Mapping Eucalyptus plantation in Guangxi, China by using knowledge-based algorithms and PALSAR-2, Sentinel-2, and Landsat images in 2020

Abstract

Eucalyptus plantations promote the economic development of forestry in southern China, but many studies have reported their negative environmental impacts, such as high water resource usage of certain species of Eucalyptus plants and losses in biodiversity. To date, annual maps of Eucalyptus plantations at large scales with high spatial resolutions are not yet available. Here, we investigated the spectral properties of Eucalyptus plantations and developed a knowledge-based Eucalyptus plantation mapping algorithm. We produced annual maps of Eucalyptus plantation at 10-m spatial resolution in the Guangxi Zhuang Autonomous Region (Guangxi), China, using our proposed algorithm and images of ALOS PALSAR-2, Sentinel-2, and Landsat (ETM+/OLI) in a single year. First, we generated annual evergreen forest maps using PALSAR-2 and Landsat/Sentinel-2-based vegetation index time series data. Second, we distinguished Eucalyptus plantations from the evergreen forest layer using the unique biophysical features of Eucalyptus plantations, the Sentinel-2 red edge bands, and Landsat/Sentinel-2-based enhanced vegetation index (EVI). Our resultant 2020 Eucalyptus plantation map had high producer's, user's, and overall accuracies of 0.85, 0.89, and 0.96, respectively. There were 3.10 × 106 ha of Eucalyptus plantation in Guangxi in 2020. Among the 14 administrative units, Wuzhou City had the largest Eucalyptus plantation area in Guangxi, followed by Nanning, Baise, and Chongzuo cities. We demonstrated the potential of knowledge-based mapping approaches for identifying evergreen forest and Eucalyptus plantations in complex and fragmented landscapes where cloud cover is frequent. Our 10-m Eucalyptus plantation map is the most current dataset available and can be used to assist the sustainable production of Eucalyptus, ecological assessments, and conservation.