Publicações

Brazil is one of the largest producers of sugarcane and the Sao Paulo State is currently responsible for 60% of the national production. Areas previously used for livestock and citrus production have been gradually substituted by sugarcane due, mainly, to the current Brazilian government incentive for the sugarcane ethanol fuel production. The usual pre-harvest sugarcane burning in Brazil emits a significant amount of aerosols to the atmosphere. The land use/cover change and the presence of aerosols in the atmosphere have the potential to affect the local energy budget and therefore the rainfall pattern in the region. In this work we analyze the changes in precipitation patterns over the past 5 decades and their possible associations with these new land use scenario in São Paulo State, Brazil. We used precipitation time series from the network rainfall data of the Brazilian National Water Agency (Agência Nacional das Águas – ANA) and the National Oceanic and Atmospheric Administration (NOAA), a gridded data composed with multi sources information, with 1 degree of resolution. Statistical techniques (Lanczos filter) were used to filter the low frequency oscillation in the rainfall time series and the data was grouped in 10-year subsets. The annual distribution was analyzed comparing each decade in order to assess the changes in the precipitation pattern over the last 50 years in the most important area of sugarcane plantation. Preliminary results indicate a change in the annual distribution along this time period and possible association with sugarcane production expansion is under investigation.

The Brazil-Malvinas Confluence (BMC) region in the South Atlantic Ocean is characterized by the thermal gradients of the meeting between the warm Brazil Current (BC) and the cold Malvinas Current (MC). The region is considered a key region for understanding the linkage between Antarctica and South America. This paper aims to describe the atmospheric wind modulation forced by the thermal contrasts of the BMC region. Winds were available from the QuikScat satellite and NCEP/NCAR reanalysis. Sea surface temperature (SST) was derived from the AMSR sensor onboard the Aqua satellite. Analysis of the satellite data demonstrated that wind intensity fields were directly related to the SST fields attesting the coupling between the ocean and the atmosphere at the BMC region. The satellite data confirms former evidence (Pezzi et al., 2005, 2009; Acevedo et al., 2010) that wind intensity increases over the warm waters of BC and decreases over the cold waters of the MC. The reanalysis data was not as able as the satellite data to represent the atmospheric modulation in terms of wind intensity. This work suggests that the use of satellite data is important when studying ocean-atmosphere coupling over regions of intense oceanic mesoscale activity.

This paper presents a compilation of the main results obtained by this research group in twenty years of field research on forest clearing fires in the Brazilian Amazon region. Experiments were conducted in Manaus (state of Amazonas) (2 experiments), Tomé-Açu (state of Pará) (1 experiment), and Alta Floresta (state of Mato Grosso) (8 experiments). The research congregated 6 Brazilian and 2 American research institutions. Results on the following main themes of the research are presented: – Biomass consumption and emission rates of CO2 under fire; – Emission factors of combustion gases produced in the burns; – Log smoldering; – Under-story fire propagation; – Charcoal formation; – Changes of composition of the water percolated through the soil; – Mercury emissions by fires. The paper also discusses the newly approved thematic project that will support research for the following 4 years. Experiments will be conducted in Cruzeiro do Sul and Rio Branco (state of Acre) and again in Alta Floresta.

