Uso de imagens Sentinel para estimativa do estoque de carbono e biomassa acima do solo no bioma Caatinga

• Ano de defesa: 2021
• Autor(a) principal: Lima, Maria Maiany Paiva
• Orientador(a): Costa, Carlos Alexandre Gomes
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Palavras-chave em Português: Produtividade primária líquida; Floresta tropical sazonalmente seca; Sentinel-1 e 2; Sensoriamento remoto; Fitossociologia; Primary net productivity; Seasonally dry tropical forest; Sentinel 1 and 2; Remote sensing; Phytosociology
• Link de acesso: http://www.repositorio.ufc.br/handle/riufc/58011

Biomass is a vital natural resource for the functioning of the biosphere, its quantification and monitoring are necessary for better planning and management of use. The methods based on remote sensing (SR) data to quantify it provide a synoptic view and they are applicable into large and inaccessible areas, of which performance depends on the characteristics of the sensor and target. Therefore, the objective was to evaluate the carbon stock and aboveground biomass in the strata of the Caatinga, a Seasonally Dry Tropical Forest (FTSS), using SR data and data obtained in the field. The methodology was divided into three stages: i) determination of biomass and carbon stock in situ; (ii) analysis based on remote sensing data from the MSI/Sentinel-2 sensor and the SAR-C/Sentinel-1; (iii) and analysis of the relationship between field and SR data. The relationship was accessed by means of multiple linear regression generating empirical models for obtaining biomass and carbon stock, and based on these models, the mapping of these parameters was carried out. The biomass showed a strong relationship with the diameter of the tree stems and as for the SR data, the best predictors were the red-edge bands and their derived indexes. As for the models for estimating woody biomass, the one based on images of the rainy season showed a better performance than the one using images of the dry season, whose adjusted coefficients were 0.78 and 0.53 respectively. But both resulted in an adequate spatialization of the biomass of according to the use of the soil and the different physiognomies of vegetation found in the area. As for herbaceous biomass, the best model presented an adjusted determination coefficient equal to 0.54 with homogeneous spatialization and generic identification of preserved and anthropized areas. Therefore, the generated models were able to predict the herbaceous and woody biomass for any season in the FTSS area.