Skip to Main content Skip to Navigation
Master thesis

Apport de la segmentation à l’estimation d’attributs dendrométriques à partir de données de LiDAR aérien

Abstract : Recent developments in the field of airborne laser scanning offer new possibilities to map different forest attributes and improve the accuracy of inventory estimates. In this work, two segmentation methods of the LiDAR point clouds were used to delineate « crowns »and assess the possible gain in precision associate to the use of these new variables for dominant height, basal area and volume estimations. Three (3) contrasting forest stands were studied. The information obtained using segmentation variables were compared to those obtained by the « classical » method using different height deciles or moments of LiDAR points taken from the entire plot. Calibration plots (33 in total) were selected in stands of maritime pine (Pinus pinaster) in Landes (11 plots); conifer in Aillons (12 plots) and hardwood stands in the « l’observatoire pérenne de l’environnement » at Bure (10 plots). The results show that the segmented crowns would better characterize basal area and volume. The best models obtained were based on the joint use of traditional metrics and segmentation ones. We obtained for G, R ² = 0.91 and an average error = 5.8 m² / ha; for the total volume, R ² = 0.95 and an average error = 76.9 m³ / ha. For dominant height, the 80th height decile of the last returns, and the density of first returns reaching the 30th heigth decile (i.e close to the ground) showed higher accuracy than models using means of local maxima. So it seems that segmentation allows an improvement of forest attributes estimations. The robustness of these findings remains to be evaluated on a larger number of stands and more varied LiDAR acquisition configurations.
Document type :
Master thesis
File URL :
http://docnum.univ-lorraine.fr/public/BUS_M_2014_ANDRE_ANA-CRISTINA.pdf
Complete list of metadata

Cited literature [63 references]  Display  Hide  Download

https://hal.univ-lorraine.fr/hal-01841051
Contributor : Memoires Ul Connect in order to contact the contributor
Submitted on : Tuesday, July 17, 2018 - 8:54:45 AM
Last modification on : Monday, October 19, 2020 - 2:42:32 PM
Long-term archiving on: : Thursday, October 18, 2018 - 12:23:06 PM

File

BUS_M_2014_ANDRE_ANA-CRISTINA....
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01841051, version 1

Collections

Citation

Ana Cristina André. Apport de la segmentation à l’estimation d’attributs dendrométriques à partir de données de LiDAR aérien. Sciences de l'environnement. 2014. ⟨hal-01841051⟩

Share

Metrics

Record views

106

Files downloads

110