La télédétection spatiale de la couleur de l’océan implique l’élimination de la contribution atmosphérique, appelée correction atmosphérique (CA). Au-dessus des eaux claires, cette dernière se base sur l’hypothèse que l’eau est totalement absorbante dans le Proche Infra-Rouge (PIR) pour estimer la réflectance atmosphérique et déterminer la réflectance marine. À l’opposé, au-dessus des eaux côtières turbides, la contribution du signal marin n’est pas négligeable dans le PIR. De ce fait, différentes méthodes alternatives ont été proposées. La thèse se consacre à l’évaluation des algorithmes de CA proposés et leur amélioration pour le capteur Ocean and Land Colour Instrument (OLCI) au-dessus des eaux côtières en exploitant sa synergie avec le capteur Sea and Land Surface Temperature Radiometer (SLSTR). Dans ce but, des mesures radiométriques in-situ ont été acquises dans deux zones côtières françaises contrastées : Manche orientale et Guyane Française, avec le spectro-radiomètre ASD, suivant un nouveau protocole d’acquisition et de post-traitement. Le post-traitement s’est basé sur le coefficient de variabilité et la différence relative de la médiane dans le contrôle de qualité, en plus du score d’assurance de qualité (QAS). Suivant l’analyse statistique basée en partie sur l’angle spectral moyenné (SAM), l’inter-comparaison radiométrique de l’ASD avec les autres radiomètres (TriOS-above et Radeau), a révélé la cohérence des mesures ASD. L’utilisation de ces dernières a permis l’inter-comparaison de la performance de cinq algorithmes de CA, où l’algorithme Polymer est le plus performant d’après un système à points avec une seule métrique. Cependant aucun algorithme a atteint le maximum de points soulignant la grande marge de progression à accomplir, surtout en eau côtière. Dans cette optique, trois relations spectrales de la réflectance des aérosols ont été testées sur une base de données simulées suivant la synergie OLCI/SLSTR. Une autre relation, Full Spectrum AC (FSAC) a été développée initialement en combinant deux relations existantes, après l’élimination de l’hypothèse du pixel noir dans l’ultra-violet (UV) et l’intégration d’un schéma itératif. L’inter-comparaison des relations révèle la cohérence de FSAC qui est légèrement moins performante qu’une relation publiée. L’application de FSAC sur des images OLCI/SLSTR ouvrirait des perspectives dans l’amélioration de la CA au-dessus des eaux côtières. / The ocean color remote sensing involves the removal of the atmospheric contribution, the so-called atmospheric correction (AC). Over clear waters, the latter is based on the hypothesis that the sea water is totally absorbent in the Near Infra-Red (NIR), to estimate the atmospheric reflectance and to determine the water reflectance. By contrast, over coastal turbid waters, the marine signal is not negligible in the NIR. Accordingly, different alternative methods were proposed. The thesis is committed to evaluate the proposed AC algorithms and their improvement for the Ocean and Land Colour Instrument (OLCI) sensor over coastal waters exploiting its synergy with the Sea and Land Surface Temperature Radiometer (SLSTR) sensor. For this purpose, radiometric in-situ measurements were acquired in two contrasted French coastal areas : Eastern English Channel and FrenchGuiana, with the ASD spectro-radiometer, according to a newly developed measurement and post-processing protocol. The post-processing was based on the coefficient of variability and the median relative difference, in addition to the Quality Assurance Score (QAS). Following the statistical analysis in part based on the Spectral Angle Mean (SAM), the radiometric inter-comparison of the ASD and other radiometers (TriOS-above and TriOS in-water), shows the consistency of the ASD measurements. The use of these measurement leads to carry out the performance inter-comparison of five AC algorithms, where the Polymer algorithm is the most efficient according to a unique metric scoring system. However, neither algorithm obtained the maximum score, highlighting the big room for improvement, especially for coastal waters. With this in mind; three spectral relationships of aerosols reflectance were tested with a simulated data set based on OLCI/SLSTR synergy. Another relationship, Full Spectrum AC (FSAC) was initially developed combining two existing relationships, after excluding the black pixel hypothesis in the Ultra-Violet and integrating a iterative scheme. The relationships inter-comparison shows consistency ofFSAC which is slightly less performing than one published relationship. The application of FSAC on OLCI/SLSTR images could have perspectives to improve the AC over coastal waters.

