基于长时间序列Landsat卫星影像的大兴安岭森林覆盖时空动态及其归因研究

Forest cover is a key parameter that can reflect forest dynamics and is thus of great significance for global greening and carbon storage. The forest of Great Xing'an Mountains in northeast China are typical mountain forests located in a climatic zone, resulting in significant differences in climatic conditions between the north and south and the windward and leeward slopes. Furthermore, the northern Great Xing'an Mountains is a hotspot to cause forest fire in China. These factors primarily drive the dynamics of forest cover, which effects could be further enhanced due to global changes. However, few studies have focused on the multiple factors driving the dynamics of temperate forests in a large scale. Here, this project would combine remote sensing and field survey data, as well as the key phenological spectrum information for identifying the Great Xing'an Mountains forest with time series of Landsat images, to establish an algorithm for accurately extracting the forest cover, and then analyze the spatio-temporal dynamics of forest cover (1984–2019). Also, this study will reveal how multiple factors including climate (precipitation, temperature, light, snow), forest fire, human activities (land use/cover conversions), geography and soil conditions (elevation, soil type, etc.) affect forest cover changes, and focus on the integrated impacts of "climate change-forest fire disturbance-human activities" on the spatio-temporal dynamics of forest cover in the Great Xing'an Mountains. This project aims to provide reliable data in support for accurately evaluating the spatio-temporal dynamics of the Great Xing'an Mountains, and to provide a scientific basis for the protection of regional forest ecosystems.

森林覆盖是能够反映森林动态的关键参数，对研究全球绿化、碳存储等具有重要意义。我国东北大兴安岭为典型的山地森林并处于气候分带，气候条件在南北方向和迎风背风坡均存在显著差异，同时又是森林火灾高发区，这些要素均为影响森林覆盖变化的主要因素，而这一作用在全球变化加剧的驱动下，可能进一步增强。然而，目前研究却很少关注多因素对大兴安岭森林的综合驱动效应。因此，本项目拟联合野外调查和遥感数据，分析基于Landsat数据识别森林的关键物候期特征光谱，构建提取森林覆盖的有效算法，进而分析长时间序列（1984–2019年）大兴安岭森林覆盖时空动态特征，揭示气候、林火、人类活动、地理和土壤条件等多因素如何影响森林覆盖变化，重点探讨“气候–林火–人类活动”等动态因素对森林覆盖时空动态的综合驱动效应。本项目旨在为评价大兴安岭森林时空动态提供可靠的数据支撑，为区域森林生态系统保护提供科学依据。

我国北方森林是重要的碳汇，对维持生物多样性也发挥着重要作用。由于受到气候变化、林火和人类活动干扰等影响，森林覆盖发生了显著变化。对此，我们通过利用基于Landsat影像数据和机器学习算法，实现森林覆盖分类精度高达90%以上。通过分析其1987–2020年之间的动态变化，结果表明我国北方森林永久森林面积为404168平方公里，森林覆盖净增加190209.5平方公里，森林覆盖净减少9198.75平方公里，发生森林扰动的面积为284971平方公里。另外，通过分析多环境因素对森林叶片生物量的影响，结果表明降水、温度、树龄和野火等因素能够很好地解释70%以上森林叶片生物量的空间变异。本项目为科学评价我国北方森林时空动态和其森林生态服务价值提供可靠的数据支撑。