Modelos de interação biosfera-atmosfera de última geração, como o modelo integrado de biosfera-IBIS (Integrated Biosphere Simulator), usam a classificação de tipos funcionais de plantas (TFP) para determinar as classes de vegetação que pela sua vez interagem com a camada limite. O IBIS utiliza 12 TFPs (que vão de árvores tropicais a gramíneas C3) definidos através de parâmetros bioclimáticos e ecofisiológicos. Esses TFPs são reagrupados em 4 estratos fitofisionômicos: campestre (estrato 1) campo-cerrado e cerrado (estrato 2), cerradão e florestas decíduas (estrato 3) e florestas perenifólias (estrato 4); onde para cada estrato é definido um valor específico de LAI. No caso da classe savana, o modelo atribui maior peso de importância e LAI para os estratos 1, 3, 4 e 2, respectivamente. Simulações do modelo IBIS em escala global, em geral, apresentam bom desempenho no hemisfério norte e médio no hemisfério sul, especialmente na America do Sul (AS). Uma das causas deste desempenho médio na AS é devido ao fato que a vegetação de savana usado pelo modelo tem as características das savanas africanas e não as do cerrado brasileiro. Neste sentido, o objetivo do projeto é identificar e caracterizar os principais TFPs usados pelo modelo IBIS para determinar a classe savana (bioma cerrado). A metodologia consistiu na aplicação de técnicas análise multitemporal de dados índice de vegetação NDVI do sensor MODIS/TERRA no período 2000-2009 para determinar os mesmos estratos fitofisionomicos usados pelo modelo. Também foram usadas métricas de fenologia para caracterizar esses estratos. Os resultados mostram uma distribuição espacial dos estratos de Noroeste para o Sudeste que seguem um gradiente de maior a menor cobertura vegetal ao longo do bioma. O estrato dominante no bioma é o 2, com 52% de cobertura, seguido do 1 (24%), 3 (15%) e 4 (9%). A análise da sazonalidade da vegetação mostra que o estrato 1 apresenta sazonalidade marcante, seguido do 2, 3 e finalmente o 4 com sazonalidade menos definida pelo fato de representarem árvores perenes que obtém água das camadas profundas do solo, diferentemente dos outros estratos. O uso das métricas mostram que os valores mínimos de NDVI (período de maior estiagem) para todos os estratos foram registrados em 2005. Já os valores máximos (período de maior precipitação e produção de biomassa), corresponderam aos anos de 2005 e 2009 nos 4 estratos. Igualmente, a diferença anual do valor mínimo e máximo de NDVI sempre foi maior em 2005. Isto indica que nesse ano, os valores extremos de NDVI foram mais acentuados do que nos outros anos analisados, indicando em geral para todos os estratos, maior cobertura vegetal na época de chuvas, devido provavelmente a maior precipitação nessa estação, e menor cobertura na de secas, devido a uma estiagem mais prolongada.

Os Modelos Digitais de Terreno (MDT), representados por TIN (Triangulated Irregular Network), são gerados a partir de um conjunto de pontos amostrais irregularmente espaçados. A quantidade de pontos determina a complexidade do modelo, quanto maior a quantidade de pontos maior a complexidade do modelo e maior a sua resolução. A geração desse modelo em multirresolução pode ser obtida pela determinação de um subconjunto dos pontos utilizados na sua geração. Esse subconjunto deve garantir que o modelo gerado apresente as mesmas características do modelo completo, como declividade e orientação. Este trabalho apresenta uma metodologia para determinação de um ranking, em um conjunto de pontos amostrais, para criação de TIN multirresolução, garantindo que o modelo gerado apresente as mesmas características do modelo completo.

Most fires in Amazonia result from the combination between climate and land-use factors. They occur mainly in the dry season and are used as an inexpensive tool for land clearing and management. However, their unintended consequences are of important concern. Fire emissions are the most important sources of greenhouse gases and aerosols in the region, accidental fires are a major threat to protected areas, and frequent fires may lead to permanent conversion of forest areas into savannas. Fire-activity models have thus become important tools for environmental analyses in Amazonia. They are used, for example, in warning systems for monitoring the risk of burnings in protected areas, to improve the description of biogeochemical cycles and vegetation composition in ecosystem models, and to help estimate the long-term potential for savannas in biome models. Previous modeling studies for the whole region were produced in units of satellite fire pixels, which complicate their direct use for environmental applications. By reinterpreting similar remote-sensing input data using a statistical approach, we were able to calibrate models for the whole region in units of probability, or chance of fires to occur. The application of these models for years 2005 and 2006 provided maps of fire potential at 3-month and 0.25-deg resolution as a function of precipitation and distance from main roads. In both years, the performance of the resulting maps was better for the period July-September. During these months, most of satellite-based fire observations were located in areas with relatively high chance of fire, as determined by the modeled probability maps. In addition to reproduce reasonably well the areas presenting maximum fire activity as detected by remote sensing, the new results in units of probability are easier to apply than the results from previous fire-pixel models.

