MAPPING AND QUANTIFYING POST-FIRE CARBON BUDGET IN AMAZONIA

全球气候变化和热带森林的大规模损失可能是当代最紧迫的环境问题。一些全球环流模型表明，亚马逊地区可能容易受到极端干旱的影响。在亚马孙地区干旱频率增加的情况下，火灾渗入森林可能是森林变化的主要因素。这一过程通过影响植被结构，改变碳库和通量，并对大气产生反馈，对全球碳循环产生重要影响，但迄今为止量化较少。缺乏关于这些过程的系统信息（空间和时间）是估计这一碳源大小和随后对亚马逊森林生物群系及其生态系统功能的维持的一个关键限制。因此，我在这个研究项目中的主要研究目标是将来自多时相卫星图像的信息与关于净生态系统生产力（NEP）组成过程的广泛实地数据相结合，以：(1)生成巴西亚马逊地区第一张受火灾影响的森林地图；(2)量化不同时间尺度、不同火灾频率和强度下森林隔室间碳的排放、吸收和移动量；(3)在考虑森林火灾影响的情况下计算亚马逊森林的碳收支，并了解这种变化如何受到气候和毁林率的控制；(4)了解森林火灾对碳平衡的时空影响。为了实现这一建议的目标，第一步将是分析多时相卫星图像，绘制受火灾影响的森林伤痕图，并量化该地区火灾的频率和强度。生物量地图将用于识别亚马逊地区不同的森林类别。最后，这些地图将被合并成一个单一的分层地图。这将表征该地区火灾损失的空间变异性。这张地图将是确定将在现场进行详细调查的区域的基础，也将在后期用于将现场数据推断到整个亚马逊地区。第二步研究不同火灾频率和强度对碳收支的影响。这将包括汇编现有的野外数据，并在关键的野外地点，直接测量碳平衡或NEP的组成过程，即植被通过光合作用吸收的二氧化碳与生态系统通过呼吸过程释放的二氧化碳之间的差异。这些测量将提供在不同时间尺度上从森林隔间排放、吸收和移动的碳量的详细信息。一旦分析了来自不同来源的数据，顺理成章的下一步就是整合这些数据集，以便了解亚马逊地区受火灾影响地区新能源政策组成过程的年度变化。最初将测试两种主要方法：(1)分区平均方法，其中NEP的平均值可以与每个火灾类别相关联；(2)决策树方法，该方法可以通过结合火灾类别信息以及经验校准的NEP及其组成过程与降雨量、生物量和森林类型之间的关系来实现。在这个阶段，考虑到森林火灾的影响，所有的信息都将用于计算亚马逊森林的碳预算。这项工作的最终目标是得出近年来受火影响的森林立地的碳吸收、释放和净平衡的年度空间明确值，量化决定这种净碳平衡的各组成过程的相对影响，并了解气候变化和人类活动（砍伐森林）如何影响其时空格局。

Global climate change and the large-scale loss of the tropical forests are probably the most urgent of contemporary environmental problems. Some global circulation models suggest that Amazonia may be vulnerable to extreme drying. In the event of increased drought frequency in the Amazon region, the leakage of fires into forests is likely to be the major agent of forest transformation. This process has an important influence on the global carbon cycling by affecting vegetation structure, changing carbon pools and fluxes, and causing feedbacks to the atmosphere, but it is thus far poorly quantified. The lack of systematic information (spatial and temporal) on these processes is a critical limitation when estimating the magnitude of this carbon source and consequent maintenance of the Amazon forest biome and its ecosystem function. Hence, my main research aim in this fellowship is to combine information from multi-temporal satellite images with extensive field data on the component processes of the net ecosystem productivity (NEP) to: (1) Generate the first map of the fire-affected forests in the Brazilian Amazon; (2) quantify the amount of carbon being emitted, absorbed and moved from and among forest compartments at different time-scales and at different fire frequency and intensity; (3) calculate the carbon budget of Amazonian forests taking into account the effect of forest fires, and understand how this variation is controlled by climate and deforestation rates; and (4) understand how the carbon balance varies temporally and spatially because of forest fires. To achieve the aims of this proposal, the first step will be to analyze multi-temporal satellite images to map the scars of forests affected by fire and quantify the frequency and intensity of fire in this areas. A biomass map will be used to identify different forest classes in the Amazon. Finally, the maps will be merged in one single stratified map. This will characterize the spatial variability of fire damage in the region. This map will be the basis for defining the areas that will be investigate in detail in the field and will be also used in a later stage to extrapolate field data to the whole Amazon. The second step will tackle the effect of different fire frequency and intensity on carbon budget. This will involve a compilation of existing field data and in key field sites, direct measurements of the component processes of the carbon balance or NEP, which is the difference between the CO2 assimilated by the vegetation through photosynthesis and the CO2 released by the ecosystem through the respiration process. These measurements will provide detailed information on the amount of carbon being emitted, absorbed and moved from and between forest compartments at different time-scales. Once the data from the different sources have been analysed, the logical next step is to integrate the datasets in order to understand the annual Amazon-wide variation of the component processes of NEP in the fire-affected areas. Two main approaches will be tested initially: (1) a zonal average method, where, an average value of NEP can be associated to each fire class; and (2) a decision tree method, that can be implemented by combining information about the fire classes and empirically-calibrated relationships between NEP and its component processes with rainfall, biomass and forest type. At this stage all the information will be available for calculating the carbon budget of Amazonian forests, taking into account the effect of forest fires. The ultimate goal of this work is to arrive at annual spatially explicit values of uptake, release and the net balance of carbon in fire-affected forest sites in recent years, to quantify the relative influences of the component processes that determine this net carbon balance and to understand how climate variation and human activities through deforestation, affects its spatial and temporal patterns.