Análise de dados multiespectrais obtidos por RPAS na caracterização espectral e mapeamento fenológico de macrohabitat de vereda no Parque Nacional da Chapada dos Guimarães - Mato Grosso/Brasil

• Autor(a) principal: Silva, Nubia da
• Data de Publicação: 2021
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional da UFMT
• Texto Completo: http://ri.ufmt.br/handle/1/4603
• topic: RPAS; Áreas úmidas; Biblioteca espectral; OBIA; Fenologia; CNPQ::CIENCIAS AGRARIAS::RECURSOS FLORESTAIS E ENGENHARIA FLORESTAL; RPAS; Wetlands; Spectral library; OBIA; Phenology

Wetlands provide important ecosystem functions that are the basis for valuable ecosystem services. Vereda, a Cerrado phytophysiognomy, is the wetland macrohabitat often associated with the presence of the buriti palm (Mauritia flexuosa), known to occur in hydromorphic soils that are gradually being affected by anthropic actions, mainly related to agricultural expansion, biological invasions and fires. Assessing footpath conditions is critical to developing conservation and restoration strategies, but it is a difficult task given the heterogeneity of habitats and field data collection is often costly and time-consuming. Remote Sensing is recognized as an effective tool to study spatial and temporal changes in the landscape. Remotely-Piloted Aircraft Systems (RPAS) provide aerial imagery with sub-decimetric resolution and offer a potential data source for habitat mapping. Therefore, this study aimed to explore the potential of multispectral images obtained by RPAS to: (1) correlate the efficiency of multispectral sensor aboard RPAS with field hyperspectral sensor, (2) analyze the spectral response of footpath habitats along a seasonality gradient and (3) feature extraction using segmentation and supervised classification techniques. Chapter 1 addresses the spectral characterization of footpath habitats through comparative analysis between multispectral and hyperspectral data and application of vegetation indices (VI) over the twelve-month period (May 2020 to April 2021). It was possible to characterize the spectral signatures of species that comprise the path, verifying that data obtained by RPAS can be an effective alternative for spectral analysis, but more distinct variations were observed for some species mainly related to plant anatomy, in addition to environmental conditions (incidence of light, particles in the air) during the flyovers may have interfered with the variations between the two sensors. The band mathematics (IV) made it possible to characterize the behavior of vegetation along the seasonal gradient, inferring the beginning and end of the senescence period of the studied habitats for the months of June and October, respectively, marked by the reduction in the incidence of rainfall in the region, showing that water restriction directly reflects on photosynthetic capacity. Chapter 2 addressed the multiscale segmentation accompanied by the Object-Based Image Analysis (OBIA) approach in the discrimination of habitats within the vereda physiognomy in the dry and rainy season (September and January). The classifications obtained reached a Kappa index of 0.92 and 0.87 for September and January, respectively. The multispectral sensor was sensitive in discriminating the habitats that make up the path and the application of the classification in two stations provided important information about the response of vegetation to water restriction in the region.