Publicações

The amount of aboveground biomass (AGB) held in vital components of vegetation play a significant role in the carbon cycle of tropical forests. Reducing uncertainty of terrestrial carbon cycle depend strongly on the accurate estimate of AGB. Lidar remote sensing provides the most precise methodology to quantify AGB at large scales, but the effects of the spatial resolution of airborne lidar data on AGB estimation is unknown. Here we examine the impact of the minimum spatial resolution threshold of lidar data to reduce the uncertainty of AGB estimations in tropical forest. For that we used a sizeable airborne lidar data from Tapajos National Forest (TNF) and ten permanent field plots. We compared two approaches: (1) we used general lidar allometric equation of AGB estimation developed for the Amazon, testing the best spatial resolution of lidar measurements at 25, 50 and 100 meters and compared with our ground data of forest inventory from TNF; (2) we developed and tested a new local lidar allometric equation to quantify AGB in TNF. Although the use of lidar cloud cover at 50 m provides unbiased estimates of AGB, our results demonstrated that local forest structure plays a significant role in this general allometric equations. Our results underscored three conclusions. First, the effects of the spatial resolution of airborne lidar data on AGB estimation were significant. We found that a minimum size-area of 50 meters of lidar is necessary to produce an unbiased estimate of AGB in a local tropical forest of Central Amazon. Second, our adjusted allometric equation for TNF, which was based in mean canopy height model, reduced the uncertainty of AGB from RMSE%: 36.8% to RMSE%: 26.2% (local model). Finally, this study highlights the need of lidar allometric equations based on local forest structure to reduce the uncertainty of AGB estimations.

A avaliação do regime de vento de forma confiável para uma determinada região por meio de coleta de dados com precisão adequada apresenta um alto custo financeiro. Os modelos meteorológicos de mesoescala são ferramentas que demandam investimentos relativamente baixos e vem apresentando resultados favoráveis. Porém, também possuem limitações, por exemplo, a dificuldade em representar a turbulência na CLP (Camada Limite Planetária), esta dificuldade está intrinsecamente relacionada à impossibilidade em reproduzir os processos físicos de trocas turbulentas de momento, energia e massa, entre superfície e atmosfera. Visto essas limitações este trabalho apresenta uma série de testes de sensibilidade entre o modelo meteorológico de mesoescala WRF (Weather Research and Forecasting) configurado com diferentes parametrizações físicas (camada limite, convecção, microfísica e radiação) e a velocidade do vento medida em estações meteorológicas automáticas do INMET (Instituto Nacional de Meteorologia) situadas no Nordeste do Brasil (NEB). O modelo WRF foi configurado com 3 domínio (grades) de resolução 5 x 5 km sobre a região em estudo, a localização desta grades se obteve a partir de uma análise de agrupamento (Cluster Analysis) realizada previamente com dados de intensidade do vento de 121 EMA do NEB para o período de 2008 a 2013. Os testes de sensibilidade configuraram um total de 12 experimentos, para os 3 primeiros dias do mês de maio de 2009. Ao analisar os resultados observa-se que independente das parametrizações adotadas, há uma forte tendência a superestimação da intensidade dos ventos pelo modelo WRF para as 13 localidades escolhidas para realizar análise estatística do erro quadrático médio e viés (RMSE e ME). Como erra esperado algumas parametrizações são mais adequadas para um domínio do que para outro, em função do NEB ser influenciado por diferentes sistemas meteorológicos, por exemplo, o ciclo anual da precipitação do litoral leste é diferente ao da região central do NEB.

Almost 20 years ago, the Amazon forest-dieback hypothesis has been proposed indicating that a large-scale loss of Amazon rainforest may be caused by climate change which may lead to substantial changes in ecosystem functioning and structure. Here, we revise the likelihood of a potential Amazon forest dieback based on a systematic literature review. We find that still large uncertainties exist about the impacts and drivers of such event. These uncertainties include the effects of increasing temperature and atmospheric CO2 concentration, and long-term drought on forest stability, the role of nutrient cycling and potential phosphorus limitation on forest productivity and potential climate feedbacks, in particular, the potential disruption of local rainfall recycling. We assess, in the light of these uncertainties, scenarios of potential socio-economic impacts that would result from a large-scale Amazon forest dieback. For our assessment, we consider the economic losses arising from changes in the provision of ecosystem services, decreasing crop yields, reduction of hydroelectric power generation potential, reduction of fish stocks and interruption of shipping waterways. Long-term economic losses are estimated between USD $1,367 to $6,928 billion (16%-80% of Gross Brazilian Amazon Product), arising mainly from changes in the provision of non-market value ecosystem services, decreasing crop yields and reduction of hydroelectric power potential. Trade-off gains coming from such a forest dieback would sum less than half of the losses (USD $576 to $2,880 billion). We conclude that, from a risk-analysis perspective, even with a low probability of occurrence, the high socio-economic impacts of an Amazon forest dieback make it, per se, a high-risk process.

