Skip to main content

Table 1 Comparison of ACD-estimation models derived from area-based (1,2,3) and tree-centric (4,5) LiDAR statistics

From: Airborne laser scanning of natural forests in New Zealand reveals the influences of wind on forest carbon

Model (ACD=) a b c \(\sigma _{1}^{2}\) \(\sigma _{2}^{2}\) R2 RMSE BIAS
1 aTCHb 6.58±0.67 1.49±0.040   40.8 3.30 0.75 35.2 -1.6
2 \(a TCH^{b} \widehat {BA}^{c} \widehat {\rho }^{b_{3}}\)   See text     0.76 34.3 -1.8
3 aTCHb(1+cCover10) 6.27±0.64 1.51±0.04 0.93±0.16 40.7 3.02 0.74 35.9 -2.7
4 \(a \sum {(CD_{i} \cdot H_{i})^{b}} \) 0.023±0.0037 1.57±0.033   40.5 3.53 0.73 35.6 0.8
5 \( a + b \cdot (\overline {CD_{i} \cdot H_{i} }) \) 30.39±3.3 3.4±0.062   24.7 4.41 0.78 32.3 -0.1
  1. Terms in the models include basal area (BA) and wood density (ρ) recorded in plots, Top-of-the-Canopy Height (TCH) and residual variation in canopy cover (Cover10) estimated from ALS data, as well as Crown Diameter (CD) and Height (H) of individual overstorey trees obtained by segmentation of the ALS dataset. Mean ± 1 SD of coefficient values are given. Goodness of fit is assessed using R2, RMSE (%) and bias (%)