Using satellite data to monitor REDD+ projects: developing methodologies and error estimation for Africa

热带国家的森林砍伐速度很快，占人类温室气体排放量的12- 15%。这比加剧气候变化和物种灭绝带来更多的问题。众所周知，在热带国家，森林的破坏会减少降雨，增加土壤侵蚀的速度，并增加洪水的风险。联合国正在实施一项旨在制止热带森林砍伐的重大政策进程：即减少森林砍伐和退化造成的排放（REDD）。REDD涉及发达国家向发展中国家直接付款，使它们能够在不破坏森林的情况下提高其人口的生活水平。这也有利于发达国家，因为这将降低气候变化的速度，保护这些生物多样性和稳定气候的森林。REDD谈判的一个主要症结是测量和监测森林面积，特别是“生物量”的方法。这是一个地区生物物质的数量，直接对应于植被中储存的碳的数量。重要的是能够在若干不同的时间点评估一个地区的生物量：在过去，以便可以计算森林砍伐的历史速度；在今天；并在未来测量森林砍伐，从而计算（并验证）适当的支付。测量森林生物量最准确的方法是测量每棵树的位置、直径和高度。这是昂贵和耗时的，因此大多数项目被限制在一个项目地点最多大约100个地块。因此，卫星数据被用来制作过去和现在的生物量地图，使用实地数据来测试和提高估计生物量的准确性。目前很少有独立的工作来评估现有的许多不同卫星监测系统的准确性。有三种主要的类型需要考虑。首先是各种光学卫星传感器，本质上类似于数码相机，从太空拍摄地球的图像；其次是雷达卫星传感器，它向地面发送微波辐射的ping信号，并“监听”它们的返回；第三是激光雷达数据，它与雷达类似，但使用激光脉冲。所有的系统都有不同的优点和缺点：光学数据不能穿透云层，只能确定森林面积而不能确定生物量，因此可能会错过退化；雷达数据是一项新技术，不同地区和不同时间的生物量图的一致性尚不清楚；激光雷达可以绘制出最精确的生物量地图，但目前还没有卫星系统可以收集这些数据，因此必须从飞机上收集，这大大增加了成本。我们的目标是利用REDD森林项目在三个具有不同植被类型的非洲国家（莫桑比克、乌干达和加蓬）客观地估计这些不同方法所涉及的误差，并寻找使用和组合它们的数据以提高准确性的新方法。我们已经与这些REDD项目的管理层建立了合作关系，这些项目处于不同的发展阶段，他们同意让我们访问他们的所有数据，并允许我们影响地面收集制度并分析他们地点的卫星数据。该项目将量化用于评估REDD项目中生物量变化的不同卫星方法的准确性。本研究得出的数据和分析将被广泛使用，供试图找到最佳方法的项目执行者和希望估计项目或国家监测制度准确性的投资者和独立监测机构使用。该项目有可能产生巨大影响，通过降低所有森林监测系统的成本和提高其准确性，可能有助于降低全球森林砍伐的速度。

Deforestation is occurring at a rapid rate in tropical countries, accounting for 12-15 % of human emissions of greenhouse gases. This causes more problems than exacerbating climate change and the loss of species. Destruction of forests is known to reduce rainfall, increase the rate of soil erosion and increase the risk of flooding in tropical countries. A major United Nations policy process is underway to halt tropical deforestation: it is called Reducing Emissions from Deforestation and Degradation (REDD). REDD involves developed countries making direct payments to developing countries, enabling them to increase their population's standard of living without destroying their forests. This also benefits the developed world, as it will reduce the rate of climate change and preserve these biodiverse and climate-stabilising forests. One major sticking point in the negotiations for REDD is methods for measuring and monitoring forest area, and in particular 'biomass'. This is the amount of living material in an area, and corresponds directly to the quantity of carbon stored in the vegetation. It is important to be able to assess the biomass of an area at a number of different time points: in the past so that the historical rate of deforestation can be calculated; at the present day; and into the future to measure deforestation and therefore calculate (and verify) appropriate payments. The most accurate way to measure the biomass of forests is to measure the location, diameter and height of every tree. This is expensive and time-consuming, and thus most projects are constrained to a maximum of about a hundred plots over a project site. Therefore satellite data is used to produce biomass maps for the past and present, using field data to test and improve the accuracy of estimated biomass. There is little independent work being done to assess the accuracy of the many different satellite monitoring systems available. There are three major types to consider. First the various optical satellite sensors, which are essentially similar to digital cameras, taking images of the Earth from space; secondly radar satellite sensors, which send pings of microwave radiation at the surface, and 'listen' for their return; and thirdly LiDAR data, which is similar to radar but uses pulses of laser light. All of the systems have different advantages and disadvantages: optical data cannot see through cloud, and can only determine forest area rather than biomass and so may miss degradation; radar data is a new technology and the consistency of biomass maps in different areas and at different times is unknown; LiDAR produces the most accurate biomass maps, but there is currently no satellite system that collects it, so it has to be collected from an aircraft, greatly increasing costs. We aim to objectively estimate the errors involved in these different methods using REDD forest projects in three African countries with differing vegetation types: Mozambique, Uganda and Gabon, as well as looking at new ways to use and combine their data to increase accuracies. We have already established collaborations with the management of these REDD projects, which are in various stages of development, and they have agreed to give us access to all their data, and to allow us to influence ground collection regimes and analyse satellite data over their sites. The project will quantify the accuracy of different satellite methods for assessing changes in biomass in REDD projects. The data and analyses emerging from this study will be widely used, by project implementers trying to find optimal methodologies, and by investors and independent monitoring agencies wishing to estimate the accuracy of the monitoring regime in a project or country. The project has the potential for a huge impact, potentially contributing to a reduction in the rate of global deforestation by reducing the cost and increasing the accuracy of all forest monitoring systems.