ABSTRACT

Forest cover change analyses have an essential role in forest management. Thus, this study adopted Landsat satellite imagery to assess the decadal spatiotemporal forest cover changes that occurred between 1989 and 2019 and predicted the 2029 land cover distribution of the Nkandla forest reserve, facing encroachment threats. The support vector machine algorithm and Land Change Modeling were utilized to classify and detect changes that occurred between 1989–1999, 1999–2009, 2009–2019. The Markov Chain Model and Multi-Layer Perceptron were adopted for the future land cover prediction. Consistent changes through inter-transitioning between the land cover types (closed canopy forest, open canopy forest, grassland, and bare sites) were detected. The closed canopy forest increased from 883.46 ha to 1059.23 ha, whereas the open canopy forest declined from 1091.89 ha to 910.60 ha between 1989 and 2019. Generally, the observed changes were caused by ecological processes and human disturbances. The future cover prediction indicated that the closed canopy forest will decline between 2019 and 2029, whereas the open canopy forest, grassland, and bare sites will increase. The information provided through this study will support the management of the Nkandla forest to ensure its continual supply of ecosystem services of national and global importance.

摘要

森林覆盖变化分析在森林管理中具有重要作用。因此，本研究采用 Landsat 卫星图像来评估 1989 年至 2019 年发生的十年间森林覆盖时空变化，并预测 Nkandla 森林保护区面临侵占威胁的 2029 年土地覆盖分布。利用支持向量机算法和土地变化建模对 1989-1999、1999-2009、2009-2019 年间发生的变化进行分类和检测。采用马尔可夫链模型和多层感知器进行未来土地覆盖预测。通过土地覆盖类型（封闭的树冠森林、开放的树冠森林、草地和裸地）之间的相互过渡，发现了一致的变化。封闭的冠层林从883.46公顷增加到1059.23公顷，而开放的冠层森林在 1989 年至 2019 年期间从 1091.89 公顷减少到 910.60 公顷。通常，观察到的变化是由生态过程和人为干扰引起的。未来覆盖率预测表明，2019 年至 2029 年，封闭的林冠森林将减少，而开放的林冠森林、草地和裸地将增加。通过这项研究提供的信息将支持 Nkandla 森林的管理，以确保其持续提供具有国家和全球重要性的生态系统服务。