Novel 3D geometry and models of the lower regions of large trees for use in carbon accounting of primary forests

Abstract

There is high uncertainty in the contribution of land-use change to anthropogenic climate change, especially pertaining to below-ground carbon loss resulting from conversion of primary-to-secondary forest. Soil organic carbon (SOC) and coarse roots are concentrated close to tree trunks, a region usually unmeasured during soil carbon sampling. Soil carbon estimates and their variation with land-use change have not been correspondingly adjusted. Our aim was to deduce allometric equations that will allow improvement of SOC estimates and tree trunk carbon estimates, for primary forest stands that include large trees in rugged terrain. Terrestrial digital photography, photogrammetry and GIS software were used to produce 3D models of the buttresses, roots and humus mounds of large trees in primary forests dominated by Eucalyptus regnans in Tasmania. Models of 29, in situ eucalypts were made and analysed. 3D models of example eucalypt roots, logging debris, rainforest tree species, fallen trees, branches, root and trunk slices, and soil profiles were also derived. Measurements in 2D, from earlier work, of three buttress 'logs' were added to the data set. The 3D models had high spatial resolution. The modelling allowed checking and correction of field measurements. Tree anatomical detail was formulated, such as buttress shape, humus volume, root volume in the under-sampled zone and trunk hollow area. The allometric relationships developed link diameter at breast height and ground slope, to SOC and tree trunk carbon, the latter including a correction for senescence. These formulae can be applied to stand-level carbon accounting. The formulae allow the typically measured, inter-tree SOC to be corrected for not sampling near large trees. The 3D models developed are irreplaceable, being for increasingly rare, large trees, and they could be useful to other scientific endeavours.

Keywords: 3D; allometric; buttress; humus mound; land-use emissions; primary forest; root volume; soil carbon.

Figure 1.

Study area. Catchments (outlined in black) in Tasmania, Australia where data were acquired. Green area shows approximate distribution of Eucalyptus regnans-dominated forests. Pseudo Plate Carree projection, spheroid WGS1984, lat/long coordinates.

Figure 2.

3D model from Photoscan with ArcGIS markup, showing buttress region and humus mound. Eucalyptus regnans, DBH = 4.95 m, shown before logging in Dean (2003) and Supporting Information—Fig. S2E. (A) Top view, lines: outermost = footprint, innermost = hollow, in between and convoluted = corrected 1.3 m contour, moderately convoluted = 1.3 m contour, least convoluted = convex hull at 1.3 m, radial = humus profiles. (B) Graphed humus profiles and line of best fit, (C and D) oblique view, (D) example triangular and rectangular pyramids used to calculate humus volume for a flute.

Figure 3.

Buttress area characterization from 3D model, example 1. Eucalyptus regnans (DBH 4.38 m, Tyenna Valley). (A) Orthophoto created using terrestrial photography and Photoscan, top view. Red line = 1.3 m corrected contour, blue line = convex hull at 1.3 m matching DBH tape, brown line = footprint, magenta line = hollow. Humus mound removed, soil removed from one major lateral root. Coarse roots extend beyond footprint. Roots continue spiral grain of trunk and are plaited, curling around hemi-epiphytic trees and aiding host tree stability. (B) 3D model of root slice of large lateral within the footprint, ring age count = 350(±40) years. (C) Google Earth® satellite image shows felled trunk, stump and neighbouring stumps, during logging (scale bar = 40 m) (insets enlarged in Supporting Information—Fig. S5).

Figure 4.

Buttress area characterization from 3D model, example 2. Eucalyptus regnans (DBH 3.24 m, height 64 m, Styx Valley) on steep slope (24°) with minimal humus mound: volume = 0.96 m³. (A) Orthophoto created using terrestrial photography and Photoscan. (B) Topography (DEM), 1.3 m corrected contour, convex hull and footprint. (C) Upper portion of tree in ‘(A)’, above neighbouring rainforest understorey, and with ample foliage and original crown.

Figure 5.

Components of geometric analysis of myrtle tree DBH 1.76 m, height 32(±4) m, pushed over during logging. Process of myrtle root excavation (A) (with Prof. Kirkpatrick) and (B); photography with scale bars for photogrammetry (C); and subsequent 3D model in Photoscan (D). Orthophotos of the 3D model: (E) top view of whole fallen tree (gap on RHS was a section removed to measure trunk hollow, but it was included in the root:shoot ratio calculation); (F) top view of rotated model showing cross-section at 1.3 m (red outline) and convex hulls (visible and estimated total, brown lines); (G) side view of model showing root volume decrease with depth.

Figure 6.

Geometric analysis from 3D model of myrtle tree in Fig. 5: ERV as a function of height from the mineral soil surface.

Figure 7.

Some attributes correlated with individual tree measurements. (A) Footprint of eucalypt trees (represented by circle size) versus DBH and ground slope. The 95 % confidence intervals, corresponding to 2.6 % of the one parameter, were too narrow to show at this scale. (B) Humus area within the humus mound of eucalypt trees and the corresponding humus volume, as functions of footprint.

Figure 8.

Potential wood area at 1.3 m versus DBH. Solid, black line is regression fit to 3D models, dotted lines are 95 % confidence intervals. ‘enclosed-hollow’ = enclosed minus hollow. Gap between upper black dashed line and black solid line = acircular area deficit.

Figure 9.

Acircular area deficit versus DBH. The deficit is likely to be more variable for younger and/or smaller mature trees. This variability will influence that of stand-level humus volume.

Figure 10.

Roots growing inside larger roots reduce ERV. The displaced SOC by the roots will be less, but this is taken into account in the modelling presented here.