Submitted by Caroline Piffer de Andrade (carol_piffer@hotmail.com) on 2018-04-28T18:02:37Z No. of bitstreams: 1 CarolinePifferdeAndrade.pdf: 3287357 bytes, checksum: beca78c3926ad0af4d896e61b2ecaf99 (MD5) / Approved for entry into archive by ALESSANDRA KUBA OSHIRO ASSUNÇÃO (alessandra@fct.unesp.br) on 2018-05-02T13:23:34Z (GMT) No. of bitstreams: 1 andrade_cp_me_prud.pdf: 3003528 bytes, checksum: 19259069721e1c717d15aa96a145d1fb (MD5) / Made available in DSpace on 2018-05-02T13:23:35Z (GMT). No. of bitstreams: 1 andrade_cp_me_prud.pdf: 3003528 bytes, checksum: 19259069721e1c717d15aa96a145d1fb (MD5) Previous issue date: 2018-03-23 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) / Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP) / O presente estudo objetivou estimar as concentrações de clorofila-a (Chl-a) no reservatório da usina hidroelétrica de Ibitinga (RHI), localizado no Rio Tietê, estado de São Paulo, Brasil, por meio de coeficientes de absorção obtidos via algoritmos quase-analíticos (QAAs). Para isso, realizou-se uma caracterização bio-ótica e biogeoquímica do RHI, por meio de dados espectrais e de qualidade da água coletados em dois trabalhos de campo, conduzidos em Julho de 2016 e Junho de 2017. Os desempenhos das versões originais QAAV5 e QAAV6 em estimar as propriedades óticas inerentes (POIs) no RHI foram avaliados. Versões re-parametrizadas para dois reservatórios localizados no sistema em cascata do Rio Tietê, QAABBHR e QAAOMW, foram também testadas para a área de estudo. Além disso, foram avaliadas as performances de esquemas compostos pelas versões do QAA já mencionadas, seguidas por quatro modelos para estimativa de Chl-a, os quais utilizam coeficientes de absorção como dados de entrada. A distribuição espacial das concentrações de Chl-a foi analisada por meio da aplicação desses esquemas em uma imagem do sensor Ocean and Land Colour Instrument (OLCI) instalado a bordo do satélite Sentinel-3A, com aquisição coincidente com o segundo trabalho de campo realizado na área de estudo. A caracterização bio-ótica demonstrou variabilidade espacial e temporal dos constituintes oticamente significativos (COSs) no RHI, com predominância da absorção pelo material orgânico colorido dissolvido (CDOM). As versões do QAA testadas para o primeiro conjunto de dados não se mostraram completamente adequadas na obtenção de coeficientes de absorção em todos os comprimentos de onda. Com relação aos esquemas para estimativa de concentração de Chl-a, apenas aqueles baseados no QAAV5 foram capazes de obter resultados razoáveis - Raiz do Erro Médio Quadrático Normalizado (REMQN) < 47.50 % - para os dados da imagem OLCI. Todos os quatro modelos para estimativa de Chl-a testados apresentaram resultados similares para os dados de saída do QAAV5. Esses resultados enfatizam o desafio gerado pela grande variabilidade ótica dos sistemas em cascata, com relação à modelagem bio-óptica. Os resultados obtidos dão suporte a futuros trabalhos, os quais podem resultar em aplicações como o monitoramento do estado trófico na área de estudo a partir de dados de satélite, com maior acurácia proveniente do uso de modelos que possam estimar consistentemente suas POIs. / This research was aimed at retrieving chlorophyll-a (Chl-a) concentrations in Ibitinga Hydroelectric Reservoir (IHR), located at Tietê River, São Paulo State, Brazil, using absorption coefficients obtained via Quasi-analytical algorithms (QAAs). For this purpose, a bio-optical and bio-geochemical characterization of IHR was carried out, through spectral and water quality data collected in two field campaigns conducted in July, 2016 and June, 2017. The suitability of two QAA native forms (QAAV5 and QAAV6) in retrieving inherent optical properties (IOPs) in IHR was assessed. Versions re-parameterized for two reservoirs also located in the Tietê River cascading system, QAABBHR and QAAOMW, were also tested for the study area. Besides that, the performances of schemes composed by the QAA versions already mentioned followed by four models that use absorption coefficients as inputs for estimating Chl-a concentration in Ibitinga Reservoir were evaluated. Spatial distribution of Chl-a in the reservoir was analyzed, since these schemes were applied in an image of the Ocean and Land Colour Instrument (OLCI) sensor onboard Sentinel-3A satellite, with acquisition date coincident with the second field campaign. The bio-optical characterization showed spatial and temporal variability of optically significant constituent (OSC) in IHR and colored dissolved organic matter (CDOM) predominance in its absorption budget. None of the QAA versions tested for the first dataset was completely satisfactory in retrieving absorption coefficients for IHR in all wavelengths. Regarding the schemes for Chl-a concentration estimates, only the ones based on QAAV5 were able to obtain reasonable results - Normalized Root Mean Square Error (nRMSE) < 47.50 % - for the OLCI image data. All four models for Chl-a estimation tested presented similar results for QAAV5 outputs. These results highlight the challenge of copying with high optical variability in cascading systems. The results obtained support further works, which can, prospectively, lead to many practical applications, as monitoring of trophic state in the study area from satellite data, with higher accuracy provided by the use of models that can consistently retrieve the IOPs for this specific water system. / FAPESP: 2012/19821-1 e 2015/21586-9