Estimativa da concentração de sedimentos suspensos no Rio Jaguaribe-CE através de imagens do satélite Rapideye

• Ano de defesa: 2017
• Autor(a) principal: Pereira, Francisco Jairo Soares
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Palavras-chave em Português: Hidrossedimentologia; Sensoriamento remoto; Reflectância da água
• Link de acesso: http://www.repositorio.ufc.br/handle/riufc/36364

The large amount of sediment transported by the rivers causes the degradation and the reduction of the water availability with damages to the population and to the environment. The in loco monitoring of these sediments in intermittent rivers in the Brazilian semi-arid region has been an obstacle to a better understanding of the magnitude of these damages. Therefore, the overall objective of this work was to estimate suspended sediment concentrations (SSC) in the Jaguaribe-Ce River, based on the relationship between field data and remote sensing. The cross section of the Iguatu fluviometric station, located in the city of Iguatu - Ce, was used as reference in this study. The methodology used in this research is divided into two main stages: 1) obtaining and analyzing data acquired in the field; and 2) acquisition and analysis of data acquired by remote sensing. Using the CSS-flow curve, CSS was estimated in the section of the Jaguaribe River for the period from 2009 to 2014 for later analysis of the relationship that it has with the orbital data. Twelve images of the RapidEye satellite were used, whose processing consisted of two steps: i) atmospheric correction; ii) reflectance analysis and extraction. For the atmospheric correction the model 6S (Second Simulation of Satellite Signal in the Solar Spectrum) was used. For the analysis and extraction of the reflectance, 21 pixels representing the water were selected, limiting a transversal strip to the Jaguaribe River. Nine models were already tested for other regions of the world that correlate CSS with orbital image reflectance. From the data of reflectance and CSS obtained in the field, empirical models of CSS estimation were developed. For this purpose, the LABFit computational model was used. From the total of 12 images, six were used to elaborate the models and the other six to validate them. For analyzes of performance of the models, the following statistical parameters were calculated: coefficient of determination (R2); mean absolute error (EMA); root mean square error (REQM); and Nash-Sutcliffe (NSE) concordance index. The results indicated that none of the available empirical models available in the literature was adequate to represent the CSS data in the Jaguaribe River. The measured CSS data ranged from 66.54 mg / L (05/26/2014) to 229.56 mg / L (4/20/2010). Of the models developed for a single band (SEB), the one that best applies is one that relates CSS to band 4 (690 - 730 nm). This is an important relationship, since RapidEye images are the first to have an exclusive range for this wavelength. The best results with double bands (DEB) were shown when bands 5, 4 and 3 were combined with each other. The best model was the DEB4 model, which represents a relation between bands 4 and 3. The best model of triple spectral bands was the model TEB2 (Bands 3, 4 and 5). Of the six best models, five use combinations of red bands. It is concluded that the double-band model (DEB4) is the one that best represents the estimation of CSS in the Jaguaribe River. It is also noticed that the use of orbital images of high spatial resolution is efficient in the estimation of CSS in semi-arid rivers. It is also concluded that the spectral band of the images of the constellation of satellites RapidEye that best represents the CSS is the band 4 (Red-Edge).