capacidade de generalização dos modelos ANN e SVM, a estatística Wilcoxon Rank Sum Test não detectou diferença significativa entre os valores de biomassa observados e preditos pelos diferentes modelos. Dessa forma, foi possível observar que os algoritmos de aprendizado de máquina conseguiram detectar e reproduzir bem a estrutura não paramétrica dos dados e fazer frente a regressão generalizada, sem a necessidade da aplicação de técnicas de redução da dimensionalidade dos dados, o que conferiu mais agilidade ao processo de modelagem.Universidade Federal de Santa MariaBrasilRecursos Florestais e Engenharia FlorestalUFSMPrograma de Pós-Graduação em Engenharia FlorestalCentro de Ciências RuraisPereira, Rudiney Soareshttp://lattes.cnpq.br/9479801378014588Hendges, Elvis Rabuskehttp://lattes.cnpq.br/5292160200165795Silva, Emanuel Araújohttp://lattes.cnpq.br/2765651276275384Schuh, Mateus Sabadi2019-06-05T14:40:29Z2019-06-05T14:40:29Z2019-01-30info:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/masterThesisapplication/pdfhttp://repositorio.ufsm.br/handle/1/16754porAttribution-NonCommercial-NoDerivatives 4.0 Internationalhttp://creativecommons.org/licenses/by-nc-nd/4.0/info:eu-repo/semantics/openAccessreponame:Manancial - Repositório Digital da UFSMinstname:Universidade Federal de Santa Maria (UFSM)instacron:UFSM2022-05-18T13:57:40Zoai:repositorio.ufsm.br:1/16754Biblioteca Digital de Teses e Dissertaçõeshttps://repositorio.ufsm.br/ONGhttps://repositorio.ufsm.br/oai/requestatendimento.sib@ufsm.br||tedebc@gmail.comopendoar:2022-05-18T13:57:40Manancial - Repositório Digital da UFSM - Universidade Federal de Santa Maria (UFSM)false |
dc.title.none.fl_str_mv |
Modelos preditivos de biomassa em Floresta Amazônica a partir de dados LiDAR Predictive biomass models in Amazon Rainforest by LiDAR data |
title |
Modelos preditivos de biomassa em Floresta Amazônica a partir de dados LiDAR |
spellingShingle |
Modelos preditivos de biomassa em Floresta Amazônica a partir de dados LiDAR Schuh, Mateus Sabadi Mensuração florestal Inteligência artificial Amazônia Sensoriamento remoto Forest measurement Artificial intelligence Amazon Remote sensing CNPQ::CIENCIAS AGRARIAS::RECURSOS FLORESTAIS E ENGENHARIA FLORESTAL |
title_short |
Modelos preditivos de biomassa em Floresta Amazônica a partir de dados LiDAR |
title_full |
Modelos preditivos de biomassa em Floresta Amazônica a partir de dados LiDAR |
title_fullStr |
Modelos preditivos de biomassa em Floresta Amazônica a partir de dados LiDAR |
title_full_unstemmed |
Modelos preditivos de biomassa em Floresta Amazônica a partir de dados LiDAR |
title_sort |
Modelos preditivos de biomassa em Floresta Amazônica a partir de dados LiDAR |
author |
Schuh, Mateus Sabadi |
author_facet |
Schuh, Mateus Sabadi |
author_role |
author |
dc.contributor.none.fl_str_mv |
Pereira, Rudiney Soares http://lattes.cnpq.br/9479801378014588 Hendges, Elvis Rabuske http://lattes.cnpq.br/5292160200165795 Silva, Emanuel Araújo http://lattes.cnpq.br/2765651276275384 |
dc.contributor.author.fl_str_mv |
Schuh, Mateus Sabadi |
dc.subject.por.fl_str_mv |
Mensuração florestal Inteligência artificial Amazônia Sensoriamento remoto Forest measurement Artificial intelligence Amazon Remote sensing CNPQ::CIENCIAS AGRARIAS::RECURSOS FLORESTAIS E ENGENHARIA FLORESTAL |
topic |
Mensuração florestal Inteligência artificial Amazônia Sensoriamento remoto Forest measurement Artificial intelligence Amazon Remote sensing CNPQ::CIENCIAS AGRARIAS::RECURSOS FLORESTAIS E ENGENHARIA FLORESTAL |
description |
