Resumo -O objetivo deste trabalho foi utilizar as técnicas de reflectância acumulada e mineração de dados, seguidas por classificação orientada a objeto, em imagens do sensor Operational Land Imager (OLI), satélite Landsat 8, para a classificação de vegetação nativa e cobertura agropecuária do Cerrado. Quatro imagens de reflectância foram utilizadas para a discriminação de seis classes -agricultura, pecuária, campo limpo úmido, savana, floresta e campo -, para a classificação do Parque Nacional das Emas, no Estado de Goiás, e adjacências. As imagens foram segmentadas para a extração de atributos espectrais de amostras e a aplicação de combinações de atributos (média + moda, todos os atributos) na mineração de dados. O programa Weka foi utilizado para a construção das árvores de decisão. Essa metodologia indicou que a diferenciação entre alvos aumentou a partir da acumulação temporal da reflectância, em todas as bandas e as classes, e a melhor imagem foi aquela do somatório das quatro datas. A classificação baseada na associação de atributos média + moda não apresentou impedimentos no processamento das regras de decisão, diferentemente da associação de todos os atributos. A classificação média + moda apresentou acurácia satisfatória (exatidão global, 69%; Kappa, 58%; e TAU, 63%). A integração dessas técnicas apresenta potencial para a diferenciação de vegetação nativa e antrópica do Cerrado.Termos para indexação: análise baseada em objeto, análise multitemporal, antropização, classificação supervisionada, mineração, sensoriamento remoto. Object-oriented classification in association with accumulated reflectance and data mining toolsAbstract -The objective of this work was to use the accumulated reflectance technique and data mining application, followed by object-oriented classification, in images of Operational Land Imager (OLI) sensor, Landsat 8, for the classification of native vegetation and agricultural coverage of Cerrado. Four reflectance images were used for the discrimination of six classes -agriculture, livestock, wetland, savannah, forest, and grassland -, for classification of Parque Nacional das Emas and surrounding areas in the state of Goiás, Brazil. The images were segmented for the extraction of sample spectral attributes and application of attribute combinations (mean + mode, all attributes) on data mining. The Weka software was used to construct the decision trees. This methodology indicated that the differentiation among targets increased from the temporal accumulation of the reflectance in all bands and classes, and that the optimal image was that of the sum of the four dates. The classification based on the attribute associations mean + mode showed no restraints in the decision rules processing, unlike the association of all attributes. The mean + mode classification showed a satisfactory accuracy (global accuracy, 69%; Kappa, 58%; and TAU, 63%). The integration of these techniques shows potential to differentiate native and anthropogenic vegetation in the Cerrado.

No abstract

Cyanobacterial blooms are related to eutrophic conditions that compromise the many uses of reservoirs. Thus, quick and effective methods for detecting the abundance of cyanobacteria in waterbodies are needed to complement conventional laboratory methods. In addition, inadequate control techniques that are applied at times of high cyanobacterial concentrations can cause the cells to lyse and release toxins into the water. In the present study we investigated the behaviour of cyanobacteria by determining phycocyanin and chlorophyll concentrations, using spectroradiometric and fluorometric techniques, in three field campaigns performed at the Nova Avanhandava Reservoir, Brazil. The sampling rate and favourable season for data collected had been determined previously by remote sensing analysis. Seasonal estimates of cyanobacteria were made because fluorometric sensors were able to record low concentrations, whereas the spectral analyses only detected phycocyanin at higher concentrations. Results of spectral analyses highlighted the subtle spectral characteristics indicating the presence of phycocyanin, even without a clear definition of the diagnostic features in the reflectance curve. Therefore, multiscale remote sensing complemented by fluorometric analysis and relevant environmental variables is an effective approach for monitoring cyanobacteria in Brazilian inland waters. of the collection and analytical methods, the spatial and temporal coverage is not sufficient to provide a picture of the real CSIRO PUBLISHING

R E S U M OFlorações de fitoplanctôn podem constituir em riscos à saúde humana e biota aquática, sendo necessários o monitoramento da comunidade fitoplanctônica e a adoção de mecanismos visando à prevenção de sua ocorrência. Neste contexto métodos tradicionais de monitoramento podem ser mais efetivos se complementados por abordagens que utilizam as propriedades ópticas dos pigmentos fitoplanctônicos por meio do Sensoriamento Remoto. Com o objetivo de avaliar o potencial de dados espectrais multifonte na detecção remota do fitoplâncton, foi selecionada uma área de estudo no reservatório de Nova Avanhandava, SP, caracterizada por intensa atividade agrícola no seu entorno. Para esta análise foram adquiridos dados hiperespectrais em campo e imagens multiespectrais Modis e RapidEye, os quais foram relacionados a variáveis limnológicas indicadoras do comportamento fitoplanctônico; clorofila a e ficocianina. Os resultados mostram que imagens multiespectrais permitem uma avaliação da biomassa fitoplanctônica pela clorofila a; contudo, para ficocianina, pigmento fitoplanctônico diagnóstico da presença de cianobactérias, dados mais refinados são necessários, tais como os hiperespectrais.Multisource remote sensing applied to the detection of phytoplankton in inland waters A B S T R A C TBlooms of phytoplankton can be a risk to human health and aquatic biota, so the adoption of monitoring methods of phytoplankton and mechanisms for preventing its occurrence are needed. Thus, traditional monitoring methods could be more effective if complemented by approaches using the optical properties of phytoplankton pigments by means of Remote Sensing. In order to evaluate the potential of multi-scale remote sensing for detection of the phytoplankton activity, a study area was selected in Nova Avanhandava reservoir, located in the Tietê River, SP. For this analysis, hyperspectral field data and multispectral images of low and medium spatial resolution (Modis and RapidEye) were acquired and were related to indicator limnological variables of phytoplankton behavior; chlorophyll a and phycocyanin. The results show that a specific spectral band of RapidEye system (690-730 nm) allowed detect chlorophyll a and to evaluate the phytoplankton biomass, however hyperspectral data are needed to detect the phycocyanin pigment, indicative of cyanobacteria. Palavras-chave:sensoriamento remoto da água cianobactérias clorofila a ficocianina Key words: remote sensing of water cyanobacteria chlorophyll a phycocyanin