Scholarlink

<p>Susceptibility of urban woodlands to plant invasions is often associated with their structural characteristics making timely detection and monitoring of invasive alien plant species (IAPs) important. The three-dimensional structure (3D) of forests can now be captured using unmanned aerial vehicles (UAV) based Structure-from-Motion (SfM) Digital Aerial Photogrammetry (DAP), a low-cost alternative to Light Detection and Ranging (LiDAR). However, there is still need to test whether ecologically meaningful linkages can be observed between IAPs occurrence and UAV-DAP derived forest structural and textural metrics as proxies of forest structure. In this study, we test whether UAV-DAP derived structural and textural metrics can explain the presence of <i>Psidium guajava</i>, an invasive alien plant using an ensemble of small models (ESMs). We assessed nine Grey-Level Co-occurrence Matrix textural features and 17 forest structural diversity metrics. Using a data set of 24 occurrences, we obtained relatively high Area under the curve (AUC) with the final ensemble of the ESMs achieving an AUC of 0.83 and True skills statistics (TSS) of 0.65. Rumple, maximum tree height (zmax), variability in tree height (zMADmedian), horizontal canopy texture (glcm mean and glcm variance) and leaf area (LAI) significantly influenced <i>P.guajava</i> habitat suitability. Results of this study, suggest that <i>P.guajava</i> invades woodland areas with less complex and heterogeneous structure. Our results highlight the usefulness of ESMs in combination with UAV-DAP derived forest structural and textural metrics in capturing forest structural characteristics linked to IAP occurrence. The approach can allow a low-cost efficient approach to monitoring IAPs in urban woodlands.</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

Coupling ensemble of small models with UAV-derived structural and textural metrics to predict occurrence of an invasive alien plant in an urban woodland reserve

  • 期刊论文

  • 发表日期：

    2025年11月12日

查看全文
  • Fadzai Michele Zengeya,
  • Odear Jafter,
  • Tinevimbo Rusike,
  • David-George Finch

摘要

Susceptibility of urban woodlands to plant invasions is often associated with their structural characteristics making timely detection and monitoring of invasive alien plant species (IAPs) important. The three-dimensional structure (3D) of forests can now be captured using unmanned aerial vehicles (UAV) based Structure-from-Motion (SfM) Digital Aerial Photogrammetry (DAP), a low-cost alternative to Light Detection and Ranging (LiDAR). However, there is still need to test whether ecologically meaningful linkages can be observed between IAPs occurrence and UAV-DAP derived forest structural and textural metrics as proxies of forest structure. In this study, we test whether UAV-DAP derived structural and textural metrics can explain the presence of Psidium guajava, an invasive alien plant using an ensemble of small models (ESMs). We assessed nine Grey-Level Co-occurrence Matrix textural features and 17 forest structural diversity metrics. Using a data set of 24 occurrences, we obtained relatively high Area under the curve (AUC) with the final ensemble of the ESMs achieving an AUC of 0.83 and True skills statistics (TSS) of 0.65. Rumple, maximum tree height (zmax), variability in tree height (zMADmedian), horizontal canopy texture (glcm mean and glcm variance) and leaf area (LAI) significantly influenced P.guajava habitat suitability. Results of this study, suggest that P.guajava invades woodland areas with less complex and heterogeneous structure. Our results highlight the usefulness of ESMs in combination with UAV-DAP derived forest structural and textural metrics in capturing forest structural characteristics linked to IAP occurrence. The approach can allow a low-cost efficient approach to monitoring IAPs in urban woodlands.