Scholarlink

<p>Developing high-precision spatial datasets of plantation forests (PF) and natural forests (NF) in Southeast Asia (SEA) is critical for understanding tropical forest loss, biodiversity degradation, agroforestry management, and regional sustainable development. However, existing land cover products and related forest thematic data often fail to distinguish between PF and NF, and often lack the high spatial resolution and consistent temporal frequency. Here, we generated an annual 10-m resolution PF and NF distribution dataset for SEA (or SEA-PNF) from 2017 to 2024, using an iterative sample-refinement strategy integrating multi-source data and temporal features. By integrating Google Alpha Earth imagery, Random Forest classifiers, and a Hidden Markov Model (HMM)-based temporal optimization algorithm, we effectively addressed classification challenges inherent to complex tropical landscapes. Validation results of our SEA-PNF demonstrate that both the producer’s accuracy and the user’s accuracy for PF and NF exceed 0.95. This dataset reveals the fine-scale spatio-temporal characteristics of PF and NF in SEA, providing essential data support for regional agroforestry changes and ecological assessment.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A 10-meter annual time-series of plantation and natural forests in Southeast Asia (2017–2024)

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月20日

查看全文
  • Cheng Rui,
  • Chiwei Xiao,
  • Yunfeng Hu

摘要

Developing high-precision spatial datasets of plantation forests (PF) and natural forests (NF) in Southeast Asia (SEA) is critical for understanding tropical forest loss, biodiversity degradation, agroforestry management, and regional sustainable development. However, existing land cover products and related forest thematic data often fail to distinguish between PF and NF, and often lack the high spatial resolution and consistent temporal frequency. Here, we generated an annual 10-m resolution PF and NF distribution dataset for SEA (or SEA-PNF) from 2017 to 2024, using an iterative sample-refinement strategy integrating multi-source data and temporal features. By integrating Google Alpha Earth imagery, Random Forest classifiers, and a Hidden Markov Model (HMM)-based temporal optimization algorithm, we effectively addressed classification challenges inherent to complex tropical landscapes. Validation results of our SEA-PNF demonstrate that both the producer’s accuracy and the user’s accuracy for PF and NF exceed 0.95. This dataset reveals the fine-scale spatio-temporal characteristics of PF and NF in SEA, providing essential data support for regional agroforestry changes and ecological assessment.