Analysing forest aboveground carbon dynamics in the Amazonia forests using dense time-series of satellite data and artificial intelligence

概述：亚马逊森林拥有最大的森林碳库，但这一估计仍然缺乏量化（Espirito-Santo et al. 2014）。用于估计热带森林，特别是巴西亚马逊地区地上生物量空间变化的遥感方法通常很少，而且空间分布往往有限。目前的生物量图之间存在显著差异（图1），导致参考这些特定的生物量图时，土地利用变化计算的碳排放量具有很高的不确定性。因此，我们迫切需要一种新的方法来估算森林生物量及其变化，而不是过度依赖森林样地数据。云和高性能计算的最新进展与先进的人工智能（AI）算法和对新卫星任务的重大投资并行。机器学习（Le Cun et al. 2015）此前已被应用于高光谱图像分类（Hu et al. 2015）、Sentinel-1 SAR图像的CORINE土地覆盖制图（Balzter et al. 2015）以及使用SAR和光学图像组合的森林生物量制图（Rodriguez-Veiga et al. 2016）。机器学习可以自动检测卫星图像中的森林变化。寻找空间和时间模式的范例，而不是历史上对卫星图像中光谱信息的关注，允许识别不同类型的森林动态（扰动和演替）。人工智能还可用于准确估算森林清查数据中难以测量的空间森林生物物理参数（Rodriguez-Veiga et al ., 2017）（图1）。这个跨学科的研究项目旨在探索使用机器学习来量化几个巴西森林遗址密集时间序列卫星数据中地上生物量碳的变化。多光谱光学和合成孔径雷达（SAR）传感器的时间序列叠加将被输入到人工智能中。人工智能将根据从现场森林清单收集的测量数据和对非常高分辨率图像的视觉解释进行训练。研究问题:1。亚马逊森林地上生物量的碳储量和通量是多少？基于卫星时间序列信息，人工智能在量化亚马逊森林动态方面的训练能有多精确？3. 估算相关的地上生物量损失或增加有多准确？图1：地上生物量的全球和泛热带地图（Rodriguez-Veiga等，2017）。

Overview: Amazon forests hold the largest pools of forest carbon, but this estimate remains poorly quantified (Espirito-Santo et al. 2014). The remote sensing methods adopted to estimate the spatial variation of above-ground biomass in tropical forests, notably the Brazilian Amazon, are usually scarce and often limited in their spatial distribution. There are notable differences among current biomass maps (Figure 1), leading to high uncertainties in the carbon emissions calculated from land-use changes when referring to these specific biomass maps. Therefore, we urge a new approach to estimate forest biomass and its changes, without heavy reliance on forest plot data.Recent advances in cloud and high-performance computing are paralleled with advanced artificial intelligence (AI) algorithms and significant investment in new satellite missions. Machine learning (Le Cun et al. 2015) have previously been applied to hyperspectral image classification (Hu et al. 2015), CORINE land cover mapping from Sentinel-1 SAR images (Balzter et al. 2015) and forest biomass mapping using a combination or SAR and optical images (Rodriguez-Veiga et al, 2016). Machine learning enables automatic detection of forest changes of satellite images. The paradigm of looking for spatial and temporal patterns instead of the historic focus on spectral information in satellite imagery allows the identification of the different types of forest dynamics (disturbance and succession). AI can also be used to accurately estimate from space forest biophysical parameters that are difficult to measure in forest inventory data (Rodriguez-Veiga et al, 2017) (Figure 1). This interdisciplinary studentship aims to explore the use of machine learning to quantify changes of aboveground biomass carbon in dense time-series satellite data of several Brazilian forest sites. Time-series stacks of multispectral optical and synthetic Aperture Radar (SAR) sensors will be input into the AI. The AI will be trained based on measurements collected from in-situ forest inventories and visual interpretation of very high resolution images.Research questions:1. What are the carbon stocks and fluxes of aboveground biomass of the Amazon forests?2. How accurately can an AI be trained to quantify forest dynamics of the Amazon based on satellite time-series information? 3. How accurately can the associated aboveground biomass loss or gain be estimated? Figure 1: Global and pantropical maps of aboveground biomass (Rodriguez-Veiga, et al., 2017).