Reimer Florian, Asner Gregory P, Joseph Shijo
Center for Development Research (ZEF), Group Börner, Rheinische Friedrich-Wilhelm University, Walter-Flex-Str. 3, 53113 Bonn, Germany.
Department of Global Ecology, Carnegie Institution for Science, 260 Panama Street, Stanford, 94305 CA USA.
Carbon Balance Manag. 2015 Feb 3;10(1):5. doi: 10.1186/s13021-015-0015-8. eCollection 2015 Dec.
Conservation and monitoring of tropical forests requires accurate information on their extent and change dynamics. Cloud cover, sensor errors and technical barriers associated with satellite remote sensing data continue to prevent many national and sub-national REDD+ initiatives from developing their reference deforestation and forest degradation emission levels. Here we present a framework for large-scale historical forest cover change analysis using free multispectral satellite imagery in an extremely cloudy tropical forest region. The CLASlite approach provided highly automated mapping of tropical forest cover, deforestation and degradation from Landsat satellite imagery. Critically, the fractional cover of forest photosynthetic vegetation, non-photosynthetic vegetation, and bare substrates calculated by CLASlite provided scene-invariant quantities for forest cover, allowing for systematic mosaicking of incomplete satellite data coverage. A synthesized satellite-based data set of forest cover was thereby created, reducing image incompleteness caused by clouds, shadows or sensor errors. This approach can readily be implemented by single operators with highly constrained budgets. We test this framework on tropical forests of the Colombian Pacific Coast (Chocó) - one of the cloudiest regions on Earth, with successful comparison to the Colombian government's deforestation map and a global deforestation map.
热带森林的保护与监测需要有关其范围和变化动态的准确信息。云层覆盖、传感器误差以及与卫星遥感数据相关的技术障碍，持续阻碍着许多国家和次国家级的减少毁林和森林退化所致排放（REDD+）倡议制定其毁林和森林退化排放参考水平。在此，我们提出了一个框架，用于在云层极多的热带森林地区利用免费多光谱卫星图像进行大规模历史森林覆盖变化分析。CLASlite方法实现了从陆地卫星图像中对热带森林覆盖、毁林和退化的高度自动化制图。至关重要的是，CLASlite计算得出的森林光合植被、非光合植被和裸地的覆盖分数提供了森林覆盖的场景不变量，从而能够对不完整的卫星数据覆盖进行系统镶嵌。由此创建了一个基于卫星的森林覆盖综合数据集，减少了由云层、阴影或传感器误差导致的图像不完整性。这种方法可由预算极为有限的单个操作人员轻松实施。我们在哥伦比亚太平洋沿岸（乔科省）的热带森林——地球上云层最多的地区之一——对该框架进行了测试，并成功地与哥伦比亚政府的毁林地图以及一张全球毁林地图进行了比较。
