Loading...
Publication
Spatio-temporal variability in dune plant communities using UAV and multispectral data
| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 1.75 MB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
O mapeamento da vegetação, através da identificação do tipo e distribuição das comunidades e espécies vegetais, é crucial para analisar a cobertura vegetal e os padrões espaciais. A compreensão das variabilidades espaciais e temporais das plantas dunares em ligação com a morfodinâmica permite uma maior compreensão do dinamismo e evolução dos ambientes costeiros. Tal análise pode contribuir para o desenvolvimento de planos de gestão costeira que ajudam a implementar a biodiversidade costeira e estratégias de protecção. Esta dissertação apresenta uma abordagem para avaliar a utilização de imagens multiespectrais e explorar a variabilidade da vegetação dunar costeira com dados recolhidos à distância por um Veículo Aéreo Não Tripulado (UAV). Foram escolhidas quatro zonas de estudo diferentes na parte oriental da Península de Ancao, distribuídas alongshore, e cobrindo a backhore e a crista das dunas até à base do lee das dunas. Foram utilizados dados de campo e de UAV, em diferentes épocas, nomeadamente ao longo de um período de dois anos. Foi utilizada uma abordagem de classificação em duas etapas, baseada num índice de vegetação de diferença normalizada e num classificador de Floresta Aleatória. Os resultados mostram desempenhos de classificação de alta precisão ao condensar a cobertura do solo em menos classes e também em áreas menos densamente vegetativas. As classificações resultantes foram posteriormente processadas em termos de alterações transfronteiriças e alterações sazonais. Estas técnicas mostram um elevado potencial futuro para avaliar a vegetação das áreas de dunas costeiras e para apoiar a gestão costeira.
The mapping of vegetation, by identifying the type and distribution of plant communities and species, is crucial for analysing vegetation coverage and spatial patterns. Understanding dune plant spatial and temporal variabilities in connection with morphodynamics gives further insight in dynamism and evolution of coastal environments. Such analysis can contribute to the development of coastal management plans that helps to implement coastal biodiversity and protection strategies. This dissertation presents an approach to assess the use of multispectral imagery and explore the variability of coastal dune vegetation with remotely sensed data collected by an Unmanned Aerial Vehicle (UAV). Four different study zones were chosen at the eastern part of the Ancao Peninsula, distributed alongshore, and covering the backshore and the dune crest until the base of the dune lee. Field and UAV data were used, in different seasons namely over an extend of two years. A two-step classification approach, based on a normalized difference vegetation index and Random Forest classifier, was used. The Results show high accuracy classification performances when condensing the groundcover into fewer classes and also in less densely vegetated areas. Resulting classifications were further processed in terms of cross-shore changes and seasonal changes. These technics show a high future potential to assess the vegetation of coastal dune areas and to support coastal management.
The mapping of vegetation, by identifying the type and distribution of plant communities and species, is crucial for analysing vegetation coverage and spatial patterns. Understanding dune plant spatial and temporal variabilities in connection with morphodynamics gives further insight in dynamism and evolution of coastal environments. Such analysis can contribute to the development of coastal management plans that helps to implement coastal biodiversity and protection strategies. This dissertation presents an approach to assess the use of multispectral imagery and explore the variability of coastal dune vegetation with remotely sensed data collected by an Unmanned Aerial Vehicle (UAV). Four different study zones were chosen at the eastern part of the Ancao Peninsula, distributed alongshore, and covering the backshore and the dune crest until the base of the dune lee. Field and UAV data were used, in different seasons namely over an extend of two years. A two-step classification approach, based on a normalized difference vegetation index and Random Forest classifier, was used. The Results show high accuracy classification performances when condensing the groundcover into fewer classes and also in less densely vegetated areas. Resulting classifications were further processed in terms of cross-shore changes and seasonal changes. These technics show a high future potential to assess the vegetation of coastal dune areas and to support coastal management.
Description
Keywords
Alterações transfronteiriças Alterações sazonais Floresta aleatória Vegetação