Akpo H. A., Atindogbé G., Maxwell C. Obiakara M. C., Gbedolo M.A., Laly F. G., Lejeune P., Fonton N. H.
There is an increase in the use of photogrammetric point clouds for tree attribute mensuration. Stem diameter and circumference can be estimated from point clouds using stem disks of varying thicknesses along the bole. However, there is a dearth of information on the effect of the thickness of point cloud-based stem disks on the accuracy of diameter and circumference estimations. In this study, we outlined a GIS-based procedure for analysing Structure from Motion-derived photogrammetric point clouds with a view to providing an optimal disk thickness for accurate circumference estimates. Geo-referenced point clouds were created from photographs of 30 trees belonging to five savanna species. For each tree, 20 horizontal stem disks, with increasing thicknesses of 1 to 20 mm were made at breast height using the open source QGIS software. The resulting cross-sections were manually delineated and digitised. The difference between reference (manually measured) and point cloud-based circumferences at breast height was expressed as mean absolute percent error (MAPE) and compared across tree species, size and disk thickness. We found significant effects of species identity, tree size and disk thickness on MAPE. A stem disk of 7 mm in thickness provided consistently lower MAPE values (< 6%). This suggests that the accuracy of tree stem circumference estimations from photogrammetric point clouds depends on stem disk thickness.
Consultez la notice complète de l’article sur ORBi