Simple approach to improving the extraction of canopy metrics from airborne laser scanning data for tropical forests

Zhengyang Hou, Qing Xu, Chao Zhang, Matti Maltamo, Timo Tokola

Research output: Contribution to journalArticle

1 Scopus citations

Abstract

We aim to improve the predictive mapping of stem volume with airborne laser scanning (ALS) data acquired in Laos by adapting the area-based approach (ABA) to a tropical context. Separating laser returns of bushes from main stories with a cut-off threshold is a step very important to the ABA. The adaptation focused here on applying global and plot-adaptive cut-off thresholds to improve the extraction of canopy metrics. In order to select the optimal global cut-off threshold for removing understory bushes and ground objects, a sensitivity analysis of the modeling efficacy to the global cut-off threshold was conducted in the range from 0 to 5 m at 0.1-m intervals. To account for structural variation between plots, a simple plot-adaptive method was proposed for adjusting the threshold of each specific plot. The results showed that the optimal global cut-off threshold, which implicitly assumed the forest structure being homogeneous for all plots was 3.6 m. A model based on the plot-adaptive cut-off thresholds achieved better accuracy (RMSE 28%) than did the optimal global threshold-based model (RMSE 30%). It is concluded that the ALS-based canopy metrics extracted using the plot-adaptive method describe the structural heterogeneity of tropical forests adequately, whereas the global thresholding method is contingent on the forest structure being simple.

Original languageEnglish (US)
Article number016019
JournalJournal of Applied Remote Sensing
Volume10
Issue number1
DOIs
StatePublished - Jan 1 2016

Keywords

  • airborne laser scanning
  • area-based approach
  • feature extraction
  • forest inventory
  • tropical forests

Fingerprint Dive into the research topics of 'Simple approach to improving the extraction of canopy metrics from airborne laser scanning data for tropical forests'. Together they form a unique fingerprint.

Cite this