According to climate and hydrological conditions, large tropical areas would potentially be covered by forests. Remote sensing and field surveys, however, show how these regions may differ from their potential original states because land use, such as logging, agriculture and pastures. Related processes also contribute to the replacement of forests by secondary and degraded vegetation, and to the establishment of savannas. Fires are such a process normally associated to the current land-use practices in the tropics. Because savannas are adapted to fires and benefit from their occurrence, forest-to-savanna transitions are potentially very stable near deforestation areas. Here we present a method to account for deforestation and fires on the distribution of tropical and seasonal forests and savannas in the Brazilian Amazonia, and underline some strategies for implementing it in long-term vegetation models. We analyzed soil-hydrology and fires occurrence data together with the results from the potential vegetation model CPTEC-PVM2, and explored two factors used to determine the potential distribution of major biomes in that model. One is a soil-moisture seasonality index (D) which is lower (higher) for places where the soil presents longer (shorter) dry periods. Other is an above-ground wetness index (H) which is lower (higher) for longer (shorter) periods of dry atmospheric conditions. The analysis indicated the most common values for both indexes in areas potentially covered by tropical forests in the Brazilian Amazonia that had low, medium and high fire activity in the years 1998-2005 according to active-fire detections with the Tropical Rainfall Measuring Mission – Visible and Infrared Scanner (TRMM-VIRS). For each fire-activity class, we determined the distribution of values for both indexes and found the values below which deforestation/fire degradation were relatively more frequent. We now plan to use these relations when accounting for deforestation/fires in projections of the spatial distribution of the major biomes in the region. For that, we selected and discuss the following strategy that we consider appropriated. First, we assume that land-use dynamics and policy currently observed in the Brazilian Amazon will remain in the future. Second, we consider that deforestation/fires will affect forests only in grid cells of tropical and seasonal forests that are located adjacent (<150km) to savannas, because the access to the forests is facilitated by the savannas where fires are also a natural feature. In summary, if a place projected to be covered by tropical or seasonal forest presents both indexes H and D below certain thresholds, and the area is adjacent to a savanna, then tropical forest will be adjusted to seasonal forest. In case of a seasonal forest it will be adjusted to savanna. Although our method may represent an underestimation of fire effects by not considering direct transitions from tropical forests to savannas, it allows for quantifying changes in vegetation cover including human factors, and based on simple general relations derived using data from the region.

A disponibilidade de dados hiperespectrais trouxe expectativas nos meios acadêmicos e empresariais quanto à potencialidade de sua aplicação no setor florestal. O objetivo deste trabalho foi avaliar a aplicação de dados hiperespectrais do sensor Hyperion EO-1 na caracterização espectral de plantios de Eucalyptus spp. e os efeitos da geometria de iluminação nesse processo. Para avaliar a geometria de iluminação foram utilizados dados topográficos provenientes do SRTM (Shuttle Radar Topography Mission) e as condições de iluminação presentes no momento da aquisição da imagem de satélite, gerando assim um fator de iluminação para a área de estudo. Os dados hiperespectrais foram convertidos para valores de FR de superfície, os quais passaram a constituir os dados fundamentais de todo o trabalho. A área de estudo contemplou os plantios do gênero Eucalyptus spp., pertencentes à empresa Fibria S.A., localizados no município de Capão Bonito-SP. Foram coletados dados de diferentes espécies e em diferentes etapas de desenvolvimento, o que caracteriza diferenças estruturais nos dosséis dos plantios. As componentes principais foram aplicadas nos dados e foram confeccionados gráficos dos fatores referentes às duas primeiras componentes, dos quais foram extraídas informações estruturais sobre os plantios.

The feedback mechanisms between climate and vegetation have an important role in the dynamics of the biosphere-atmosphere interaction. In this sense, the knowledge of spatial distribution, temporal dynamics and biophysical characteristics of the vegetation types, are important elements to improve the understanding of how vegetation interacts with the atmosphere. The main objective of this study is to determine the relationship of environmental variables such as precipitation and fire with the patterns of spatial and temporal distribution of the main vegetation type, in the Brazilian tropical savanna. The methodology consists of the spatial and temporal analysis of the following variables: precipitation, density of fire spot and vegetation indices (normalized difference vegetation index-NDVI, enhanced vegetation index-EVI and land surface water index-LSWI) for the years of 2002, 2005 and 2008. The temporal resolution is 23 periods of 16 days each to complete a year and the spatial resolution is 1 km to the vegetation indices and density of fire spots, and 20 km for precipitation. Points randomly distributed along the entire biome were sampled (N=800) to perform statistical analysis. Time series and multivariate analysis techniques were applied to relate the influence of environmental variables (precipitation and fire) with the distribution pattern of 4 vegetation types (forest, savanna with trees, shrubs and herbaceous) in the study area. The results show seasonality in precipitation, density of fire spot and vegetation indices of the vegetation types analyzed along the biome. In general, the analysis show a direct positive relationship between precipitation and leaf biomass production (derived from vegetation indices), and on the other hand, both variables (precipitation and vegetation) show a negative relationship with the density of fire spots. This means that increases in precipitation are related to increases in vegetation cover and decreased in density of fires spots. It was also found high density of fires spot in the dry season in deciduous trees, shrubs and herbaceous which indicates the high removal of CO2 (greenhouse gas) of the land cover to the atmosphere somehow influencing the dynamic equilibrium of this (atmosphere) in the region of the Brazilian tropical savanna.