LiDAR (Light Detection and Ranging) remote sensing data combined with machine learning techniques has presented great potential for modeling large-scale forest atributes. In this context, the work aims to evaluate the application of machine learning techniques in the construction of models that relate LiDAR metrics and forest inventory data in the prediction of biomass in tropical forest. Initially, above-ground biomass was computed by an adjusted allometric equation, using biometric variables inventoried in 85 sample units at Fazenda Cauaxi, in the municipality of Paragominas / PA. The biomass of the plots (variable of interest) was related to 87 LiDAR metrics (explanatory variables), obtained by processing the LiDAR points clouds. This database was randomly divided into 70% for model adjustment and 30% for validation. The predictive performance of three different machine learning techniques (Random Forest - RF, Support Vector Machine - SVM and Artificial Neural Networks - ANN) was compared to a Generalized Linear Model (GLM) technique, traditionally used in nonparametric estimations. The results indicated that the information derived from the airborne LiDAR survey proved to be efficient and perfectly applicable to the modeling process of biomass in a tropical environment. With the exception of the RF model, with R² of 0.60, the machine learning models obtained better performance in the training stage. The value of 0.99 for the R² and the superior performance in the other adjustment quality indicators (RMSE, Syx, BIAS and DM), gave the ANN model the condition of better adaptation to the training data. In the validation stage, the GLM and RF models that presented the worst indicators in relation to the adjustment, showed superior performance, while the ANN estimates showed the greatest distortion. In general, the Spearman correlation between the estimated and observed values presented a behavior inversely proportional to the degree of adjustment of the models in the training stage, varying from 0.57 to 0.87 for the ANN and GLM models respectively. In spite of the lower adjustment of the RF model and the lower generalization capacity of the ANN and SVM models, the Wilcoxon Rank Sum Test did not detect a significant difference between the biomass values observed and predicted by the different models. In this way, it was possible to observe that the machine learning algorithms were able to detect and reproduce well the nonparametric data structure and to cope with generalized regression, without the need for data dimensionality reduction techniques, which gave more agility to the modeling process. |
publishDate |
2019 |
dc.date.none.fl_str_mv |
2019-06-05T14:40:29Z 2019-06-05T14:40:29Z 2019-01-30 |
dc.type.status.fl_str_mv |
info:eu-repo/semantics/publishedVersion |
dc.type.driver.fl_str_mv |
info:eu-repo/semantics/masterThesis |
format |
masterThesis |
status_str |
publishedVersion |
dc.identifier.uri.fl_str_mv |
http://repositorio.ufsm.br/handle/1/16754 |
url |
http://repositorio.ufsm.br/handle/1/16754 |
dc.language.iso.fl_str_mv |
por |
language |
por |
dc.rights.driver.fl_str_mv |
Attribution-NonCommercial-NoDerivatives 4.0 International http://creativecommons.org/licenses/by-nc-nd/4.0/ info:eu-repo/semantics/openAccess |
rights_invalid_str_mv |
Attribution-NonCommercial-NoDerivatives 4.0 International http://creativecommons.org/licenses/by-nc-nd/4.0/ |
eu_rights_str_mv |
openAccess |
dc.format.none.fl_str_mv |
application/pdf |
dc.publisher.none.fl_str_mv |
Universidade Federal de Santa Maria Brasil Recursos Florestais e Engenharia Florestal UFSM Programa de Pós-Graduação em Engenharia Florestal Centro de Ciências Rurais |
publisher.none.fl_str_mv |
Universidade Federal de Santa Maria Brasil Recursos Florestais e Engenharia Florestal UFSM Programa de Pós-Graduação em Engenharia Florestal Centro de Ciências Rurais |
dc.source.none.fl_str_mv |
reponame:Manancial - Repositório Digital da UFSM instname:Universidade Federal de Santa Maria (UFSM) instacron:UFSM |
instname_str |
Universidade Federal de Santa Maria (UFSM) |
instacron_str |
UFSM |
institution |
UFSM |
reponame_str |
Manancial - Repositório Digital da UFSM |
collection |
Manancial - Repositório Digital da UFSM |
repository.name.fl_str_mv |
Manancial - Repositório Digital da UFSM - Universidade Federal de Santa Maria (UFSM) |
repository.mail.fl_str_mv |
atendimento.sib@ufsm.br||tedebc@gmail.com |
_version_ |
1805922070507290624 |