Comparison of TLS against traditional surveying method for stem taper modelling. A case study in European beech (Fagus sylvatica L.) forests of mount Amiata

Traditionally, taper equations are developed from measurements collected through a destructive sampling of trees. Terrestrial laser scanning (TLS) enables high levels of accuracy of individual tree parameters measurement avoiding tree felling. With this
study, we wanted to assess the performance of two approaches to calibrate a taper function: using stem diameters extracted from
TLS point clouds and measured at different tree heights with the traditional and usual forest instruments. We compared the performance of four taper equations built with data collected by TLS and traditional survey in a European beech (Fagus sylvatica L.) forests
of mount Amiata (Tuscany Region, Italy). We computed the volume of stem sections 1.00 m long by integrating the most performing
TLS-based taper equation and by the Huber, Smalian and cone formulas applied on the diameter and height values measured with
the traditional field surveys. We conducted the analysis of error distribution in volume estimates computed integrating the most
performing TLS-based taper function along the stem. We tested if the differences in the volume estimate of the two methods were
significant. Schumacher and Hall (1933) equation was the most performing taper function both in case of using TLS and traditional
surveyed data, being the TLS-based function more performant (rRMSE = 9.17% vs 6.90%). Its performance did not increase when
diameter values were extracted from TLS point clouds with a higher frequency (i.e. 25.0 cm vs 1.00 m). By integrating the TLS-based
Schumacher and Hall (1933) function, the sections with the highest error resulted from 5.00 to 7.00 m of stem height (i.e. RMSE from
14.72 to 19.14 dm3
and rRMSE from 13.00 to 17.76%). This study case represents the first attempts to develop a taper equation for
European beech of mount Amiata using values of stem diameter and height extracted from the TLS point cloud. The results demon-strated that TLS produces the same stem volume estimates as traditional method avoiding falling trees.