Biological nitrogen fixation (BNF) is the main pathway by which nitrogen (N) is delivered to unmanaged ecosystems. Despite the importance of BNF on understanding and modeling key ecosystem functions such as carbon storage and climate feedbacks, information on BNF rates for natural ecosystems in Latin America (LA) is quite limited. In this study, we estimated total BNF rates for major biomes in LA using a literature-derived dataset and a modeling approach. We compiled published data on free-living and symbiotic BNF rates (kgN ha-1 y-1) from primary and mature sites located in different biomes in LA. We found enough empirical data to estimate total BNF rates only for Moist, Tropical Dry, and Temperate forests and for desert biomes in LA. Furthermore, we used the conservative global model of Cleveland et al. (1999) and mean annual evapotranspiration data (ET) from NASA GLDAS to estimate potential total BNF rates for other biomes, and further evidence on the extent of BNF rates in forest and desert biomes in LA. The model overestimated total BNF rates but still had a reasonable predictive power (r2=0.56). We found a large spatial variably on BNF rates across biomes in LA. Literature-derived rates were higher for Moist forests (10±2) than Tropical Dry forests (4±2), and similar between Temperate forests (2±2) and deserts (2±3). Modeled rates were higher for Moist forests (14±1), mangroves (12±1), Pantanal (11±1), and savannas (10±1), followed by Tropical Dry and Coniferous forests (9±1). Modeled rates were lower for the Mediterranean biome (2±2), followed by Montane grasslands (4±2), deserts (4±2), and Temperate grasslands (6±1) and forests (5±1). This study provided the first evidence on the extent of BNF rates for a range of biomes in LA. The information obtained in this study will greatly contribute to the refinement of our understanding of Latin Americas N balance and to model critical ecosystem functions for a variety of societal needs. This study was supported by the project Nitrogen cycling in Latin America: drivers, impacts and vulnerabilities (Nnet, IAI/CRN3005 and FAPESP 2012/06416-1), and collaborators.

Peatlands cover around 570 million hectares worldwide and although most are located in boreal environments, tropical peatlands cover 170 million hectares and store about 860 Gt of carbon. Peat soils store a large amount of carbon and have acted as sinks of atmospheric CO2 for thousands of years but can also act as major sources of GHG emissions into the atmosphere when impacted by anthropogenic activities such as drainage, deforestation, forest fires, and changes from native to agricultural and livestock systems. These changes, besides disturbing the functions of this ecosystem, affect the dynamics of CO2 and CH4 flows. Therefore, the objective of this study is to evaluate the effect of land use change and land cover (LULC) on the dynamics of carbon emissions in tropical peatlands in the Vale do Paraíba region, São Paulo, Brazil. The sampling program was carried out in the lowlands of the Paraíba do Sul River, over different land cover: pastureland, native forest and irrigated rice crop. We used the static chamber technique to estimate GHG fluxes during two seasonal periods (wet season of 2017 and dry season 2018). Our results show that the highest CO2 (g CO2 m-2 yr-1) fluxes were observed in the dry season in pastureland (3,210), followed by native forest (2,174) and irrigated rice crop (2,074). The higher values in the dry season was regulated mainly by peat moisture, organic matter content and groundwater level. The CH4 fluxes (gCH4 m-2 yr-1) were mainly dependent on the moisture, temperature of the peat and groundwater level. The CH4 fluxes were negative (uptake) for pastureland (-5.2) and native forest (-3.1) and positive (emission) for irrigated rice crop (3.1). In the rainy season, methane fluxes were similar for the three land covers, with emissions of 4.0, 4.2 and 3.1 gCH4 m-2 yr-1 for native forest, pastureland and irrigated rice crop, respectively. These results suggest an increasing trend of CO2 emissions in situations of higher temperature and deeper groundwater table, mainly driven by the drainage and deforestation of peatlands in the region. On the other hand, long-term drought conditions may promote methanotrophic activities and consumption of CH4, decreasing CH4 emissions. Thus, the role of the peatlands as sink or source of GHG to the atmosphere will depend mainly on the dynamics of the LULC changes and land management.

In this paper we analyze images of a lightning flashover induced by a downward negative lightning flash that struck a transmission line located in São Paulo, Brazil. The lightning flash and the flashover were recorded by a high-speed camera running at 10,000 images per second. The approach of the negative downward leader did not launch an induced positive upward leader prior to the return stroke. One of the AC transmission lines that run over a railway line was struck. The flashover exhibits an oscillatory behavior. The flickering of the image occurs at exactly 120 Hz. The return stroke that struck the line had an estimated peak current of a moderate intensity. It was however, followed by a continuing current lasting 17 ms and an intense M-component that may have contributed to the occurrence of the flashover.