A carregar...
Publicação
Modelação preditiva da vegetação natural potencial do concelho de Loures
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 15.42 MB | Adobe PDF |
Orientador(es)
Resumo(s)
O conceito de Vegetação Natural Potencial (VNP) e a sua representação
cartográfica assume uma importância primordial para a maioria dos países europeus nas
questões relacionadas com o restauro de habitats. Dada a relação existente entre as séries de
vegetação de um território e os factores ambientais, este artigo visa o desenvolvimento de
modelação predictiva da VNP para o concelho de Loures (AML). Atendendo à possibilidade
de integração de um vasto conhecimento empírico, utilizou-se uma abordagem de modelação
precedida de classificação de séries de vegetação (classification-then-modelling). Para
averiguar a relação entre 6 séries de vegetação e um conjunto de 8 variáveis ambientais
recorreu-se a Modelos de Distribuição de Espécies (SDM) aplicados ao nível da comunidade,
suportados em Sistemas de Informação Geográfica (SIG) e em modelos de regressão,
machine learning e rule-based. Os resultados obtidos permitiram aferir o modo como os
gradientes ecológicos determinam a ocorrência das séries de vegetação. A cartografia predictiva
da VNP resultante do modelo da Máxima Entropia, foi ainda validada com a cartografia
oficial da VNP do concelho de Loures (precisão global de 88%). Dado que a gestão e
conservação da biodiversidade é frequentemente desenvolvida a escalas de grande detalhe,
o SDM possibilita a integração de observações directas de comunidades vegetais, e uma
interpretação da distribuição local da VNP ao longo de gradientes ambientais.
The concept of Potential Natural Vegetation (PNV) and its mapping have become extremely important within the scope of habitat restoration in almost every European country. The aim of this study is to predict the PNV in Loures based on the vegetation series and the main environmental variables. The modelling approach is based on the distribution of communities referred to as classification-thenmodelling. Subsequently, several statistical model-fitting techniques, such as regression models, machine learning and rule-based, were successfully applied to the survey data (vegetation series; and 8 environmental/predictor variables). The spatial database was organized as a Geographic Information System (GIS) and was also used to perform the Species Distribution Models (SDM) at community level. The results show a high correspondence between the vegetation series and the environmental gradients. The predicted PNV maps based on the Maximum Entropy Model were validated with the official map of the PNV of Loures, and presented an overall accuracy of 88%. Often, conservation planning and biodiversity resource management is carried out at more detailed scales, where SDM allows integration of community direct observations and improve our interpretation of PNV local distributions along environmental gradients.
Le concept de Végétation Naturelle Potentielle (VNP) et sa représentation cartographique ont une importance primordiale dans la plupart des États européens, quand il s’agit de restaurer des habitats. En prenant en compte le rapport existant entre les séries de végétation et les facteurs ambientaux du territoire, on a cherché une modélisation prédictive de la VNP du concelho de Loures (AML). Tout en intégrant une importante connaissance empirique, on a introduit dans ce modèle une classification des séries de végétation (classification-then-modelling). Pour établir la relation existant entre 6 séries de végétation et 8 variables ambientales, on a eu recours à des Modèles de Distribution des Espèces (SDM), applicables au niveau des communautés en Systèmes d’Information Géographique (SIG) et en modèle de régression, machine learning et rule-based. Les résultats montrent comment les gradients écologiques déterminent les séries de végétation. La cartographie prédictive de la VNP, selon le modèle d’Enthropie Maximale, a ainsi été validée à 88 % par rapport à la cartographie officielle de la VNP du concelho de Loures. Étant donné que la gestion et conservation de la biodiversité sont souvent pratiquées à des échelles de grand détail, le SDM permet d’intégrer les observations directes de communautés végétales et l’interprétation de la répartition locale de la VNP au long de gradients ambientaux.
The concept of Potential Natural Vegetation (PNV) and its mapping have become extremely important within the scope of habitat restoration in almost every European country. The aim of this study is to predict the PNV in Loures based on the vegetation series and the main environmental variables. The modelling approach is based on the distribution of communities referred to as classification-thenmodelling. Subsequently, several statistical model-fitting techniques, such as regression models, machine learning and rule-based, were successfully applied to the survey data (vegetation series; and 8 environmental/predictor variables). The spatial database was organized as a Geographic Information System (GIS) and was also used to perform the Species Distribution Models (SDM) at community level. The results show a high correspondence between the vegetation series and the environmental gradients. The predicted PNV maps based on the Maximum Entropy Model were validated with the official map of the PNV of Loures, and presented an overall accuracy of 88%. Often, conservation planning and biodiversity resource management is carried out at more detailed scales, where SDM allows integration of community direct observations and improve our interpretation of PNV local distributions along environmental gradients.
Le concept de Végétation Naturelle Potentielle (VNP) et sa représentation cartographique ont une importance primordiale dans la plupart des États européens, quand il s’agit de restaurer des habitats. En prenant en compte le rapport existant entre les séries de végétation et les facteurs ambientaux du territoire, on a cherché une modélisation prédictive de la VNP du concelho de Loures (AML). Tout en intégrant une importante connaissance empirique, on a introduit dans ce modèle une classification des séries de végétation (classification-then-modelling). Pour établir la relation existant entre 6 séries de végétation et 8 variables ambientales, on a eu recours à des Modèles de Distribution des Espèces (SDM), applicables au niveau des communautés en Systèmes d’Information Géographique (SIG) et en modèle de régression, machine learning et rule-based. Les résultats montrent comment les gradients écologiques déterminent les séries de végétation. La cartographie prédictive de la VNP, selon le modèle d’Enthropie Maximale, a ainsi été validée à 88 % par rapport à la cartographie officielle de la VNP du concelho de Loures. Étant donné que la gestion et conservation de la biodiversité sont souvent pratiquées à des échelles de grand détail, le SDM permet d’intégrer les observations directes de communautés végétales et l’interprétation de la répartition locale de la VNP au long de gradients ambientaux.
Descrição
Palavras-chave
Vegetação Natural Potencial Séries de vegetação Modelação espacial Ordenamento do território Loures (Portugal)
Contexto Educativo
Citação
Gutierres, F., Gabriel, L., Emídio, A., Mendes, P., Neto, C., & Reis, E. (2015). Modelling the potential natural vegetation in the Loures municipality (Lisbon Metropolitan Area). Finisterra, 50(99), 31-62. https://doi.org/10.18055/finis3146.
Editora
Universidade de Lisboa, Centro de Estudos Geográficos